Insight Editions · No. 274
More Money Than God
Hedge Funds and the Making of a New Elite
Sebastian Mallaby
Sebastian Mallaby's history of hedge funds from Alfred Winslow Jones to the 2008 crash, through Steinhardt, Soros, Robertson and Renaissance, read as evidence that markets are far from efficient.
A study guide to More Money Than God by Sebastian Mallaby. Shared free for personal reading; not for sale and not endorsed by the authors or publishers. Please read and buy the original.
The Book on One Page
Mallaby tells the history of hedge funds from Alfred Winslow Jones's 1949 "hedged fund" to the 2008 crash as a series of edges: each generation found a way to beat supposedly efficient markets, grew rich, and saw the edge fade once others copied it or size destroyed it. His verdict is that hedge funds, because they are small enough to fail and run with their owners' money at stake, manage risk better than the too-big-to-fail banks that taxpayers had to rescue, and so should be encouraged, not smothered. The danger comes when a fund grows huge or goes public.
Read each chapter through Mallaby's single test: where does the edge come from, why does it fade, and who pays when it fails? The answer to the last question, investors rather than taxpayers, carries the book's policy conclusion.
The argument in eight moves
- 1. Markets are beatable, but not for long. Every chapter names a real edge (Jones's incentive system, Steinhardt's block trading, Commodities Corporation's trend following, Soros's reflexivity, Robertson's stock picking, Renaissance's code-breaking) and shows imitators or growth eroding it (Intro, Ch. 1-5, 13).
- 2. The best profits come from official mistakes. Untenable pegs and policy errors (Plaza 1985, sterling 1992, Thailand 1997, the subprime bubble) gave Soros, Druckenmiller and John Paulson their killings; traders ask about payouts, policy makers only about odds (Ch. 4, 7, 9, 15, 16).
- 3. Trading on positioning and forced sellers. Paul Tudor Jones read how others were placed and when they had to sell, becoming the trigger for stampedes in 1987 and Tokyo 1990 (Ch. 6).
- 4. Leverage makes overreaction mandatory. The 1994 bond rout and LTCM's collapse in 1998 showed that borrowed money turns small shocks into fire sales, and that precise risk models are brittle because crises create their own correlations (Ch. 8, 10).
- 5. Hedge funds are not superpredators. In Asia and Russia they were often cautious or wrong, and in the dot-com bubble fighting the trend ruined Tiger while Druckenmiller surfed it too long (Ch. 9, 11).
- 6. Institutions turned an elite into an industry. David Swensen's Yale model brought endowment money, event-driven funds such as Farallon promised smooth returns from illiquid assets, and multistrategy alpha factories such as Citadel and Amaranth scaled up; Amaranth's $6 billion loss hurt only its investors, and Citadel bought its book in two days (Ch. 12, 14).
- 7. The crisis was a banking failure. Hedge funds mostly shorted subprime while banks, and bank-run funds such as Bear's, bought it; Citadel put out the Sowood fire, and the August quant quake hurt quants, not the system; the industry fell 19% in 2008, about half the S&P's drop, closed about 5,000 funds in a decade and needed no bailouts, while Citadel survived by matching its funding (Ch. 15, 16).
- 8. Policy: encourage the small, watch the big. Bank safety nets breed risk and regulation is fallible, so push risk toward small partnerships; supervise only funds with over $120-200 billion of leveraged assets, or listed ones (Conclusion, Appendix).
Chapter map
| Ch | Title | One-line payload |
|---|---|---|
| Intro | The Alpha Game | Hedge funds exploit inefficiencies; small enough to fail, unlike banks |
| 1 | Big Daddy | A. W. Jones's hedge, leverage and pay-for-performance; the go-go collapse |
| 2 | The Block Trader | Steinhardt profits by supplying liquidity, partly by collusion |
| 3 | Paul Samuelson's Secret | Commodities Corporation survives on risk rules and trend following |
| 4 | The Alchemist | Soros's reflexivity: Plaza triumph, 1987 humiliation |
| 5 | Top Cat | Robertson's stock-picking edge, proven by his Tiger cubs |
| 6 | Rock-and-Roll Cowboy | Tudor Jones trades positioning, forced sellers and his own influence |
| 7 | White Wednesday | Druckenmiller and Soros break sterling and the ERM, 1992 |
| 8 | Hurricane Greenspan | A quarter-point hike and leverage sink Steinhardt and Askin, 1994 |
| 9 | Soros versus Soros | Speculator versus statesman in Asia and Russia, 1997-98 |
| 10 | The Enemy Is Us | LTCM: careful risk models fail because traders create correlations |
| 11 | The Dot-Com Double | Tiger shuts fighting the bubble; Druckenmiller rides it too long |
| 12 | The Yale Men | Swensen's endowment model and Farallon's illiquid, smooth returns |
| 13 | The Code Breakers | Renaissance, D. E. Shaw and the computers that find patterns |
| 14 | Premonitions of a Crisis | Alpha factories; Amaranth's $6bn gas loss; Citadel buys the book |
| 15 | Riding the Storm | Paulson shorts subprime; banks buy it; Sowood and the quant quake |
| 16 | "How Could They Do This?" | Lehman, the short ban and Citadel's survival without a bailout |
| Concl | Scarier Than What? | Too big to fail is the danger; "Don't regulate" all but the giants |
| App | Do the Tiger Funds Generate Alpha? | Tiger cubs' 11.9% a year and 1.42 Sharpe point to skill |
The Alpha Game
Mallaby opens with a two-sided argument: hedge funds earn their fortunes by exploiting market inefficiencies, and in doing so they mostly make prices saner, yet their leverage can also turn them into amplifiers of panic. Writing in early 2010, after the crash, he rejects the tempting verdict that they caused it. Hedge funds are not the culprit of 2007-2009 but a partial model for safer finance: small, self-funded firms that are "safe to fail, even if they are not fail-safe."
From A. W. Jones to Cliff Asness
The industry's founder had none of the credentials of a modern financier. Alfred Winslow Jones skipped business school and Wall Street apprenticeships; instead he crewed a tramp steamer, studied at the Marxist Workers School in Berlin, carried out covert work for an underground anti-Nazi cell called the Leninist Organization, married twice, honeymooned at the front of the Spanish Civil War, and drank and travelled with Dorothy Parker and Ernest Hemingway. Only at 48 did he scrape together $100,000 for a "hedged fund," which then produced extraordinary profits through the 1950s and 1960s. Mallaby's claim is that Jones stumbled, half by accident, onto a structure that has lasted ever since and will keep thriving whatever the critics say.
Fifty years on, Clifford Asness was Jones's mirror image. He had the business-school training, the quantitative-finance PhD and a star career at Goldman Sachs. He launched at 31, set a start-up record by raising $1 billion, and, unlike the secretive Jones, talked freely to television and told the New York Times that being worth millions was not unpleasant. By the eve of the 2007 subprime crash his firm, AQR Capital Management, managed $38 billion. He kept a shelf of plastic superheroes in his office and embodied the brash new finance.
- Performance fee. Jones kept one fifth (20%) of investment profits for himself and his team, which sharpened his lieutenants' incentives.
- Light regulation. By steering clear of red tape he kept the freedom to switch methods whenever opportunities shifted.
- Hedging by short selling. He paired purchases of promising stocks with borrowing and selling unpromising ones, so the fund was partly insulated from general market swings.
- Leverage. Having hedged out much market risk, he felt safe to magnify his bets with borrowed money.
Asness stressed the deeper point: the long/short plus leverage combination is not tied to equities. It works for bonds, futures, swaps and options, and any blend of them. More by luck than design, Jones built a platform for strategies far more complex than he could have imagined.
What counts as a hedge fund
Mallaby admits no definition is perfect, and many episodes in the book involve neither hedging nor leverage:
- Breaking the pound. When George Soros and Stan Druckenmiller bet against sterling, there was no need to hedge.
- Shorting housing. John Paulson's bet against the US mortgage bubble was likewise an unhedged directional wager.
- Siberian palladium. When a commodities trader negotiated to buy the Russian government's entire stock of non-gold precious metals, leverage mattered less than the security on the armoured train carrying the metal from Siberia.
Even so, the Jones platform has proved exceptionally congenial: freedom to go long or short in any instrument in any country lets funds pounce wherever opportunity appears, leverage lets them size each bet, and performance fees give a raw incentive to make money.
More money than God
At his death in 1913, J. Pierpont Morgan, nicknamed "Jupiter" for his power over Wall Street, left about $1.4 billion in current dollars. The top hedge-fund managers of the early 2000s amassed that much in a couple of years, far outstripping investment bankers and even private-equity barons.
- $54 million. Record award to Goldman Sachs chief Lloyd Blankfein.
- Just under $400 million. Blackstone's payout to Stephen Schwarzman, the private-equity king.
- $240 million. Reported take of the 25th (lowest) name on Alpha magazine's top-25 hedge-fund earners list.
- Over $1 billion each. Reported earnings of the top three hedge-fund managers.
- $2.15 billion. Combined 2006 pay of six hedge-fund managers near Stamford, per the Stamford Advocate.
- $150 billion. Total personal income of everyone in Connecticut that year.
The press turned from dot-com millionaires to hedge-fund excess. Ken Griffin of Citadel bought a $50 million Bombardier jet and fitted it with a crib for his toddler. Louis Bacon of Moore Capital bought an island in Great Peconic Bay, tagged the local mud turtles with transmitters to study their mating, and hosted English-style pheasant shoots. Steven Cohen of SAC Capital filled his estate with a basketball court, indoor pool, skating rink, a two-hole golf course, a vegetable plot, works by van Gogh, Pollock and Keith Haring, and a cinema whose ceiling reproduced the stars on his wedding night. Mallaby casts them as the new Rockefellers and Carnegies, the latest act in America's mix of creativity and greed.
Loners and contrarians
Hedge funds, Mallaby argues, are vehicles for individualists too ambitious for big institutions.
- Asness left a rising career at Goldman for the freedom and rewards of his own shop.
- Jim Simons of Renaissance Technologies, the industry's top earner in the 2000s, took orders from no one, rarely wore socks, and was fired from the Pentagon's code-breaking centre for attacking his bosses' Vietnam policy.
- Griffin, the second-highest earner of 2006, began trading convertible bonds from his Harvard dorm, the finance version of the tech "entreprenerd."
- Julian Robertson hired college athletes half his age and raced them up mountains at Rocky Mountain retreats.
- Michael Steinhardt could reduce staff to tears; when one said he wanted to kill himself, Steinhardt asked if he could watch.
Philanthropists and ideologues
Like the robber barons, the new moguls reached beyond finance. Soros was the most ambitious: his foundations backed independent voices in ex-communist states, drug decriminalization and a rethink of laissez-faire economics. Paul Tudor Jones created Robin Hood, an early venture philanthropy that picked innovative New York anti-poverty charities, set hard benchmarks and paid for performance. Bruce Kovner chaired the neoconservative American Enterprise Institute; Steinhardt funded a new secular Judaism. But their largest mark was on finance itself: hedge funds are the frontier of financial innovation and leverage, of triumphs, falls and the debates they spark.
Edge funds: the case against efficient markets
From the mid-1960s to the mid-1980s the academic consensus held that markets are efficient, prices follow a random walk, and winning managers are mostly lucky. The logic is strong: if a price rise were predictable, smart investors would already have bought and the rise would have happened, so prices already contain all known information and only surprises move them. Mallaby concedes that plenty of funds have no real edge once the marketing and the lucky streaks are stripped away. But for the successful funds that dominate the industry the theory is wrong; they "could drop their h and be called edge funds."
- Stock picking. Despite the finance literature, A. W. Jones, Julian Robertson and many Robertson protégés clearly beat the market by choosing better stocks.
- Exploiting the theory's footnotes. Efficiency assumes perfect liquidity: a seller at the fair price always finds a buyer. In the 1970s and 1980s a pension fund dumping a big block had to offer a discount; Steinhardt milked those discounts systematically, turning a footnote into a legend.
- Trends and irrationality. Later research admitted that prices trend, that liquidity is imperfect and that investors are not fully rational, all things traders had long exploited.
Genius does not understand itself
Managers often obscure their real edge. Paul Tudor Jones still credits his call on the 1987 crash to his young colleague Peter Borish, who overlaid the 1980s market on the chart before 1929 and found the lines matched. Mallaby finds this inadequate: Borish admitted massaging the data to make the lines fit, and he forecast the crash for spring 1988, so following him literally would have meant ruin in October 1987. Jones won for other reasons, explored later. The tennis coach Vic Braden made the same complaint about top players: none could consistently explain what they actually did.
Mallaby's method throughout is to test managers' accounts against the record: self-reported explanations of success are unreliable evidence, a caution for anyone tempted to copy a famous trader's stated method.
The academics convert
From the 1980s scholars came round. Economist Scott Irwin obtained a detailed commodity price series from a small Indianapolis firm and, after painstaking work, announced that prices move in trends. Unknown to him, Commodities Corporation had analysed the same data almost twenty years earlier, reached the same conclusion and programmed a computer to trade it. The 1987 crash, when America's corporate valuation shifted by a fifth in one day, made it hard to defer to market prices; in 1990 Harvard's Andrei Shleifer and Lawrence Summers mocked that the efficient-markets hypothesis had crashed with everything else on 19 October 1987.
The academy's turn changed the industry's personnel. Where professors once predicted hedge funds would fail, they now queued to join them:
- Asness studied at Chicago under Eugene Fama, a father of the efficient-market hypothesis, who by 1988 was leading the revision with Kenneth French by finding non-random, tradable patterns.
- Myron Scholes and Robert Merton, Nobel laureates whose option-pricing formula grew out of efficient-markets theory, joined Long-Term Capital Management.
- Shleifer co-founded the investment firm LSV with two fellow professors.
- Summers, between running Harvard and advising President Obama, worked for the quant fund D. E. Shaw.
The biggest effect of the new consensus was not academics moving to hedge funds but institutions acquiring permission to fund them. Before 1987 most money came from rich individuals; afterwards it came from endowments, told by consultants that the market could be beaten. David Swensen of Yale reasoned that systematic patterns could be harvested by funds identified in advance, and that their returns would reduce overall risk through diversification, the uncorrelated returns called alpha.
Swensen's managers were diverse indeed: in 2002 the West Coast fund Farallon bought Indonesia's largest bank, unfazed by a currency collapse, a political revolution and Islamist extremism that had scared most Westerners away.
Stabilizing heroes or weak links
The inefficient-market view also gave hedge funds a social function they never sought. If alpha exists, savings are being misallocated:
- Value versus growth. Fama and French showed dull value stocks were underpriced relative to hyped growth stocks, so solid firms paid too much for capital and flashy ones too little. Asness, by buying value and shorting growth, narrowed that bias.
- Block discounts. Capricious prices raised risk for investors, who then charged more for capital. Steinhardt buying Ford after an illogical block-sale dip assured the grandmother holding Ford stock a fair price.
- Automation. Jim Simons and David Shaw computerised Steinhardt's art through statistical arbitrage.
The more efficient the market, the better capital flows to productive uses and the fewer bubbles and violent corrections occur. In this telling, hedge funds contributed to the Great Moderation.
The opposite worry: in markets prone to bubbles, the wildest players might make things wilder.
- 1994. A tiny 0.25-point Fed rate rise sent bonds into a spin; leveraged funds caught on the wrong side dumped positions, the turmoil spread to Japan, Europe and emerging markets, several funds sank, and for some hours even Bankers Trust looked in danger.
- 1998. LTCM and its Nobel laureates failed; fearing a chaotic bankruptcy would topple Lehman Brothers and others, regulators organised the burial.
- Currency raids. After the East Asian crisis, Malaysia's Mahathir Mohamad said decades of national development had been wrecked by "a moron like Soros."
By 2000 the two views had fought to a stalemate, and the heart of the argument was leverage, a much larger version of Jones's. Borrowed money gave funds the firepower to trade heavily and correct prices, but it also meant losing trades could burn through thin capital and force fire sales. Britain and the United States stressed the stabilizing side, others the destabilizing one; the countries that liked hedge funds most, Mallaby notes drily, were the ones that hosted them.
The crisis of 2007-2009
The crash produced the sharpest recession since the 1930s, and hedge funds were swept in. In July 2007 the credit fund Sowood blew up; in August about a dozen quant funds tried to cut positions at once, whipsawing equities and losing billions. 2008 was worse: Lehman's bankruptcy trapped hedge-fund money inside the shell; nobody lent after the Lehman shock; governments imposed clumsy short-selling bans; investors demanded their money and some funds blocked withdrawals with gates. It looked as if hedge funds had produced not the Great Moderation but the Great Cataclysm.
Why the crisis does not condemn hedge funds
Mallaby calls that conclusion almost certainly mistaken: the system is broken, but hedge funds are not the core problem.
- Monetary failure. The Fed targeted consumer prices and ignored asset bubbles, a lesson the 1994 blowup had already hinted at. With tighter credit in the mid-2000s, household debt would not have climbed from 66% of GDP (1997) to 100% a decade later, mortgage lenders would not have ignored repayment capacity, Fannie Mae and Freddie Mac would probably not have collapsed, and Citigroup and Merrill Lynch would not have gorged on bad mortgage securities.
- Too big to fail. Bailouts went to Citigroup, Goldman Sachs, Morgan Stanley, AIG and guaranteed money-market funds, not hedge funds. Even LTCM got no public money. A supersized fund could become systemic, so the largest and most leveraged should be regulated, but most are too small to matter.
- Symmetric pay. High-water marks mean no performance fee until past losses are recouped, and managers invest their own money; bank traders mostly gamble with other people's.
- Modest leverage. The average fund borrows 1-2 times capital, "highly leveraged" ones under 10 times; Goldman and Lehman ran about 30 to 1, and Citi higher by some measures.
- Paranoid structure. Scrappy owners; no deposit insurance to gather savings; no central-bank lender of last resort, so less reliance on short-term funding; mark-to-market books that register risk immediately; no underwriting or advisory fees to conflict judgment.
That last point leads to his definition: true hedge funds are independent. Bank-owned "hedge funds" lack the paranoia and focus that give the real thing its character.
The prescription
Mallaby grants that regulators in early 2010 had good instincts: finance had absorbed too much human talent and taken risks that cost societies dearly. But he asks who will manage risk better: banks that blew up or were rescued, or fund companies whose money-market products needed a government guarantee? Bailouts have deepened moral hazard, since rescued banks expect rescue again. Capitalism works only when firms absorb the consequences of their own risks; when they keep the upside and socialise the losses, failure becomes almost certain. Policy should therefore restrain financial supermarkets, favour focused boutiques that live or die by their risk management, move capital out of taxpayer-underwritten institutions, and shrink firms too big to fail. The history of Jones and his successors, he concludes, shows that a partial alternative already exists.
Mallaby's contrast maps onto modern prudential tools. Bank leverage of 30:1 is what Basel III's leverage ratio and capital buffers now target; the high-water mark is a clawback-style alignment that banks later imitated with bonus deferral and malus. His point that hedge funds mark to market while banks wait for defaults foreshadows expected-loss accounting (IFRS 9 / CECL).
Mallaby wrote before Dodd-Frank took shape and with sympathy for his subjects. "Safe to fail" relies on funds staying small; he concedes a supersized fund could become systemic, and later episodes (Archegos's 2021 collapse cost banks over $10 billion) showed that leverage supplied by prime brokers can carry hedge-fund-style risk back into banks.
- Industry size. Per HFR, global hedge-fund capital hit a record $5.6 trillion at end Q2 2026, roughly triple the 2007 peak.
- AQR. Assets peaked at $226 billion (2018), fell sharply in the "quant winter," and recovered to about $179 billion by December 2025.
- People. Steven Cohen's SAC Capital pleaded guilty to insider trading in 2013 and paid $1.8 billion, reopening as Point72; Jim Simons died in May 2024; Julian Robertson in August 2022.
Big Daddy
Hedge funds were not invented in the 2000s boom: Alfred Winslow Jones built the template in 1949, and by 1968 he had compounded almost 5,000%. Mallaby shows that Jones's founding theory, market timing by chart reading, lost money; his real edge was a hedged structure plus a Darwinian system that paid stock pickers for measured performance, and that edge dissolved once imitators copied it and his own managers abandoned hedging in the go-go market of the late 1960s.
The 2000s were a rerun
On the eve of the first crash of the twenty-first century, a few dozen hedge-fund managers had become the unofficial kings of capitalism. Globalization made wealth, the wealth was parked in quiet funds, and in just the three years 2003-2006 the money in the top 100 hedge funds doubled to $1 trillion, enough to buy every share on the Shanghai Stock Exchange or a full year of Canada's output. Everyone assumed the phenomenon was brand new.
Mallaby shows it was not, by setting two New York magazine articles side by side. The 2004 piece called running a hedge fund "the going Wall Street dream" and praised managers who call the market right 22 days in a row; the 1968 piece called hedge funds the logical extension of the "go-go cult of success" and celebrated a manager making 20% a week for seven weeks. Both complained of secrecy. The hedge-fund manager was the It Boy of the 2000s, displacing the buyout barons of the 1980s and the dot-com wizards of the 1990s, but he had been the star of an earlier era too, as "Adam Smith"'s account of the 1960s boom attested.
An unlikely patriarch
The 1968 article named Jones the industry's "big daddy." In 1949, money management was ruled by cautious "trustees" at firms like Fidelity and Prudential, whose job was to conserve capital, not grow it; John Brooks pictured the ideal trustee as sober, white-haired and unsmiling. Jones was different: a restless experimenter with careers, a friend of writers and artists, and a former flirt with Marxism who went on to found the most capitalist of financial vehicles.
1900 Born at the ninth hour of the ninth day of the ninth month, son of the American who ran General Electric's Australian operations (family lore: they owned Australia's first car); raised in Schenectady, New York.
1923 Graduates from Harvard, drifts: a year as purser on a tramp steamer, then export buyer, then statistician for an investment counsellor.
1930 Passes the foreign-service exam; arrives in Berlin as vice-consul in December, as German output falls 8%, unemployment hits 4.5 million and the Nazis win 107 Reichstag seats.
1931 Meets Anna Block, anti-Nazi socialite from a Jewish banking family, working for the Leninist Organization.
1932 Secret marriage discovered; forced to resign in May; back in Berlin that autumn as "Richard Frost."
1933 In London as "H. B. Wood," urging a pacifist-leaning Labour Party to confront Hitler.
1934 Divorced; enrols in a Columbia sociology PhD; marries Mary Elizabeth Carter of Virginia.
1937 Honeymoon in civil-war Spain, hitchhiking to the front with Dorothy Parker; Hemingway gives them a bottle of Scotch.
1938-39 Directs 1,700 field interviews in strike-torn Akron, Ohio.
1941 Thesis published as Life, Liberty and Property, a standard sociology text; joins Fortune.
1942 Fortune essay warns that wartime statism must be dismantled after the war.
1949 "Fashions in Forecasting" appears in Fortune (March); his fund is already running.
Berlin and the underground
In Berlin Jones wrote two studies on German workers' food and housing. Anna Block, his entry to the resistance, was a resourceful figure: she once evaded the Nazis by working out of a hospital maternity wing, and later, in the Paris underground, bet she could bluff her way into London's finest hotel with only a cardboard box for luggage. When Jones met her she was seeking a third husband, and he became the servant of her political and personal aims. The British grew suspicious when they learned he had attended the Marxist Workers School, run by the German Communist Party; a State Department official confirmed that Jones had shown an interest in communism while a diplomat. The resistance proved more romantic than practical, and so did the marriage, which ended within months. Jones kept ties to the German left through the early 1940s and may have worked with US intelligence.
From radical to centrist
Mallaby's striking observation is that Jones came out of the turmoil of the Depression (Wall Street so quiet you could hear backgammon dice through the windows) more level-headed than before. His doctoral question was whether the European calamity could happen in America: how far Americans' economic conditions shaped their attitudes to property, whether the nation was united or "a house divided." His Akron interviews, tested statistically, showed that sharp economic divisions did not produce polarized worldviews, a repudiation of his youthful socialism and a vote of confidence in American democracy. At Fortune he blended right and left: protect the free market as conservatively as possible and pursue welfare as radically as necessary.
The forecasting premise, and Cowles's warning
A 1948 Fortune assignment turned Jones to finance. The resulting essay, "Fashions in Forecasting" (March 1949), attacked the old method of valuing stocks from economic data such as freight-car loadings and commodity prices, citing sharp market moves with no change in the data. He argued instead that investor psychology creates predictable patterns: money is a medium for greed, fear and envy, a barometer of the crowd, a premise natural to a sociologist.
In his model, a feedback loop drives prices: a rising market breeds optimism, which lifts prices further, creating a trend that can be ridden until prices reach an unsustainable level and greed flips to fear. The forecasters he profiled had chart signals for these turning points: the Dow rising while most stocks fall, or prices rising on falling volume, meaning buyers are running out. All believed charts held the secret because patterns repeat.
Jones read the academic evidence selectively. Alfred Cowles, a father of statistical economics, had reviewed thousands of professional recommendations in 1933 and 1944; the first paper, "Can Stock Market Forecasters Forecast?", carried a three-word abstract: "It is doubtful." Jones cited Cowles's faint evidence of monthly trends but omitted that Cowles found no trends at three-week intervals and judged any pattern too weak to trade profitably.
The two men did agree on one thing: forecasting destroys the trend it predicts. If a seer could tell that prices would rise for days to a certain level, money would rush in and push prices there at once, killing the trend; forecasters would speed up the market and work themselves out of a job, leaving prices to move gently with fundamentals.
Jones unknowingly forecast the history of hedge funds: wave after wave of innovators found profit opportunities, and once enough investors understood an insight, the profit faded as markets grew more efficient. Jones himself would make markets more efficient, but not in the way he expected.
The hedged fund
Jones was not drawn by love of finance; he was more absorbed by politics and his Connecticut garden. In his late forties, with two children and expensive New York tastes, he needed money. He had left Fortune to start a magazine, but two blueprints found no backers. Plan B: he raised $60,000 from four friends, added $40,000 of his own, and began investing.
His next twenty years produced one of the most remarkable records in history. By 1968 his cumulative return was just under 5,000%: $10,000 invested in 1949 had become $480,000. In the five years to 1965 he returned 325% against 225% for the hottest mutual fund, and over ten years to 1965 he earned almost twice as much as his nearest competitor; by some measures he rivalled Warren Buffett.
The office was a shabby one-and-a-half rooms on Broad Street, rented from the insurance business of investor Winslow Carlton, a dapper man who sometimes drove Jones down the East Side from his 30 Sutton Place apartment in a roofless Packard convertible, the pair swapping market predictions. The furniture: a Royal typewriter, a dictionary on a stand, a glass-domed stock ticker, a hand-cranked calculator and a couch for post-lunch naps.
"Speculative means for conservative ends"
Conventional managers loaded up on stocks when bullish and held cash when bearish. Jones improved both options. Bullish, he borrowed to be, say, 150% long. Bearish, he sold short instead of retreating to cash. Both techniques had been used by 1920s operators with their own money, but 1929 gave them a racy reputation unfit for trustees of others' savings. Jones's innovation was to combine them without raciness: by keeping some shorts as routine insurance, even when the charts were calm, he protected the portfolio against market falls and so could own more good stocks than an unhedged buyer for the same risk. A traditional manager had to sell hot names like Xerox or Polaroid when the market wobbled; a hedged fund could keep profiting from stock selection even when the market looked dear.
- Leverage in bull phases. Borrow to hold stocks worth more than capital, e.g. 150% long.
- Shorts instead of cash. Reduce exposure by selling borrowed stock rather than holding cash.
- Permanent hedge. Keep some shorts at all times, freeing the fund to buy more good stocks.
- Volatility weighting ("velocity"). Size each position by its swings relative to the S&P 500.
- Return attribution. Separate profit from stock selection and profit from market exposure.
- Separate risk from selection. One decision about which stocks to own, another about how much risk, adjusted via leverage.
The 1961 prospectus example
In a 1961 private prospectus Jones illustrated the magic. Two investors each have $100,000, equal stock-picking skill and a bullish view. The conventional one puts $80,000 into his best stocks and $20,000 into bonds: net exposure $80,000. Jones's investor borrows $100,000 for a $200,000 war chest, buys $130,000 of good stocks and shorts $70,000 of bad ones. He is better diversified (more names on the long side) and has less market exposure (net $60,000). Now assume both pick longs that beat the index by 10 points, and the hedger's shorts lag the index by 10 points.
| Scenario | Conventional ($80k long) | Hedged ($130k long / $70k short) |
|---|---|---|
| Market +20% (longs +30%, shorts +10%) | +$24,000 = +24% | +$39,000 - $7,000 = +32% |
| Market -20% (longs -10%, shorts -30%) | -$8,000 = -8% | -$13,000 + $21,000 = +8% |
The result seems to defy the rule that higher returns require higher risk: the hedger earns a third more in the bull case despite taking less risk, and does better in the bear case because of it. The catch, which Jones acknowledged, is that a poor stock picker has his incompetence magnified.
Why nobody copied it
- Stigma. Jones called short selling a little-known procedure that "scares away users for no good reason." Its taint since the crash survived into 2008, when regulators restricted it. He argued that short sellers dampen bubbles by selling overpriced stocks and cushion falls by buying them back, a defence hedge funds would repeat for decades.
- Asymmetry and treatment. Picking shorts is the same analysis inverted (slow earnings growth, charlatan managers), but shorting faced tougher tax and regulatory treatment, and its maximum gain is 100% (the stock going to zero), against unlimited upside on a long.
- Volatility matching. A real hedge needs a refinement few understood, and this is where Jones was far ahead.
Velocity: sizing by volatility
Buying $1,000 of an inert stock and shorting $1,000 of a jumpy one is no hedge: if the market rises 20%, the first may rise 10 points and the second 30. So Jones measured each stock's swings against the S&P 500, calling it "velocity." Sears Roebuck's significant moves since 1948 were 80% of the market's, a relative velocity of 80; General Dynamics scored 196. To short 100 General Dynamics shares at $50 while staying market-neutral, the fund needed 245 Sears shares at $50. Velocity said nothing about whether a stock was a good buy, only how big a position must be to have a given effect on the portfolio. Jones's small staff computed it by hand for about 2,000 firms every two years.
- Longs. $130,000 should have gained $1,300; they gained $2,500, so stock selection added $1,200 (1.2%) on $100,000 of equity.
- Shorts. $70,000 should have lost $700; the loss was only $400, so short selection added $300 (0.3%).
- Market. Net long $60,000 × 1% = $600 (0.6%).
- Total. $2,100 = 2.1%, of which 1.5 points came from selection (later called alpha) and 0.6 from market exposure (beta).
Each evening, sometimes helped by his children, Jones copied closing prices from the World Telegraph or the Sun into a dog-eared leather book in pencil and ran this reasoning, using the statistical skills honed in Akron.
Ahead of the professors
Mallaby's claim is that Jones's rough methods anticipated three Nobel-winning breakthroughs of the 1950s and 1960s.
- Markowitz (1952). Harry Markowitz, a 25-year-old graduate student, founded modern portfolio theory in "Portfolio Selection": maximize risk-adjusted return, and measure risk by the correlations among holdings. Jones's velocity weighting and balanced longs and shorts were a crude version. But Markowitz was unusable on Wall Street: 1,000 stocks needed almost half a million correlation calculations, and in the mid-1950s even 25 stocks overwhelmed the memory available to Yale's economics department.
- Sharpe (1963). William Sharpe's "A Simplified Model for Portfolio Analysis" replaced the full correlation matrix with one correlation between each stock and the market index: precisely Jones's velocity, which he had used for more than a decade.
- Tobin (1958). James Tobin's separation theorem held that stock choice should be separate from risk appetite. The 1950s adviser kept widows out of go-go Xerox and executives out of stodgy AT&T; Tobin said both should buy the same best stocks, the cautious one committing part of his savings, the bold one borrowing to buy more. Jones had run his fund this way nine years earlier.
Secrecy, tax and regulation
Jones's undercover years had taught him to stay under the radar, and he had three motives for doing so.
- Protecting ideas. Visiting brokers were cross-examined about their tips but left with no clue what the Jones men thought.
- The 20% fee as a tax device. Lawyer Richard Valentine of Seward & Kissel, a creative but absent-minded genius (he once explained a tax idea at length to a colleague's five-year-old on the phone), saw that a share of profits instead of a flat fee could be taxed as capital gains: about 25% rather than up to 91% at the top income rates of the day. Jones charged 20% of the upside, claiming it copied Phoenician merchants who kept a fifth of a voyage's profit. Labelled a "performance reallocation" to avoid ordinary income tax, it was adopted by every generation since.
- Escaping regulation. Jones declined to register under the Securities Act of 1933, the Investment Company Act of 1940 and the Investment Advisers Act of 1940, arguing his funds were private. Registration would have restricted borrowing and short selling, the two pillars of his strategy, and capped fees. He never advertised, raised money by word of mouth (often over dinner) from intellectual friends such as Lenin biographer Louis Fischer and Sam Stayman, inventor of a bridge bidding convention, and in 1961 set up a second partnership rather than let the first exceed 100 members.
The cost of stealth was lasting suspicion. By the mid-1960s John Brooks was calling hedge funds Wall Street's last bastions of secrecy and privilege, "the parlor cars of the new gravy train." Thanks to the pattern Jones set, Mallaby notes, hedge funds have been mysterious and resented ever since.
Where the money really came from
Ignoring Cowles cost Jones. His market calls failed as often as they worked:
- 1953, 1956, 1957. He lost money on timing, levering up when the market did poorly and vice versa.
- 1960. Cowles revised his own research and decided that even faint monthly trends did not exist; Jones carried on regardless.
- Early 1962. Jones was 140% net long when the market fell; he turned bearish and it rose.
- August 1965. He was net short 18% of capital just as a hot rally began.
Later traders would prove trend following could pay, and later academics would revise Cowles, but Jones never profited from the charts that were his fund's founding premise. His own attribution system showed exactly how much the bad calls cost. The hedge structure was brilliant but not itself a profit source. The profits came from the people he hired to fill it.
The Darwinian stock-picking machine
Jones knew he was no stock picker: an investment novice bored by balance sheets. So from the early 1950s he had brokers run paper "model portfolios" of favourite longs and shorts, phoning in changes as if managing real money. His attribution method measured each one's results precisely, and he paid brokers by performance, which made them call Jones with hot ideas before anyone else.
Wall Street in the 1950s was sleepy. Hardly any students took finance; Harvard's investment course was nicknamed "Darkness at Noon" because it was given the unpopular lunchtime slot. Trustees were paid on assets under management, not results, and decided by committee. Jones broke the mould: individualism instead of collectivism, adrenaline instead of complacency. By the 1960s, with half a dozen in-house stock pickers, he held few meetings (committees bored him), gave each manager a segment of capital and a target market exposure, and rewarded the best performers most.
- Paid research. Brokers' model portfolios were tracked and paid generously; a young broker could see his salary double if his picks made money.
- Merit-based capital. Successful segment managers received more capital the next year, raising their chance of a big bonus; laggards received less.
- Fee split by results. Performance fees were divided according to whose segment did best.
- Skin in the game. Partners had to keep their own money in the funds, so wealth as well as income rode on results.
- Top-down risk control. Jones set each segment's market exposure centrally while leaving stock choice to the manager.
Mutual funds, by contrast, had no system to track which tips worked, and spent thousands paying salesmen to gather assets, leaving little to reward research. Information then spread slowly (no blast emails, no cable pundits), so hustle paid. Jones man Alan Dresher went to the SEC to read company filings the moment they were released; astonishingly, he was alone, while the rest of the Street waited for the filings to arrive by post.
Without grasping its significance, Jones created the competitive multimanager structure, Darwinian individualism plus top-down risk control, later reinvented by a Princeton firm in the 1970s (Chapter 3) and copied by dozens of funds. Markets may be efficient given existing institutional arrangements; Jones won by blowing up those arrangements, scrapping committees and paying people to perform.
Unravelling
The aloof dilettante
In early 1964 Jones, now in his sixties and graduated from a Dodge station wagon via a Citroën DS to a huge Mercedes, invited a young analyst to lunch at his club and asked whether he washed his hands before or after using the urinal. "After," the analyst said; wrong answer, Jones replied, that was conventional and irrational. He was mangling a joke then going round. The analyst, future Wall Street grandee Barton Biggs, took against him at once: Jones seemed aloof, conceited and ignorant about stocks, harvesting young men's work while turning up sporadically. Biggs ran a model portfolio but never liked him.
Jones disdained the monomaniacs who simply went on making money. He believed Edward de Vere, 17th Earl of Oxford, wrote Shakespeare and named his poodle Edward; tunnelled through his weeping willows and nursed his grass tennis court like a sick infant; founded a philanthropy for the "humiliated poor"; and began a sequel to Michael Harrington's The Other America. Dinner guests were intellectuals and UN diplomats discussing Russian power in Yugoslavia, not stocks. Committed Wall Streeters resented him.
Defections and imitators
Soon after the Biggs lunch, an in-house manager left to found the rival City Associates: a rational move, since the hedged concept was easy to copy and there was no need to share the spoils with a dilettante. Jones set his lawyers on the defector, then spent Christmas 1964 on a Himalayan tiger shoot with Indian UN friends, complete with elephant rides and Hindu hosts singing carols. Meanwhile Biggs persuaded Jones's longest-serving manager to quit and start another fund, taking clients with him.
Every great investor's edge eventually unravels as rivals copy it. Jealousy over sharing the profits led to more defections, and an industry was born.
- 40 → 200-500. Imitator firms at the start of 1968, and estimates for 1969; many leaders had worked for Jones or been his brokers.
- $11 billion. Industry assets per The Economist, five times the figure two years earlier.
- "Hedge fund." The corrupted form of Jones's "hedged fund" entered the Wall Street vocabulary.
- 26%, 22%, 47%. Jones's net returns in the three years from summer 1966.
- "Something in the millions." Jones's reported earnings; a City Associates partner bought a penthouse, helicopter, wine cellar and bodyguards.
Paradoxically, the copying first raised Jones's profile. A flattering 1966 Fortune profile called him possibly the best money manager of the day, investors scrambled to get in, ambitious analysts came seeking jobs, and in 1968 New York crowned him "big daddy."
Regulators circle
As in later booms, regulators took notice. In 1968 the NYSE and AMEX considered limiting their members' dealings with hedge funds; in January 1969 the SEC sent a questionnaire to 200 hedge funds asking who they were, how they were organised and what their trading did to the market, with the aim of forcing registration. Mallaby finds the complaints not quite rational: hedge funds held half the short interest in some stocks, yet nobody explained why restraining overvaluation was harmful; they turned portfolios over faster than mutual funds, yet extra liquidity was treated as a vice; a Jones manager supposedly bought a block of a go-go stock in the morning and sold it after lunch, harming no one.
The hedged fund drops its "d"
The strong late-1960s returns hid decay. Stock pickers were leaving, and those who stayed, swept up by the bull market, treated shorting as a sucker's game, stopped hedging against an S&P fall, pushed leverage to the limit and ignored velocity limits on hot, volatile stocks. With Jones increasingly absent, the young men did what they liked: the hedged fund dropped its "d."
Jones had also been lucky. He launched as the crash trauma faded: stock owners rose from 1 in 25 American adults in 1950 to 1 in 8 by the end of the decade, retail brokerages spread, and the S&P 500 climbed from 15 at his 1949 launch to 108 in late 1968. The trustees gave way to go-go managers who believed the Fed, the SEC and Keynesian budgets had abolished the business cycle. Jerry Goodman ("Adam Smith") captured the mood in the Great Winfield, a semi-fictional investor who hands his money to twentysomethings whose virtue is having no memory and no fear.
The multimanager structure had no brake. The more the market rose, the more the performance system rewarded the managers taking the most risk, a flaw later multimanager funds would rediscover. In May 1969 the market began to fall, losing a quarter of its value over the next year. For the year to May 1970 Jones told clients he had done worse than the market: down 35%.
At his 70th birthday in September 1970, under a marquee on his country lawn, with a grandchild on the way and his daughter's engagement announced, Jones kept fretting that his managers would see him spending money "like a drunken sailor." After two decades, his edge was gone; the markets had caught up with him.
Jones's nightly ledger is performance attribution in embryo: decomposing P&L into beta (market) and alpha (selection) is now standard in bank trading-desk and asset-manager reporting. His velocity sizing is a beta-adjusted hedge ratio. His failure is the classic lesson that incentive schemes rewarding recent P&L encourage risk drift unless limits are enforced centrally.
The source text omits the two return tables and Jones's velocity quotation (likely printed as images); the table above is reconstructed from Mallaby's stated assumptions and ignores borrowing costs and bond interest. Mallaby's "first hedge fund" label is also contestable: Benjamin Graham's Graham-Newman partnership (1920s-1950s) already shorted stocks and took a profit share, though without Jones's systematic hedge and leverage. The 1966 Fortune profile he quotes was written by Carol Loomis.
The Block Trader
The inflationary shocks of 1969-1974 wiped out the first hedge-fund generation, but Steinhardt, Fine, Berkowitz thrived by actually hedging, by betting against consensus, by reading monetary data, and above all by supplying liquidity to the new institutional investors through block trading. Mallaby's judgment is that Steinhardt's edge was real, partly collusive, and on balance stabilizing: the service he provided outweighed his wrongdoing.
The end of the first era
For two postwar decades American finance had been almost quaintly stable: the dollar was pegged to gold and interest rates moved in a narrow, regulated band. Inflation broke that world. Below 2% in the first half of the 1960s, it reached 5.5% in spring 1969; the Federal Reserve slammed on the brakes and crushed the stock market. In 1971 the Nixon administration conceded that inflation had eroded the dollar and abandoned the gold standard, so money's value could change day to day, fuelling still more inflation. Another squeeze followed and the market crashed again in 1973-74. Go-go, as Mallaby puts it, was gone.
- Two thirds. Capital lost by the 28 largest hedge funds between the end of 1968 and 30 September 1970; their "hedged" label had been a lie, their performance built on borrowing and the bull market.
- 150. Funds left by January 1970, down from 200-500 a year earlier; the 1973-74 crash killed most of the rest.
- 1977. Institutional Investor asked where all the hedge funds had gone; the SEC had dropped its regulatory campaign against a sector now too small to matter.
- 68. Hedge funds a Tremont Partners survey could find as late as 1984.
- $100m+ → $35m → $25m. A. W. Jones's assets: late 1960s, 1973, and a decade later. Performance fees vanish in an era of non-performance.
The troika
The first winner in the new uncertainty was Steinhardt, Fine, Berkowitz & Company. Its dominant partner, Michael Steinhardt, became a legend for his trading and his personality. Raised by a single mother in tough Brooklyn, with a confrontational, gambling-addicted father whose traits he later brought to trading, he entered the University of Pennsylvania at 16, graduated at 19, and by about 25 was, in his own estimate, the "hottest analyst on Wall Street." Short, barrel-chested and volcanic, he turned red from face to temples before unleashing abuse that left colleagues shaking.
In 1967 he quit broking and launched a fund with two equally young friends, Jerrold Fine and Howard Berkowitz. At first they looked like any Jones imitator: a pool table in the office, a creed of youthful superiority, and a lawyer's warning, ignored, that the name sounded like a Jewish delicatessen. In their first full year they loaded up on the era's story stocks: King Resources (a charming CEO claiming new oil and gas finds), National Student Marketing (a bet on the youth market) and tech names ending in "Data" or "-onics." They finished up 84% after fees, and each founder became a millionaire.
Hedging for real, then going short
In 1969 the bull market ended; other go-go funds blew up, and Fred Mates, 1968's top mutual-fund manager, ended up tending bar. Almost alone, the troika had sensed that the postwar expansion had overreached. Fifteen years earlier Benjamin Graham had declined to buy Haloid (the Xerox maker), seeing no reason to pay for future growth; Haloid had sextupled in two years and the growth-stock cult had been unquestioned since. In 1969 the troika judged the cult had gone too far: paying a premium for bright prospects was one thing, extrapolating supersonic growth forever another.
| Period | Steinhardt, Fine, Berkowitz | Market |
|---|---|---|
| First full year (1968) | +84% net; each founder a millionaire | go-go boom |
| 1969 | all but a sliver of capital kept (genuinely hedged) | S&P -9% |
| 1970 | gain | S&P -9% |
| July 1967 to 1971 | +361%, 36 times the index; only fund in Fortune's top 28 to grow | |
| Year to Sept 1972 | -2% (net short, waiting) | S&P +9% |
| Year to Sept 1973 / 1974 | +12% / +28% after fees | S&P -2% / -41% |
| 1975 | +54% after fees, turned bullish | strong rebound |
| 1967 to fall 1978 | $1 → $12, 24.3% a year net | $1 → $1.70 |
At the start of 1969 they shorted enough story stocks to balance their longs: unlike most "hedge funds," they really hedged. They kept almost all their capital in 1969, made money in 1970, turned bullish in 1971 to catch the bounce, and were the only fund on Fortune's 1971 list of the top 28 to have grown through the shakeout.
In 1972 they turned bearish for deeper reasons. America's postwar confidence was cracking: the Nixon administration was covering up Vietnam, masking inflation with an unworkable wage-and-price-control programme, and blue-chip companies were masking their numbers with accounting tricks. Fine and Berkowitz, who led research, kept finding red flags in annual-report footnotes, and Leonard Spacek, chairman emeritus of Arthur Andersen, called financial statements "a roulette wheel." Berkowitz dismissed early-1970s research reports as simplistic nonsense; Fine recalled capital gains dressed up as operating earnings. The market priced political and financial delusions, so this time the troika went heavily net short and waited.
The crash came late. The fund was down 2% in the year to September 1972 while the S&P rose 9%. Then the payoff: the S&P fell 2% and then 41% in the years to September 1973 and 1974, while the troika made 12% and 28% after fees. Desperate sellers phoned, urging them to cover their shorts and support prices; Steinhardt, who ran trading, refused. The old prejudice returned: they were called arrogant, greedy and un-American, betting against American companies as near-treason. Steinhardt remembers the vilification as the peak of his professional satisfaction.
Contrarian by temperament
The troika showed the contrarian capacity of later hedge funds. A. W. Jones invented the structure to control market exposure; his go-go imitators used it to maximize exposure; the troika made it a vehicle for betting against conventional wisdom, and aggressive contrarianism became a company creed.
Steinhardt enjoyed baiting people. He showed guests his upstate farm's Falkland flightless steamer duck, which bit anyone who came close; he phoned brokers with juicy orders for nonexistent stocks and left them scrambling. In the 1980s he let a psychiatrist offer "organizational therapy"; staff interviews produced phrases like "battered children" and "rage disorder," and the therapy ended when Steinhardt lost his temper and threw the man out.
He loved offending the consensus even more: a Republican in the liberal 1970s, leaning Democrat in the Reagan 1980s, a funder of the centrist Democratic Leadership Council that helped elect Bill Clinton, then its opponent once Clinton won. He professed no faith in God yet gave millions to Jewish causes. In markets he trawled for unconventional ideas and backed them on a terrifying scale.
New analyst John LeFrere visited IBM and urged buying before Friday's quarterly results. Steinhardt, who disliked the stock's aimless action on the ticker, said he hated it; LeFrere, a big man who reckoned he could out-bench his boss, insisted the results would be good. Asked how much, LeFrere proposed 10,000 shares (about $3.5 million at $365, the most he thought conceivable). Steinhardt ordered 25,000, demanded to know how convinced LeFrere was, and bought another 25,000: roughly $18 million, perhaps a quarter of the firm's capital and five times the suggestion. IBM jumped 20 points on the results, an instant $1 million profit.
The quality Steinhardt prized above all was the nerve to take a position. Early on the troika's big bets rested on plain intellectual confidence: all three had studied at Wharton, knew they were very smart, and happily shorted any firm whose footnotes hid something.
Tony Cilluffo, the eccentric
In 1970 Steinhardt hired Frank "Tony" Cilluffo, a mathematically gifted autodidact from Wall Street's wild fringe: a Brooklyn City College dropout who had spent his youth devising a horse-race prediction system, then worked at Arthur Lipper and Company, a brokerage known for serving the notorious con man Bernie Cornfeld. He was a devotee of Kondratiev wave theory.
- The claim. Capitalist economies move in long cycles: upswings during bursts of innovation and abundant investment, downswings as investment dries up and old capital loses value.
- The inventor. Nikolai Kondratiev founded Moscow's Institute of Conjuncture in 1920.
- The dated upswings. 1789-1814, 1849-1873 and 1896-1920.
- Cilluffo's forecast. Advances last about 24 years, so the postwar bull market that began in 1949 would hit the rocks in 1973.
- Mallaby's verdict. Nobody could say why innovation cycles should repeat so precisely across centuries: "conjuncture" rested mainly on conjecture. The mainstream's scorn only increased Cilluffo's faith, which suited a boss who wanted contrarians and did not care how they got there.
Cilluffo sat with Steinhardt on the trading desk amid unfinished lunches, both chain-smoking. Cilluffo timed a Dunhill at eight minutes thirteen seconds and smoked four packs a day; Steinhardt lit two cigarettes at once in bad markets. Where intensity made Steinhardt explode, it made Cilluffo superstitious: during winning streaks he ate the same lunch every day (two toasted English muffins with jam for two years, then a long run of cream-cheese-and-olive sandwiches), gulped coffee, and declared that an $80 stock would hit $10 by summer. A colleague thought him brilliant or crazy; Steinhardt's memoir says Tony had a direct line to God, if there was one.
The 1973 prediction came true, whether by luck or insight. That bear market finished off the postwar order; the rest of the decade brought stagflation, the number of American stockholders fell by seven million, and a Business Week cover in summer 1979 announced "The Death of Equities." The partnership prospered anyway: after 12% and 28% in the bear market it turned bullish in time to make 54% in 1975, and stayed solid for three more years.
By its eleventh year the record was among the best ever: $1 in 1967 became $12 by 1978, against $1.70 in the market, a net 24.3% a year, matching A. W. Jones in his heyday, but earned against an economic headwind.
Luck or edge?
Mallaby puts the efficient-market objection squarely. If 200 people each flip eleven coins, about five will get nine heads; among the 200-plus funds launched in the late 1960s, one calling the market right in nine of eleven years (missing only 1969 and 1972, when it lost a fraction) could be chance.
Steinhardt's own explanations are unsatisfying. Markets are inexact, he says, and laymen rival professionals. He credits an "often inchoate judgment," a sixth sense built by daily immersion since his absent father gave him stock certificates for his bar mitzvah at 13. Mallaby finds this plausible but untestable, and weakened because Steinhardt's biggest errors came late in his career, when experience should have protected him. He also cites intensity, an "overriding need to win every day." His tantrums blasted over the firm's "hoot-and-holler" intercom, and he once screamed at an analyst whose stock made a fortune for recommending it too early. Fear may have driven the team forward, but the opposite theory, that his temper choked idea-sharing and drove talent away, seems at least as strong.
Fine and Berkowitz credit better analysis, more care and harder work. Mallaby grants that stock-picking skill drives other famous funds despite academic doubts, but it explains less here than for Jones: by the late 1960s there were dozens of hedge funds, and pension funds and endowments had become far more professional. Jones won with a novel pay-for-performance system; the troika had no monopoly on analysis.
The success is hard to explain, even for the partners, but it was not luck. Over 28 years Steinhardt lost money in only four, odds of about one in eleven thousand. Mallaby's resolution fits a commonsense reading of efficient markets: the market is hard to beat except with an approach that others have not yet exploited. Two such approaches stand out.
Innovation one: monetary analysis of stocks
From the 1960s Cilluffo tracked monetary data to anticipate the stock market. A decade later this was standard: fast money growth predicts inflation, which forces the Fed to raise rates, which pulls money out of stocks into deposits and bonds, so the market falls, and rate-sensitive stocks (home builders, equipment suppliers) fall hardest. But 1960s equity investors had learned their trade when inflation never topped 2%; the Fed was marginal to their thinking. The eccentric outsider was the exception.
By 1970 he had a crude model. He watched the big Federal Reserve member banks: the moment they switched from reporting spare lending capacity to reporting they had hit their reserve limits, lending was maxed out, so money growth would slow, the economy would weaken and stocks would suffer. History showed stocks began falling about two months after the crossover. It also worked in reverse: a switch to free reserves meant an imminent rally.
Cilluffo had grasped the rules of the high-inflation, post-gold-standard world before it fully arrived. He anticipated both the 1973-74 collapse and the rebound, reinforcing colleagues' conclusions from conventional analysis, and because those three years produced the bulk of the firm's 1970s profits, Mallaby credits him with a large share of the decade's success. Colleagues barely understood him because he explained badly. They knew the wiry man on the desk hated Kaufman & Broad, the biggest home builder, and that the firm was short 100,000 shares; they did not necessarily know why (rising rates crush home builders). Steinhardt empowered him anyway, and the short earned over $2 million.
Innovation two: block trading
The second edge came from a shift in who held America's savings. Until the 1960s individuals dominated the market.
- 10 million → 3×+. Workers covered by company pensions in 1950, and more than triple that by 1970.
- $130 billion. Pension-fund assets by 1970, growing $14 billion a year.
- $2 billion → $50 billion+. Mutual-fund assets, 1950 to the late 1960s.
- Under 5% → roughly 15%. Block trades' share of NYSE transactions, 1965 to 1970.
Professionalization made it harder to win on stock knowledge, since six other professionals probably had the same facts. But it created a new opportunity in trading. Before institutions, a small order (say 50 Ford shares) went to the NYSE specialist in that stock, who bought slightly below where he expected to resell. When pension and mutual funds wanted to trade 100,000-share blocks, specialists lacked the capital. Brokers led by Oppenheimer and Goldman Sachs began making markets themselves, matching clients or buying with their own capital.
Why block trading paid
Institutions had little choice but to pay well. Selling 100,000 Ford shares bit by bit would push the price down as they sold, and a leak midway would collapse it, so they preferred to hand the whole block to Goldman or Oppenheimer at a substantial discount. Brokers could earn a fat commission by finding a buyer, or take the stock onto their books and resell later. Their need was a buyer with the guts to play in size.
At most firms trading was a dull back-office job, but at Steinhardt's the boss himself ran the desk. When Goldman or Oppenheimer called with a block at a tempting discount, he bought, and the more he dealt, the more they called. Brokers needed someone who could decide big and fast; junior traders elsewhere lacked authority, while Steinhardt could risk millions on his own say, and, with his father's gambling gene, loved it. Trading, he later said, went from a clerklike function in the 1950s and 1960s to one of great significance in the 1970s and 1980s.
- New markets are inefficient. Brokers shot from the hip, with no rules relating discount to block size and no risk managers yet. Offered 700,000 discounted shares of bankrupt Penn Central, Steinhardt bought and resold them, making over $500,000 in eight minutes.
- Arbitraging a newcomer. Around 1970 Salomon Brothers set out to become the third big block house and absorbed lots at wafer-thin discounts; Steinhardt bought deeply discounted blocks elsewhere and sold them to Salomon.
- The opaque third market. Trades with exchange members (Goldman, Salomon) printed on the tape; trades with non-member brokers in the "third market" were unreported. Steinhardt bought there "down one point" (a $1 discount), then asked a member broker to resell, say, 200,000 of 400,000 shares "up an eighth" (12.5 cents above the tape), so the trade looked positive to buyers who never knew of the earlier discount.
- Paid both ways. He collected fat fees for supplying liquidity (the discount) and paid little or nothing for the liquidity he consumed (reselling above market).
- Network effects. He got the big calls because he was known to get them, and could unload because he generated huge commissions for Oppenheimer, Salomon and Goldman. Unlike the easily copied Jones model, rivals faced a catch-22.
The theory says prices embody all information. That may hold over the medium to long run, but in the short run prices move on liquidity needs, not information: an insurer selling to pay storm-damage claims depresses a price; a pension fund investing fresh inflows lifts one. The models assumed perfect liquidity; Steinhardt was paid to supply it, especially while block demand was surging and market structures had not caught up.
The shady part
Liquidity is costly because a buyer facing a 400,000-share block cannot tell whether the seller knows something: an earnings cut, another institution about to dump. Ignorant of what they do not know, buyers demand a discount. Steinhardt shrank that risk through his scale: he generated so many commissions that brokers bent the rules for him. They were not supposed to identify sellers, but might hint that a seller was a plodding insurance fund (selling for liquidity, so a bargain), steer him away when the seller was a smart hedge fund with bearish information, or warn when a block was the first of a series.
Goldman block trader John Lattanzio recalled a clubby world where nobody tried to hurt anybody. Oppenheimer head trader Will Weinstein described warning a friend not to buy the first hundred thousand because four hundred more were behind it, calling this mutual protection rather than collusion; how he saw it as anything else, Mallaby says, is "a mystery."
Sometimes the collusion was more elaborate. A broker would tip Steinhardt that an institution was about to unload 500,000 shares, so he could get short in advance; when the order hit and the price fell, he covered by buying the tail end at the depressed price. The broker looked brilliant to the seller, since the sale went through near the market price, but the market had already been pushed down by Steinhardt's shorting. The seller had hired the broker to get the best price and was sold out, a clear violation of the rules. Steinhardt, three decades on, is frank: he was told things other clients were not, and got information he should not have.
How much block trading contributed cannot be measured. Steinhardt says it was much of the noise, not the profits, but the more he describes it, the more Mallaby suspects he understates it. It was the firm's most distinctive edge, and surely no coincidence that of the three gifted founders, Steinhardt, the trader, became the legend.
The regulators and the counterfactual
Had the collusion been known, might the SEC have revived its push to regulate hedge funds and changed the industry's history? Collusion was suspected but never proved. The SEC pursued Steinhardt once, alleging that in January 1970 he bought a large block of Seaboard Corporation as a favour to a broker propping its price before a public offering, on an understanding that losses would be covered; he settled in 1976 without admitting wrongdoing.
Mallaby doubts disclosure would or should have triggered a clampdown:
- Existing powers. Regulators could already pursue trading on privileged information, as Seaboard showed.
- Regulators' own fixes. Part of his edge came from oddities he did not create; in 1975 the regulators required all trades on a new consolidated tape, and in 1978 required bids and offers to be reported, dragging the third market into the light and eroding the inner circle's advantage.
- Net stabilization. The SEC's 1969 hunt for "market disruption" found none, and the troika's success stabilized markets.
Harm. Collusion sometimes cost outside investors better prices; since those included mutual and pension funds, the losers were ordinary Americans and the winners Steinhardt's millionaire clients: a "reverse Robin Hood."
Benefit. He sold the 1972 bubble and bought at the end of 1974, when the market badly needed buyers; spread monetary sophistication that made inflation-driven bubbles less likely; and offered liquidity in highly inefficient short-run markets that institutions were free to refuse, and chose to use. On balance, the service outweighed the wrongdoing.
The 1987 proof
If that seems too charitable, Mallaby fast-forwards. After Black Monday in October 1987, the big block desks pulled back and there was an outcry. In December the New York Times reported that 25,000-share blocks now moved prices as much as 100,000-share blocks had before the crash; a mere rumour that an institution was selling Ford or IBM knocked the price down. In a twist on the debates of the 1960s and 1970s, the SEC promised to investigate the troubling lack of block trading. The only thing worse than fast-trading hedge funds, Mallaby concludes, was a sudden shortage of them.
Sabbatical
Steinhardt traded like a man at war, and it exhausted him. In the fall of 1978 he took a sabbatical to shrink his waistline and expand his head, seeking meaning beyond money. A friend scoffed that he was as likely to quit Wall Street as Vladimir Horowitz to quit the piano. He stayed away until the fall of 1979, then stormed back, broke with his partners, and marched into the 1980s.
Steinhardt's business is what bank market-making and risk-capital desks do: warehouse client blocks for a spread. His abuses (trading ahead of a client's order, leaking client identity) are exactly what modern best-execution, front-running and information-barrier rules (Reg NMS, MiFID II, the Volcker Rule's market-making limits) police. His 1987 lesson recurs whenever dealer balance sheets shrink: less dealer capacity means thinner liquidity in stress.
In December 2021 Steinhardt surrendered more than 180 looted antiquities valued at up to $70 million to the Manhattan District Attorney and accepted a lifetime ban on acquiring antiquities, avoiding criminal charges.
Paul Samuelson's Secret
Paul Samuelson told Congress that random portfolios beat professional managers, yet quietly backed Commodities Corporation, a Princeton firm founded by econometricians in 1970. Its fundamental models nearly destroyed it in the 1971 corn blight episode; it was saved by strict risk controls and by trend following, a strategy academics said could not work. Mallaby's lesson is that hedge-fund innovation comes from pragmatic trial and error, not from theory: Commodities Corporation won by going with what worked.
The professor's exception
In celebrated 1967 congressional testimony, Samuelson cited a Yale PhD dissertation showing that randomly chosen portfolios tended to beat professionally run mutual funds; when the House banking chairman balked, he told him to picture dice, random numbers or a dart. In 1970 he became the third economist to win the Nobel Prize, and it did not soften him: in 1974 he wrote that most portfolio managers should quit and take up plumbing or teach Greek, conceding that such advice would not be eagerly followed.
Yet his condemnation left room for exceptions. People differ in height and looks, he wrote, so why not in their "P.Q." (performance quotient)? Such rare giants would not rent themselves cheaply to the Ford Foundation or a bank trust department; they would form small partnerships and keep the gains, in other words hedge funds. Never short of confidence (by 25 he had published more papers than his age in years), Samuelson believed he could spot them. In 1970 he became a founding backer of Commodities Corporation, paying $125,000 for a stake and joining the board, and around the same time invested with Warren Buffett.
That is Samuelson's secret: the high priest of the random walk put his own money behind active traders. His public view (most managers cannot beat a dart) and private bet (a few high-P.Q. outliers can) are consistent, and they frame Mallaby's book: the market is hard to beat except for those with a genuine edge.
A quant start-up
Commodities Corporation was among the first boutiques of hard-core "quants." Page one of its prospectus promised large-scale econometric analysis made possible only by computers. A founding trader was MIT's Paul Cootner, ironically famous for contributions to efficient-market theory, and the firm later hired an Apollo programme programmer: a literal rocket scientist. Legally a corporation, not a partnership, it was otherwise a typical hedge fund: long and short, leveraged, with huge profits shared between managers and a few investors.
- F. Helmut Weymar. Driven, anguished, mildly megalomaniacal president; MIT PhD whose thesis modelled cocoa prices; called the random walk "bullshit" and the idea that no one could profit from an edge "wacko."
- Frank Vannerson. Co-founder from Nabisco; wheat PhD; bearded and subdued (a medieval friar or friendly therapist, colleagues said); built the Technical Computer System (TCS).
- Paul Cootner. MIT efficient-market theorist; traded pork bellies; designed the first risk system; left after 1971.
- Kenneth Meinken. Former Rutgers econometrician; soybeans and feed grains.
- Hans Kilian. German who toured West Africa by Land Rover counting cocoa pods.
- Michael Marcus. Hired 1974 without a doctorate; the firm's star trend surfer.
- Bruce Kovner. Hired 1977; later founded Caxton Corporation.
- Irwin Rosenblum. CFO, who judged that Kovner made more money at the firm than anyone.
Weymar's early wins
Weymar, tall and bespectacled with a Northern European air from his German parents, had moved cities often as a child and was determined to become rich enough to owe nobody. As a graduate student he modelled the frozen-orange-juice industry; when the model said concentrate prices would double, he borrowed $20,000 and bought a warehouse consignment. Prices duly soared, but supermarkets would not buy wholesale juice from a student, so he sold it back to the warehouse, which kept a fifth of his profit.
At Nabisco he persuaded his bosses to trade on his cocoa model: buy when prices fell below the forecast, pause when they rose above it. Prices immediately fell far below the model and kept falling; following his rules, he kept buying, fending off the anxious CFO with quantitative jargon, until he held two years of Nabisco's cocoa needs. Just as his nerve was breaking, West African crops failed and prices nearly doubled; he sold part of the stockpile at a vast profit, in a period he admits was short on modesty.
The Princeton farmhouse
Weymar and Vannerson raised $2.5 million and opened in March 1970 in a farmhouse near Princeton, New Jersey, among flowering trees and lawns. Weymar wore a suit on day one; Vannerson came in polo shirt and khakis with his dog Peanuts. Weymar had a huge walnut desk and red leather chair, but Vannerson's egalitarian style won out: Agnes, the cook, attended company meetings. A three-legged German shepherd named Cocoa guarded the place; seven professionals and six support staff pitched horseshoes at lunch and played softball after the close, and seminars filled blackboards with formulas. This was about models, maths and superior information, far from Jones's stock jockeys or Steinhardt's pizza-strewn trading room.
The team was built for fundamental analysis: econometricians predicting where prices ought to be, on the theory that profits come when reality catches up. Weymar took cocoa, Vannerson wheat, Cootner pork bellies, Meinken soybeans and feed grains; specialists in separate commodities would diversify the firm. Weymar fed his cocoa model with correlations between rain and humidity and yields in the Ivory Coast and Ghana, forecasting the harvest before official reports. Kilian checked those forecasts in the field, counting the number, length and condition of pods (about 25 per tree, on trees as tall as apple trees), suffering a stiff neck, a failing Land Rover and endless struggles to obtain enough Ghanaian currency to pay local helpers.
The corn blight disaster
After a fungal corn blight hit US fields, some experts warned it would return bigger the next year, and corn futures rose. Seeking an edge, the firm funded a Rutgers plant pathologist who advised New Jersey, paying for his research and conference travel. He concluded the scare reflected only media alarmism, so the traders built vast short positions. Then a Friday-night CBS News special, alongside its Vietnam coverage, featured the Illinois state pathologist, from a far bigger corn state, forecasting a catastrophic harvest. On Monday corn futures hit the exchange's daily price limit after a handful of trades, trapping the firm; it escaped only on Tuesday, at the very top. Capital had fallen from $2.5 million to $900,000. The Rutgers man was eventually proved right: there was no blight.
The firm came within a whisker of closing. Founders fell out (Weymar wondered about the shotgun an angry co-founder kept in his office for rabbits and pheasants), and several board members wanted their money back: the PhDs seemed to lack Samuelson's high P.Q. Weymar refused to quit. As a student who had once lost everything, he had drunk beer in the Red Sox bleachers until he thought his way back; now he cast himself as a romantic novel's hero, wallowing in angst. If a three-legged dog could soldier on, so could a cocoa trader.
In July 1971 the board gave him one last chance: the firm would close if it lost another $100,000. Four of the seven founding professionals, including Cootner, left. Recovery came soon, and it laid the base for one of the era's great trading operations.
Rebuilding: risk control first
Weymar rethought his theory. An economist's faith in data had bred hubris: positions too big to survive being wrong. A smaller short would have ridden out the CBS scare and ended in profit. His new maxim: the most dangerous people are very smart traders who have never had their teeth kicked in. He may have known this included himself; a colleague likened putting Weymar in charge of risk to putting Evel Knievel in charge of road safety.
Cootner's first-year system had been mathematically elegant but unenforceable. All capital sat in one pot; traders could draw freely but paid penal interest for adding to large, volatile positions, so in theory only the very confident would double up. In a firm of strong egos, confidence proved abundant, and clerks were so far behind in tracking exposures that there was effectively no risk control.
- Traders as profit centres. Each trader gets a capital allocation sized by past performance, like Jones's segment managers.
- Position caps. A ceiling on the risk in any single position, needed because futures can be bought on a small cash margin, whereas stock buyers could borrow at most half the value (2-to-1 leverage).
- Drawdown stop. Lose half your initial capital and you must sell everything, take a month off, and write management a memo explaining your mistakes.
- No fighting the trend. At most one tenth of capital may bet against a trend, with trends defined by the TCS computer.
This system, Mallaby notes, survives more or less intact in the modern hedge funds descended from Commodities Corporation.
Trend following wins the argument
The second rethink was a new respect for price trends, which efficient-market theory denied; through the 1970s and much of the 1980s it was hard even to publish contrary views. But Vannerson had spent a year at Nabisco on a trove of daily prices for 15 commodities compiled by Dunn & Hargitt of Indiana, and by opening day in March 1970 had convinced himself that trends exist. His Technical Computer System traded on the finding, one of the first of the industry's many automated systems.
Weymar was sceptical. The idea, buy what has just risen, short what has just fallen, seemed too trivial to make serious money, even though the TCS tried to tell lasting trends from noise. By summer 1971 he had converted, for two reasons:
- Built-in discipline. A computer system must have risk controls programmed in from the start, so no overconfident trader could breach his limits.
- Better calls. Weymar's cocoa model, brilliant at Nabisco, had misjudged the market expensively in year one; the TCS, watching market patterns rather than chocolate demand or rainfall, had made money consistently from the first day.
Weymar shifted more capital to the TCS, and the one-tenth rule forced human traders to respect it. Even Samuelson put in fresh money for the TCS to invest, though trend following had little standing in academia and none in his own research.
Michael Marcus, surfer
In 1974 the firm hired Michael Marcus, the opposite of Weymar's model trader: no economics degree, no use for computers, little maths, a psychology PhD dropout, among the first hires without a doctorate. He silenced the sceptics with triple-digit returns; over ten years he grew his account 2,500%, and in some years out-earned all the other traders combined.
Marcus was intense and fiercely controlled. Convinced most foods contained poisons, he once purified himself on raw fruit and vegetables until so emaciated he hired a dietician-cum-driver. He threw lavish parties, owned about ten homes (some sold without a night spent in them), chartered a jet, converted a bus into a travelling home for an entourage, and took kick-boxing lessons from a world champion.
Fundamentals plus charts
He arrived each morning with an oversized briefcase of market reports, handwritten notes taped to key pages (pre-Post-it), reading newsletters and supply-demand data and testing scenarios: if corn rose, would wheat follow; if the weather turned cold, which crops suffered first? But he also read price charts as a window into investor psychology, focusing on how the two interacted. If fundamentals turned bad yet the chart kept rising, investors had already digested the setback, and there was nowhere to go but up.
His respect for trends came from the cotton exchange floor, where traders worked one another into shouting crescendos and then fell back exhausted. When a price broke above the previous day's high, it would often keep riding a wave of excitement, so Marcus took a large position at the crossover, protected by a stop-loss. Either the market ran or he was out: like mounting a surfboard, you either catch the wave or drop back into the water.
- Win rate. Marcus caught the wave on under half his attempts, but winners paid 20-30 times the small stop-loss losses.
- Rounded bottoms. A bumper harvest drives prices down and leaves excess in storage, an overhang that keeps prices low for a long time.
- Spiky tops. A sudden shortage forces consumption down and prices up, but a good next harvest ends it, so prices spike and collapse.
- Spike bottoms. Pork bellies and eggs cannot be stored long, so gluts are dumped rather than held.
- Gradual absorption. People digest news at their own pace, so instant processing is an ivory-tower fiction; slow digestion is why trends exist.
- Reaction to recent experience. Losses trigger anxious selling and gains euphoric buying; at congestion points, a breakout inflicts sharp losses on the wrong side, whose panic closing speeds the escape, like travellers jammed at a station door who accelerate once through.
- Generalism. Trade whichever market is producing waves rather than studying a quiet sugar or wheat market; Marcus, a former cotton specialist, preached generalization with a convert's zeal.
Marcus's success made Weymar's founding concept inoperative. With so many variables driving prices, econometric models could rarely get all of them right, and even a correct forecast could bankrupt you while waiting for the market to agree. The quants would have their day later in the story; for now, trend surfing paid.
Into currencies: the first stirrings of macro
Refusing to specialize made Marcus a pioneer in currencies. After Nixon cut the dollar's gold link, the Chicago Mercantile Exchange began trading futures in seven floating currencies in May 1972, and investor psychology produced the same waves there. Currencies were about a third of Marcus's trading by the mid-to-late 1970s and two thirds by the decade's end; colleagues and the traders the firm seeded followed. The "macro" traders who would torment central bankers from the 1980s were stirring in a Princeton farmhouse.
Betting on bad policy
Marcus was a libertarian, and the stagflationary 1970s kept vindicating the view that big government blunders. Nixon's early-1970s price controls were the equivalent of later doomed currency pegs: unsustainable policy and a bonanza for speculators.
- $110. Nixon's cap on plywood per thousand square feet amid a construction boom: traffic cones in front of a tank. Warehouses shipped wood to Canada and back or shaved slivers off planks to charge $150; builders turned to unregulated futures, which Marcus bought by the truckload and which nearly doubled.
- 1974 inflation trade. After controls ended in early 1973, leveraged commodity longs paid: coffee +25%, rice +two thirds, white sugar doubled.
- The reversal. Then a tradable downtrend: sugar -67% from its peak, cotton and rubber -40%, cocoa over -25%.
- The Carter dollar. Stimulus-fuelled inflation sank the dollar by a third against the yen and deutsche mark.
- The riyal, early 1975. A leveraged long bet on the Saudi riyal's dollar peg, rewarded with a revaluation in March 1975.
The riyal trade prefigured the great peg attacks from Britain to Thailand. Booming oil revenues flooded Saudi Arabia with money and pushed the currency up against its dollar peg. Revaluation was not certain (the kingdom could tolerate inflationary inflows), but devaluation was inconceivable. That made a long riyal position a near one-way bet: no guarantee of winning, a near guarantee of not losing. Marcus levered up and slept soundly, and made another killing.
Why it worked
By the late 1970s many traders were earning 50%-plus. Capital grew from under $1 million at the low to about $30 million; the farmhouse got a new wing and a new building. Traders hired chart researchers, and administrators multiplied at a rate old-timers thought absurd, since overhead came out of trading profits. In 1980 profits hit $42 million; after $13 million of bonuses to 140 employees, the quiet firm out-earned 58 of the Fortune 500. Weymar flew staff and families first class to a Bermuda retreat, where the traders attacked his spending.
Mallaby draws three lessons:
- Flexibility. Innovation is often credited to grand theory (Stanford engineering behind Silicon Valley, the National Institutes of Health behind pharma), but in practice it usually comes from humble trial and error, going with what works whatever the theory, even in finance. Jones expected charts to time the market and instead won through incentives; Steinhardt's troika began as equity analysts and won through block trading and monetary analysis; Weymar began with econometrics and won through trends.
- Selling insurance. Markets transfer risk from insurance seekers to insurance sellers. In the 1960s government price floors and surpluses kept commodity prices calm, and Chicago Board of Trade traders read newspapers on the soybean-pit steps. The inflationary 1970s brought a rush to hedge: food companies against high prices, growers against low ones, multinationals against dollar swings, and CBOT turnover surged. Hedgers trade to shed risk, not on a price view, so informed speculators had the upper hand; exchange position limits restricted the supply of insurance and inflated its price, and the firm also traded off-exchange with wholesalers to get round the limits.
- Trend following, above all. Vannerson's TCS showed how wrong the random walkers were and gave the firm confidence to hire Marcus-style traders and make fundamentals plus charts its creed.
Academia caught up years later. A 1986 Journal of Finance paper found profitable trend following in currencies; a 1988 study found it in commodity and currency futures. One of its authors, Scott Irwin, had been prompted by a meeting with Dennis Dunn of Dunn & Hargitt, and examined the same price series Vannerson had mined almost two decades earlier, reaching the same conclusion.
Bruce Kovner
In 1977 Marcus advertised for an assistant. The applicant had dropped out of a Harvard PhD and was driving a cab part-time; Marcus phoned Weymar at once to say he had the next president of the firm in his office. Bruce Kovner was tall and imposing, with striking range: part of a Harvard political-science circle including James Q. Wilson and Daniel Patrick Moynihan, a full-time music student for a spell, a political-campaign worker, a freelance writer for Commentary. Prompted by a friend, he had studied futures, borrowed $3,000 on a MasterCard and turned it into $22,000. Quizzed on financial texts, he had read more than his interviewers, from Charles Mackay's Extraordinary Popular Delusions and the Madness of Crowds to current newsletters. Weymar thought him intellectually rich beyond measure and hired him as a trader, not an assistant.
Over the next decade Kovner averaged about 80% a year, founded Caxton Corporation, and became a figure in arts and politics: chairman of the American Enterprise Institute and backer of the New York Sun, chairman of Juilliard, sponsor of the illustrated Pennyroyal Caxton Bible (with an engraving of Kovner as King Solomon), and, after a divorce, owner of Manhattan's former International Center of Photography, converted into a mansion with a rare-book vault and a study that doubles as a radiation-proof bomb shelter.
Temperament and organization
Marcus, the ex-psychology student, saw that Kovner had unusual physical and psychological strength: he shed distractions, did not overthink and slept well. Others made money faster but lost it faster; Kovner was consistent and nerveless. After a silver loss that would have left most traders vomiting, he walked into an administrative meeting the same day as though nothing had happened.
He also had organizational talent. He hired assistants to track numbers and draw charts, paying no more than necessary, and got the best from ordinary people: an ex-librarian monitored the link between interest rates and gold futures with orders to bet on normality's return whenever it blipped, a handsomely profitable formula. He extracted concessions from Weymar, getting the central budget to pay his assistants and extra capital allocations for their ideas, to colleagues' irritation.
Kovner's trading ideas
Kovner blended fundamentals and charts, confirming the move away from Weymar's models, and sometimes put the charts first. He argued that the best opportunities come when you have no fundamental information: if a quiet market suddenly breaks out for no visible reason, some insider knows something, and following him gets you in before the news is public. Once, when he and Marcus were short the dollar and it inexplicably strengthened, they bailed out at once; that weekend President Carter announced a dollar-support programme. Trading on announced fundamentals, as Weymar first intended, would have annihilated them.
- The setup. Find currencies that are much cheaper in the forward market than today, and buy them at the bargain forward rate; sell forwards trading at small discounts.
- The conventional view. Traders assumed a low forward rate simply forecast depreciation, so there was nothing to gain.
- The insight. A forward discount mostly reflects high interest rates: if Spanish banks paid 7%, the peseta would be about 7% dearer spot than one year forward, compensating forward buyers for forgoing the interest.
- Why it pays. High rates tend to cut inflation and attract capital, so a discounted currency is more likely to rise than fall.
- The run. The trade earned glorious profits for about a decade, until rivals caught on; it became a hedge-fund staple.
The break-up
As word of the profits spread, New York brokers approached the star traders offering to place client money directly with them, in effect asking Marcus and Kovner to run hedge funds inside Weymar's company. Weymar resisted, but the traders held the cards: his vast overhead already tempted them to go independent, and Wall Street now offered outside capital plus a fat fee. They won the argument. Kovner was soon managing millions of Wall Street money with an army of assistants, a state within a state; by the time he left to found Caxton in 1983, he was already an independent mogul in all but name: secretive, leveraged and enviably successful.
Behind him came two younger futures traders, Paul Tudor Jones and Louis Bacon. Both took seed capital from Commodities Corporation, flew in by helicopter from Manhattan for traders' dinners in the early 1980s, swapped ideas on trends and chart patterns, and adopted its risk-control procedures; Bacon later hired the firm's senior administrator Elaine Crocker as president of Moore Capital. But Bacon was too independent to fit, and Jones turned down a job: he wanted the money, not the employer. The difficulty of keeping stars, plus Weymar's spending, produced a 1984 internal revolt that forced him to promise lower overhead and a return to "the simple life." The firm carried on but never regained its verve. The centre of gravity moved from Weymar to a younger set, and from Princeton to New York.
The 1971 rulebook is still the skeleton of trading-floor risk control at banks and multimanager "pod" funds: capital allocated by track record, per-position limits, a hard drawdown stop that forces de-risking and review, and limits set by an independent model rather than the trader's conviction. The corn episode also shows liquidity risk: limit-down markets make stop-losses and VaR estimates meaningless when you cannot trade.
Mallaby's carry-trade explanation is simplified: a forward discount reflects the interest differential between two currencies, not a single country's rate. The profitability Kovner exploited is the documented forward-premium puzzle (Fama, 1984), but the trade is prone to sudden crashes when funding currencies jump, as in 1998 and 2008. Samuelson's investment is described from secondary sources; how much of his personal portfolio it represented is not stated.
Kovner handed Caxton's investment management to Andrew Law in 2008 and retired in 2011, when Law became chairman and CEO. Commodities Corporation itself had been bought by Goldman Sachs in 1997.
The Alchemist
George Soros turned Karl Popper's philosophy of fallible knowledge into a theory of markets, reflexivity, in which flawed perceptions move prices and prices then change the fundamentals they are supposed to reflect. It produced what Soros called the killing of a lifetime around the 1985 Plaza accord, then a humiliating loss in the 1987 crash, when Quantum's sheer size turned its own exit into a market-moving fire sale. Mallaby reads 1987 as Soros's investment humiliation but intellectual vindication: the crash broke the efficient-market consensus and opened the door to the hedge-fund boom, while sowing the suspicion that hedge funds destabilize markets.
Budapest, London and Popper
The London School of Economics in 1949 was full of refugees from Nazism, exiles from communism and young leaders from Britain's fading empire, all hunting for grand theories of how Europe had wrecked itself and how it could be rebuilt. Labour was building the welfare state and Marshall aid was rebuilding the continent; in the lecture halls Marxists sat beside the libertarian Friedrich Hayek, and Keynesians argued with anti-Keynesians. A historian of the school dated the birth of the LSE myth to exactly this period.
Soros arrived already marked by trauma. Born to a comfortable Jewish family in Budapest, he survived the Nazi occupation by separating from his family, taking a Christian identity and hiding with acquaintances of his father; he saw mutilated corpses in the streets and helped his family survive by selling jewellery to black-market dealers. In 1947, barely 17, he left for London, saying goodbye to parents he did not expect to see again. He worked as a dishwasher, house painter and busboy (a headwaiter promised that hard work might one day make him his assistant), and the summer before the LSE found a job he liked: lifeguard at a nearly empty pool, where he read Adam Smith, Hobbes and Machiavelli.
Popper's stamp
The LSE figure who shaped him most was Karl Popper, another refugee from Nazism (from Austria). Popper's central claim was that humans cannot know the truth, only grope towards it by trial and error, which made every dogma suspect: the Nazism and communism imposed on Hungary each claimed a certainty neither had earned. The Open Society and Its Enemies left Soros with a lifelong wish to contribute to philosophy, pointed him to a distinctive way of thinking about finance, and gave its name to his later philanthropy, the Open Society Institute. Mallaby stresses that Popper stamped himself on the most famous hedge-fund manager of all in an entirely unintended way.
Into finance
Soros left the LSE with mediocre grades and drifted through dead-end jobs, at one point selling handbags in north Wales. He wrote to every investment bank in the City of London; the establishment spurned him for his lack of social connections, and he finally got in at a brokerage run by Hungarian émigrés. In 1956 he moved to New York, planning to endure Wall Street for five years, long enough to save for a life as an independent philosopher. He proved too good at investing to quit.
1944 Survives the Nazi occupation of Budapest under a Christian identity, apart from his family.
1947 Leaves Hungary for London at 17.
1949 Enters the LSE; inspired by Karl Popper.
1956 Moves from a Hungarian-run London brokerage to New York.
1967 Head of research at Arnhold and S. Bleichroeder, specialists in European stocks.
1969 Launches the $4 million long/short Double Eagle Fund under Bleichroeder's umbrella.
1973 Sets up his own firm with Jim Rogers; fund renamed the Soros Fund.
1978 Renamed the Quantum Fund.
1980 Quantum gains over 100%; Soros splits with Rogers.
1981 Assets $381 million at the start of the year; Quantum then loses 23%, its first loss.
1984 Returns to full-time investing.
By 1967 he headed research at Arnhold and S. Bleichroeder, an old brokerage specializing in European stocks. Having got to know A. W. Jones's segment managers by pitching them ideas, in 1969 he launched his own $4 million long/short stock-picking fund, the Double Eagle Fund, run inside Bleichroeder.
Reflexivity
By then Soros had fused Popper with finance into a synthesis he called reflexivity. A listed company's details are too complex for any mind, so investors rely on guesses and shortcuts. But those shortcuts also change reality: bullish guesses lift the share price, letting the company raise capital cheaply and improve its performance. Certainty is therefore doubly elusive: people cannot see reality clearly, and reality is itself altered by their constantly shifting misperceptions.
- Imperfect cognition. Reality is too complex to know, so investors act on biased approximations; Popper's limits on knowledge apply to markets.
- Feedback. Those approximations alter the fundamentals: a rising price cheapens capital, which boosts earnings, which seems to confirm the bias. Prices can therefore run far from any "objective" equilibrium, until the gap becomes unsustainable and the trend reverses.
This put Soros at odds with academic finance, which starts from the premise that rational investors can reach an objective valuation and that once all information is priced in the market sits at an efficient equilibrium. To a Popperian, that premise ignored the most basic limits of cognition.
Not an after-the-fact story
Through the 1960s Soros hid away in the study of his weekend house trying to write his philosophy down, and Mallaby rejects later suggestions that reflexivity was a rationalization laid over success. A 1970 investment note on real-estate investment trusts (REITs) reasoned in explicitly reflexive terms. Conventional analysis, Soros wrote, forecasts earnings; but a REIT's future earnings depended on how investors perceived it. Bullish investors would pay a premium for shares in a successful trust, giving it cheap capital; that capital would boost earnings, reinforce the appearance of success and draw buyers at an ever greater premium. The trick was to watch neither earnings nor psychology alone but the loop between them, which would drive both forward until the trusts were so overvalued that a crash was inevitable. The trusts followed his boom-bust script, and his fund made one fortune on the way up and another on the way down.
Soros and Rogers
In 1973 Soros left Bleichroeder and rented an office a block from his Central Park West co-op, taking his partner Jim Rogers, an irascible workaholic analyst (interviewed decades later, Rogers talked while gasping on an exercise bike rigged with a laptop and phone). Together they hunted for moments when an unstable equilibrium might flip:
- Bank stocks. Financial deregulation was turning a dull sector into an exciting one, and Soros made a fortune from it.
- Defence stocks. In the 1973 Arab-Israeli war, Soviet weapons used by Egypt performed well, revealing a bigger challenge to the US than assumed; Soros predicted the Pentagon would win catch-up spending from Congress, plunged into defence shares and became the largest outside shareholder in Lockheed.
- British gilts. On a Swiss ski holiday he bought the Financial Times at the foot of a chairlift, read on the way up about the government's plan to bail out Rolls-Royce, and phoned his broker from the summit to buy British government bonds.
He bet big without waiting for proof. If an idea looked attractive on a quick look, others would be seduced too; and since perfect cognition was impossible, there was no point sweating details. His motto was "Invest first, investigate later."
The perfectly honed machine
By the start of 1981 the fund (renamed the Soros Fund in 1973 and Quantum in 1978) held $381 million, almost a hundred times its starting capital despite tough 1970s equity markets. The teenager who had sometimes leaned on charities in London had a $100 million personal fortune and was becoming a philanthropist. In June 1981 Institutional Investor hailed him as "the world's greatest money manager," and rivals borrowed Ilie Nastase's tribute to Bjorn Borg: they were playing tennis and he was playing something else.
Soros was not above admiring himself, writing without apparent irony that he had viewed himself with awe, and confessing he had fancied himself a Keynes or even an Einstein (reflexivity, he noted, sounds like relativity). But he was unhappy. Investing demanded a focus so intense it became physical: trouble in the portfolio announced itself when his back seized up, and he would obey the signal and sell. He likened himself to a boxer in training sacrificing all personal life, and to a sick man with a parasitic fund swelling inside his body.
The 1981 stumble
In 1980 he parted with Rogers, blaming him for driving out younger staff and frustrating Soros's hope of spreading the workload, and began searching for partners to delegate to. The distraction wrecked performance: after more than 100% in 1980, Quantum fell 23% in 1981, its first-ever loss, and redemptions halved capital from $400 million to $200 million. By September 1981 a humiliated Soros had handed the remaining money to outside managers and, like Steinhardt three years earlier, stepped back from the markets.
The real-time experiment, 1985
Soros returned to full-time investing in 1984 with a new sense of balance. He had feared that ceasing to be paranoid would hurt his performance; a psychoanalyst helped him accept his success and relax, knowing it might kill the golden goose but that not relaxing would make the success pointless. His physical identification with the fund ended. He compared it to an operation to remove a hard calcium stone from his salivary gland: once exposed to air it crumbled to powder, and so, he said, did his hang-ups.
In place of his aching back he adopted a cerebral discipline. From 16 August 1985 he kept a diary of his investment thinking, the "real-time experiment." It is dense, repetitive and full of scenarios that never happen, but Mallaby values it because it is free of hindsight bias: a true portrait of a speculator at work, and it happened to record the dollar bet Soros called "the killing of a lifetime."
Why currencies
Soros came from Wall Street stock picking, but reflexive feedback loops pushed him to think broadly. Like Michael Marcus, who left the cotton floor to become a generalist, he saw no point knowing everything about a few stocks to catch small moves; the game was to know a little about a lot and spot where the big wave might come. After Bretton Woods, floating currencies were a natural playground: the dollar's value rested on traders' perceptions, which he assumed were flawed and could reverse at any moment.
Equilibrium view (1970s-80s mainstream). An overvalued dollar hurts exports and boosts imports; the resulting trade deficit means foreigners need fewer dollars than Americans need foreign currency, so weak demand pushes the dollar down until trade balances. Speculators who guess right merely speed the adjustment; those who guess wrong slow it briefly but lose money and give up.
Soros's view. Speculative capital flows dominate and are self-reinforcing: from 1982 to 1985 the US ran a growing trade deficit yet the dollar strengthened, because hot money pushed it up and the rise drew in more. Currencies follow a perpetual boom-bust sequence: a prevailing bias reinforces itself, the trade imbalance grows, and eventually the imbalance overwhelms the bias, speculators swivel 180 degrees, and a new trend runs the other way.
Timing the turn
The question in summer 1985 was timing. Reagan had reshuffled his team for the second term, and the new officials seemed determined to talk the dollar down to shrink the trade deficit. Interest rates were falling, making dollars less rewarding to hold. In a mature cycle every speculator who wants to ride the dollar is already aboard, so with few buyers left, a handful of sellers can force a U-turn.
He agonized. Faster US growth would lift rates and make a reversal less likely; but a self-reinforcing credit contraction in the banks (falling collateral values feeding reduced lending) could slow the economy sharply and push rates down. Wondering who he was to judge, he concluded: "The only competitive edge I have is the theory of reflexivity." The theory made him weight a reflexive banking crisis heavily, so he leaned against the dollar, and technical indicators agreed. Having consulted what Mallaby calls the quasi science of economics, the quasi philosophy of reflexivity and the quasi psychology of the charts, he shorted the dollar.
- $720 million. Long yen, marks and sterling on 16 August, exceeding the fund's entire equity by $73 million. His rule of not exceeding 100% of equity in any one market came with a breezy caveat: he adjusted the definition of a market to suit his current thinking.
- -$20 million. Loss reported in the second entry, 9 September, after bullish US data lifted the dollar.
- $30 million. Overnight profit when the Plaza accord was announced on 22 September 1985; the yen rose more than 7% the next day, its biggest one-day jump ever.
- +$209m / +$107m. Extra yen and marks bought, and new dollar shorts added, by the Friday after Plaza.
- +$500m / +~$300m. Further yen and marks, and further dollar shorts, by early December.
- +35%, $230 million. Quantum's gain in the four months from August.
Mallaby's verdict on the sizing is dry: the idea that a hedge fund should actually be hedged had been casually discarded.
Holding on in September
The 9 September entry found Soros soul-searching. He kept focusing on the weak banks, and the charts hinted that the mark was forming a pattern that often preceded a sharp rise. Then he added a new dimension by putting himself in the authorities' shoes: rates would likely stay low even if growth surprised, because the Fed, as bank regulator, would not want to make capital dearer for wobbly lenders, and could afford not to because the reshuffled administration was determined to cut the budget deficit, easing inflation. He decided to keep the losing bet but cut it in half if the market moved further against him.
Mallaby's model of speculation is set out here: markets are at least somewhat efficient, so most information is already priced, and the art is to find one overlooked insight and trade big on that small edge. Soros would build a thesis and then turn on a dime, and a stray remark from a lunch guest could send him leaping up to order a trader out. Holding in the second week of September was one of those knife-edge calls; had he blinked after the first loss, his life story would have been different.
Plaza and piling on
On Sunday 22 September 1985, Treasury secretary James Baker met his counterparts from France, West Germany, Japan and Britain at New York's Plaza Hotel, and the five powers promised coordinated intervention to push the dollar down. Mallaby separates luck from skill. Soros had foreseen that Reagan's team wanted a weaker dollar but had no foreknowledge of Plaza or how the intention would play out: that was luck. What followed was skill, and the making of a legend.
Instead of banking his profit, he piled on. It was already Monday morning in Asia, so he called Hong Kong brokers to buy more yen that night. When his own traders began taking profits in their small sub-portfolios the next day, he stormed out of his office in a rare fury, ordered them to stop selling yen and took over their positions. In his reading, the world's top governments had signalled that the dollar was headed down, so he should hog as much yen as he could. The risk was that the communiqué, thin on actionable detail, would prove a paper tiger; but Soros, with a web of political contacts in Washington, Tokyo and Europe wider than any other New York manager's, was told that Plaza was serious. He wrote in the diary that he had taken maximum exposure in all directions.
The verdict on phase one
Closing the first phase in December 1985, Soros admitted his repeated forecasts of a banking collapse had been a red herring: he kept expecting developments that never materialized. But one central insight dwarfed the errors. Nothing more substantial than slippery perceptions had pushed up the dollar, so a trigger could set off a sudden reversal; because he grasped the system's instability, he understood Plaza's meaning faster than others, and the policy's missing details did not matter. A political jolt had started a trend that would feed on itself.
He joked that the $230 million profit, which he partly credited to the discipline of the diary, was the highest honorarium ever paid to an author. When the diary came out two years later inside The Alchemy of Finance, reviewers mocked its prose, though one remarked that financial alchemy beat boiling mercury with egg yolks.
1987: celebrity
The Alchemy of Finance appeared in May 1987 and made Soros a celebrity. Younger stars saw in the diary an honest picture of a speculator wrestling with imponderables. Paul Tudor Jones, the cotton prodigy seeded by Commodities Corporation who built Tudor Investment Corporation, made it required reading for his staff, and in a foreword warned Soros that having published he should beware, invoking the scene in the film Patton where the general, having studied Rommel's tactical writings, gloats that he read his book.
Soros was enjoying himself too much to worry. He had become the public intellectual he had admired at the LSE: expensively tailored, large glasses, a thick tangle of hair and a Central European accent, on magazine covers. Profiles noted that he had invested globally before most managers had found Tokyo on the map, used futures, options and forward currency contracts, went long and short with equal ease, and ruled the world's markets from an eerily quiet Manhattan floor in five languages. The Economist called him the world's most intriguing investor, and Fortune suggested he might outrank Warren Buffett as the most prescient investor of his generation. Mallaby notes the omen: 1981's flattering profile had preceded a 23% loss, and 1987's adulation preceded a calamity.
Bullish at the top
Fortune's cover of 28 September 1987 asked "Are stocks too high?" After a bull market dating from the start of the decade, the S&P's 400 industrials traded at three times book value, the highest since World War II. Soros, the magazine's first expert witness, was sanguine: overvaluation did not mean the market was unsustainable, and Japan, even more stretched, would crash first, before Wall Street. He had company: a week later Salomon Brothers forecast the bull market would last into 1988, and the week after Byron Wien of Morgan Stanley, a Soros friend, predicted a new high before the cycle ended.
It was the leveraged-buyout era, and debt-funded takeovers kept pushing prices up. The mood was captured by an unknown financier, P. David Herrlinger, who announced a $6.8 billion bid for Dayton Hudson, told reporters on his lawn it was no more of a hoax than anything else, sent the stock soaring, and was then revealed as a hoaxer and taken to hospital. The buyout mania fitted reflexivity neatly: each deal lifted every stock in the sector and the prospect of re-floating acquired firms at a profit, loans kept flowing, and prices spiralled away from fundamental value. A collapse under the debt would come eventually, but there was no sign it would come soon, so Soros played Wall Street's philosopher-in-chief on television and with journalists.
Enter Druckenmiller
On 5 October Soros received Stanley Druckenmiller, the hottest mutual-fund manager on the Street, who had read Alchemy and asked to meet. Soros offered him a job: he wanted a successor to run Quantum so he could spend more time on philosophy and philanthropy. Druckenmiller declined, but they grew close: tall, broad, plain-spoken and middle American where Soros was compact, complex and exotically middle European; in his mid-thirties to Soros's late fifties, and willing to listen while Soros pontificated. Their books differed. Druckenmiller expected a break and was short Wall Street; Quantum was short Japan but long the US, having added stock pickers who loaded up on hot takeover names (the "garbitrage" stocks), and was up about 60% by the end of September.
The week Quantum lost $840 million
- Wed 14 Oct. Soros's FT article repeats that the crash will start in Tokyo. He lectures on boom-bust theory at Harvard's Kennedy School and emerges to find the Dow down 3.8% on reports Congress might raise taxes on mergers, threatening an engine of the bull market. He later admitted that was when he should have been in the office getting out.
- Thu-Fri 15-16 Oct. Stocks slide, then dive. Convinced by his charts that the fall has gone far enough, Druckenmiller flips from short to long on Friday afternoon.
- Fri evening. At Soros's office he sees Tudor Jones's charts: markets tend to accelerate downward once an upward parabola breaks, with a parallel to 1929. Panic rises in his gut.
- Sat 17 Oct. At Jack Dreyfus's, he studies charts the patriarch's secretary keeps on individual stocks: stock after stock has just broken out of its congestion zone downward. Chartists say soldiers lead generals; by watching the index he had missed the troops.
- Mon 19 Oct. Druckenmiller sells and by late morning is short again. Black Monday: the Dow falls 22.6%, its biggest drop in its 91 years. Soros, running more money and holding illiquid garbitrage stocks, cannot get out. (One trader steadied herself by watching a hot-dog stand: while hot dogs sold, the world was not ending.)
- Mon evening. His Nikkei short cushions losses. Blaming portfolio insurance, he reasons a program-driven crash should correct soon.
- Tue 20 Oct. The market rallies and Soros buys back in. But his Nikkei short was placed in Hong Kong futures for liquidity, and Hong Kong has shut its futures exchange: he cannot cover.
- Wed 21 Oct. The Nikkei jumps 9.3%, its biggest gain since 1949. Near the New York close Druckenmiller, who has studied crash history (a sharp fall, a wild two-day rally, then a slump back to the low), tells Soros he is short. Soros, having consulted others, thinks Black Monday was a freak.
- Thu 22 Oct. London is collapsing early. Around 8 a.m. Salomon's futures desk warns Druckenmiller that an elephant may push futures, closed at 258, below 200. He places orders to cover at 195; they fill, a 25% return overnight. At about 10 a.m. he calls Soros to say some crazy person has just sold recklessly; Soros, calm, says he is licking his wounds. That weekend Barron's reveals the elephant was Soros.
The Thursday story was tied to Japan. Tokyo rose again on Thursday, Soros could not cover his short until Hong Kong reopened, and on top of Monday's New York losses Quantum faced the loss of confidence that destroys leveraged funds: lenders call loans, forced sales hit a weak market, and a death spiral follows. Seeing London fall and not wanting to be too slow twice in a week, he decided to move to cash. His trader Joe Orofino placed the order to sell Quantum's entire $1 billion of S&P futures on the Chicago Mercantile Exchange through Shearson Lehman Hutton. An order that size could not avoid moving the market: pit traders sold frantically as the fire sale began, and short sellers like Druckenmiller let prices crash before taking profits.
Going to cash that day was perhaps the worst call of his career, costing about $200 million. It capped a cataclysmic week in which Quantum went from up 60% for the year to about 10% down: $840 million vanished.
The episode exposed a weakness that would haunt the industry: size kills agility. A smaller Quantum could have bailed out on Monday as nimbly as Druckenmiller and sold on Thursday without cratering the price. A style built on turning on a dime fails once your own orders move the market.
The Lazarus act
The crash broke many managers' confidence; Steinhardt recalled being so depressed that autumn that he did not want to go on, and felt alone. Soros felt only a faint echo. He was tired of running Quantum and quitting was always somewhere in his mind, but his nerve never wavered, and his fans rank the recovery among his greatest feats. A week or two after Black Monday he put on a gutsy, leveraged short of the dollar as if nothing had happened; the dollar fell, and Quantum finished 1987 up 13% after being in the red two months earlier.
The sniggers had been loud. An anonymous source told the London Times: "It took 20 years to make George Soros a genius; four days to make him a jerk." Forbes mocked his Fortune cover, calling him rich, immensely conceited and famous, and advising readers to sell stocks whenever he appeared beside a gushing headline. Fear that the losses would end his giving spread to eastern Europe, and Soros flew to Hungary to reassure the prime minister. By year end Financial World ranked him the second-highest earner on Wall Street, behind Paul Tudor Jones, chartist and Patton fan.
The birth of modern macro
The Plaza killing plus the 1987 comeback made Soros a folk hero, and Mallaby credits his example with creating the modern "macro" hedge fund. The style had a precursor: from 1924 until his death in 1946, John Maynard Keynes ran King's College, Cambridge's endowment, the Chest Fund, in global markets, speculating in currencies, bonds and equities, long and short, with leverage. After the war, stable inflation, regulated interest rates and fixed currencies (fixed partly by Keynes himself at Bretton Woods) killed the tradition; it stirred again, tentatively, after Bretton Woods collapsed in the 1970s.
Two streams revived it. The equity tribe, investors like Steinhardt, learned that interest rates drive stocks and from the 1980s bet on bonds directly, first in the US and later abroad; their culture was fundamental analysis. The commodity tribe, traders like Marcus and Kovner, started in cotton and gold and then surfed the new currency and interest-rate futures; their culture was charts and trend following. Until Alchemy, the two traditions stayed apart. Soros's real-time experiment blended fundamentals with a belief in trends, the language of economists with the instincts of a chart watcher, so it spoke to both halves of the hedge-fund house and reminded each that the other had wisdom. Within a few years commodity man Tudor Jones and equity man Druckenmiller were both simply macro investors.
What caused the crash
For years Black Monday was blamed on financial engineering, later blurred with hedge funds in the public mind: portfolio insurance, which sold futures as prices fell, supposedly created a destabilizing loop of falling prices triggering more insurance selling. Mark Rubinstein, the Berkeley professor who co-invented it, fell into what he later recognized as clinical depression, fearing that weakened US markets might tempt the Soviet Union to attack and make him responsible for a nuclear war.
Mallaby thinks innovation was blamed too eagerly, Soros's own view included:
- History and geography. Markets had crashed periodically throughout history, and foreign markets with far less portfolio insurance also fell steeply.
- The numbers. Of the $39 billion of stock sold on 19 October through futures and cash markets, only about $6 billion came from portfolio insurers.
- Low-tech villains. Old-fashioned stop-loss orders left with brokers probably caused at least as much selling.
- Psychology. Crash fears were everywhere: the Atlantic Monthly ran "The 1929 Parallel," and the Wall Street Journal on Black Monday morning overlaid the recent decline on a 1920s chart. Tools are just tools; people bought insurance and placed stops because of the skittish mood.
The intellectual vindication
The larger lesson, Mallaby argues, was the blow to efficient-market theory. Corporate America's value bounced around like a pachinko ball for a week, with no sign of efficiency or equilibrium. Soros later wrote that reflexivity could explain such bubbles and the efficient-market hypothesis could not, and Mallaby judges him broadly right. He adds a pointed observation: efficient-market thinking grew up on American campuses in the 1950s and 1960s, the most stable enclaves of the most stable country in the most stable era in memory, whereas Soros had survived the Holocaust, war and London penury. The consensus began to come round to him.
| Attack on efficiency | What it showed | Evidence |
|---|---|---|
| Statistical | Price changes have fat tails, not a bell curve | Mandelbrot and Fama (early 1960s): 5-sigma daily moves every 3-4 years, not once in 7,000 years; the 1987 futures plunge had normal-curve odds of 1 in 10^160 |
| Institutional | Smart money cannot always correct prices | Jammed phones, threatened credit lines, scarce stock to borrow, uncertain "knowledge", finite risk capital; overvaluation can persist |
| Psychological | Investors are not Homo economicus | Repricing corporate America by a quarter in a day; Richard Thaler's anomaly features in the Journal of Economic Perspectives from 1988 |
Statistical: fat tails
Efficient-market theory rested on the assumption that price changes follow a normal distribution, the bell curve where extreme values all but vanish. In the early 1960s the maverick mathematician Benoit Mandelbrot argued the tails were fatter, and Eugene Fama, father of efficient-market theory, ran tests that confirmed it: jumps above five standard deviations, which should appear about once every 7,000 years of daily data, turned up every three to four years.
Fama and colleagues then buried the finding. It made the field's statistical toolkit useless, since modelling abnormal distributions was largely unsolved mathematics; Paul Cootner of Commodities Corporation lamented, Churchill-style, that Mandelbrot offered blood and toil rather than utopia, and that if he was right almost all past econometric work was meaningless. The profession looked away, especially as normal-curve mathematics yielded the 1973 options-pricing breakthrough by a trio of economists and launched a new industry; Fama now called the normal distribution a good working approximation.
1987 forced a rethink. Under a normal distribution, the S&P 500 futures plunge of 19 October had odds of one in 10^160: it should not occur even if markets stayed open for twenty billion years (the upper estimate of the universe's lifespan), nor across twenty successive big bangs of that length. Mandelbrot, who had quit financial economics after being brushed off in the early 1970s, returned with a vengeance; his chaos theory, in which small pieces of information generate large price moves through complex feedback loops, much like Soros's thinking, gained a cult following among money managers.
Institutional: frictions
Efficient-market theory assumed investors could always act: if IBM was worth $90 but traded at $100, informed speculators would short it until the price fell $10. That required heroic leaps: that they could borrow enough stock to short, that their "knowledge" was certain rather than risky, and that they had unlimited appetite for risk. Before 1987 the theory said a small, well-armed minority would suffice. Black Monday showed otherwise: broker phone lines were jammed, leveraged investors feared their credit lines would be pulled, and the sheer weight of selling made it crazy to buy against the trend whatever hedge funds thought of value.
So the hypothesis failed at least in crises. It also struggled to explain why the bubble had formed: in summer 1987 investors could see stocks trading at historically high earnings multiples, but bucking the market cost money, shorting was hard, and the few skilled short sellers had limited war chests. The picture of wise speculators pushing prices into line was, at a minimum, exaggerated.
Psychological: not Homo economicus
The third attack, closest to Soros's own complaint, targeted the perfectly rational Homo economicus: when investors could revalue corporate America by a quarter in a day, something other than reason was at work. Economists became open to explanations, and from 1988 Richard Thaler of the University of Chicago published regular features in the Journal of Economic Perspectives on choices that violated rational expectations. For Soros, obsessed with the limits of cognition since his London student days, it was another victory.
The double edge
The triple attack, statistical, institutional and psychological, vindicated the hedge-fund industry. It explained how Steinhardt's block trading and Weymar's commodity speculators had done so well, and showed practitioners had often been ahead of the academics. Recognition that efficiency was imperfect prompted a wave of finance professors to launch hedge funds and persuaded sophisticated endowments to fund them, triggering the industry's headlong growth after 1987.
There was a darker side. If markets were not always efficient and rational, boom-bust sequences could distort the economy and hurt ordinary workers and households; and if markets could be demons, fast-trading hedge funds might be demons on steroids. Mallaby calls that suspicion exaggerated, but it would haunt the funds throughout their golden era.
Two lessons travel straight into bank risk practice. Fat tails: normal-distribution VaR understates crash risk, which is why regulators moved to expected shortfall and stress testing (Basel's FRTB). Liquidation risk: Quantum's Thursday sale shows that position size relative to market depth must be priced into exits, the logic behind liquidity horizons and concentration add-ons in modern capital rules. The Hong Kong closure adds venue risk: a hedge on an exchange that can shut is not a hedge.
Mallaby's dismissal of portfolio insurance is contestable: the Brady Commission (1988) gave it and index arbitrage a central role in the selling cascade, concentrated in key hours, even if the dollar share of the day's total was small. The diary also flatters Soros less than it seems: it is his own record, his Plaza windfall depended on a meeting he did not foresee, and his central banking-crisis thesis never came true. The 10^160 figure is an illustration of how badly the normal model fits, not a meaningful probability.
Druckenmiller announced in August 2010, months after the book appeared, that he would close Duquesne Capital to outside investors. In July 2011 Soros announced he would return under $1 billion of outside money and convert Soros Fund Management (then about $25.5 billion) into a family office to avoid new SEC registration rules, ending a record of roughly 20% a year since 1969.
Top Cat
Julian Robertson's Tiger Management earned 31.7% a year after fees from May 1980 to its August 1998 peak, against 12.7% for the S&P 500, by doing what efficient-market theory said was impossible: analysing companies, currencies and commodities and betting on them. Mallaby's answer to the "lucky coin flipper" objection is that Robertson's edge was transferable: dozens of Tiger cubs went on to beat both the market and other hedge funds, so skill, not chance, explains the record. The same chapter shows that edge eroding as Tiger grew past its competence into foreign markets and leveraged macro trading.
Buffett's orangutans
In late spring 1984, Columbia Business School staged a clash of titans. Michael Jensen, a dean of academic finance, argued the efficient-market case before an audience packed with professional investors: stock pickers survive only because confused laymen have a psychic demand for answers, however worthless, and managers who beat the market for five straight years are simply lucky. Ask a million people to flip coins and some will throw five heads in a row; there is no skill in coin flipping, and investing is no different.
Warren Buffett's reply, Mallaby says, could serve as a hedge-fund manifesto. He first made Jensen's case better than Jensen. Imagine a national coin-flipping contest in which tails drop out each round: after ten rounds about 220,000 Americans would remain, growing cocky and boasting of their technique at parties; after twenty rounds the 215 survivors would be insufferable, writing fatuous books on the art of flipping. A professor would then point out that 225 million orangutans would produce the same result: 215 egotistical apes with twenty straight wins.
Then he cut holes in it. If the 215 winning orangutans were scattered randomly, luck explains them; but if 40 of the 215 came from the same zoo, something else is going on. Statistics that look random can look quite different once you study their distribution: a rare cancer clustered in one village makes you test the water.
Investment villages
Buffett's zoo was Ben Graham's value-investing tradition. He laid out the records of nine managers: three had worked at the Graham-Newman Corporation in the mid-1950s, and the rest had been converted to Graham's method by Buffett or his associates. He insisted he had not cherry-picked: these were all the Graham-Newman alumni with records and all the managers he had won over. Without exception, and without copying one another's picks, every one had beaten the market.
Mallaby takes the point as indisputable. Viewed as bare performance numbers, a few winners can be dismissed as ten-heads-in-a-row flippers; viewed as members of distinct intellectual "villages" or styles, their success may be clustered in a way chance cannot explain. The hedge-fund story has several such clusters, and the most famous was built by a booming North Carolinian.
The jock's jock
Many market masters are aloof: facts are facts, and social skills do not change a portfolio's bottom line. A manager visiting the macro trader Louis Bacon found an Oz-like figure hidden behind banks of screens; when Bacon later bought a private island, it hardly seemed to change anything, since he was already as isolated as possible.
Julian Hart Robertson came from another tradition: southern charmer, New York networker, with moods that could swing alarmingly. Tall, confident and athletic, a guy's guy, he hired in his own image, and you almost needed the physique. Tiger retreats in Idaho's Sawtooth Mountains meant vertical hikes and outward-bound contests: flown out on Robertson's private plane, teams were given logs the size of telephone poles, rope and two paddles, had to haul them to a lake, lash a raft and race to a buoy, and since not everyone fit on board, some had to swim in icy water. Back in New York the firm kept a private trainer, who would ask a shirking analyst bound for a business dinner in fifteen minutes whether he had time for a two-mile run and a shower first.
Learning from Jones
Like Soros and Steinhardt, Robertson was inspired by A. W. Jones. At Kidder Peabody in the 1970s he befriended Bob Burch, Jones's son-in-law, and the two sometimes took the patriarch to lunch so Robertson could quiz him on the partnership's mechanics. He was close to Alex Porter, a fellow southerner who had bunked in Robertson's apartment and became a Jones segment manager; his sister, a Fortune journalist who had written on Jones-style funds, introduced him to Carol Loomis, who had first explained the hedged structure. A few years after Tiger's launch, Burch entrusted him with $5 million, a fifth of the remaining Jones money: a wise decision, Mallaby notes.
The launch
Robertson opened Tiger in 1980, at a relatively late 48. In the first winter the heating in his tiny office broke, he caught a cold and lost his voice, and squeaked out his orders. His easygoing partner Thorpe McKenzie left no doubt who ruled, recalling that whenever they disagreed, he gave in. Robertson kept buying and selling decisions to himself rather than delegating to segment managers, but otherwise stayed faithful to Jones:
- Long/short stock picking. He bought and shorted stocks, hedging out part of market risk.
- No market timing. He dismissed talk of the market's direction as gibberish and promised clients he would prosper through stock selection.
- A macro overlay. As the fund grew he picked stocks abroad and played commodities, currencies and bonds.
- Derivative hedges. He sometimes hedged with futures and options, tools Jones lacked, while insisting, in Jones's language, that speculative instruments served conservative ends.
- 31.7% vs 12.7%. Tiger's average annual net return from May 1980 to August 1998, against the S&P 500.
- $8.5 million. Starting capital in 1980.
- 20.2% vs 6.3%. Tiger against the S&P in 1984; over half the return came from shorts.
- 51.4%. Net gain in 1985.
- 1988-1992. Five straight years beating the S&P.
- 1991. Becomes the third hedge fund to run over $1 billion, after Soros and Steinhardt.
- $7 billion. Assets by end-1993, more than Steinhardt and slightly less than Soros.
- 64%. Net return in 1993; Business Week put Robertson's personal take at $1 billion.
An affront to efficiency
Mallaby sees the record as an affront to the efficient-market hypothesis. Jones's success was understandable (rivals were committee-bound trust banks); so was Steinhardt's place in the block-trading circle, Commodities Corporation's lead in quants and trend surfing, and perhaps Soros's seer-like feel for turning points. Robertson claimed no institutional loophole, no quantitative trick, no philosophy. A letter to Robert Karr, Tiger's man in Tokyo, summed up the method with telling blandness: manage the portfolio aggressively, replacing good companies with better ones; never risk more than 5% of capital on one bet; and keep swinging through bad patches until luck returns. Robertson and a handful of lieutenants (perhaps a dozen later) simply analysed and bet, exactly what efficient marketers called impossible.
Four explanations, all partial
| Explanation | For it | Mallaby's verdict |
|---|---|---|
| Bull market | 401(k) plans born 1981; by Tiger's end holders had 75% in equities; ~1,550 US firms went private 1981-88, making remaining stock scarcer | Real tailwind (and Tiger rode merger targets), but a hedged fund should lag a rising market, and Tiger nearly always beat it |
| Small-cap focus | Analysts ignore minor firms, so prices may be inefficient | Tiger also won on big names (United Airlines); the seller of an obscure stock is usually well informed |
| Short selling | Jones-style rivals washed out in the early 1970s; brokers issued 10 buys per sell; Prime Motor Inns fell $28 to $1 | A genuine edge, but in most years Tiger beat the market on its longs alone |
| Luck | Any single record can be chance | Refuted by the cubs: 36 ex-Tiger funds ran $100 billion in 2008, plus 29 seeded funds, and beat other hedge funds 2000-08 |
The bull market. Tiger opened at the start of an equity boom. Robertson was in tune with the merger wave because his aim, seeing through the market price to true value, led him to stocks takeover artists also liked. In 1985 Tiger bought Empire Airlines at $9 and was bought out at $15, and the aviation-parts distributor Aviall at $12.50, taken out at $25. Reporting the 51.4% year, he joked that investors not planning to buy emeralds or diamonds should hide the letter from their wives. Still, the shorts meant Tiger was not built to rise as fast as the market, yet it almost always beat it.
Small companies. Some alumni credit a focus on under-analysed small caps, on the theory that a big firm like United Airlines is efficiently priced because analysts pore over it. Tiger did hunt small firms, but it also made money on big ones, spectacularly on United itself. The deeper flaw is logical: the investors who have never heard of a provincial retailer or third-tier bank are not the ones holding it. Tiger bought from people informed enough to own the stock and to want to sell it.
Short selling. Shorting was something most managers could not do, and competition was thin, since nearly all Jones-style funds had been washed out in the early 1970s. Wall Street research was congenitally bullish, with about ten buys for every sell at the big houses in the early 1980s: analysts feared losing access to the companies they covered, and their investment-banking colleagues coveted those firms' advisory fees. So Robertson phoned analysts and coaxed out their best short ideas verbally, telling them he knew the companies were their children but asking to hear about the least favourite child.
Scepticism came naturally. In July 1983 he wrote that the media, public and analysts were so bullish they were eating grass, and that it might be time to crawl into a log and slurp honey. In 1984, when the S&P rose only 6.3%, Tiger made 20.2%, more than half from shorts. In 1985 portfolio manager Patrick Duff suspected the hotel chain Prime Motor Inns was worse off than its accounts showed but, working for a conventional pension fund, could do nothing; after joining Tiger in 1989 he persuaded Robertson to short it, and within a year it fell from $28 to $1. Robertson once told an associate: "It's me and the patsies."
The ten-to-one buy bias Robertson exploited is the conflict regulators later attacked: the 2003 Global Research Analyst Settlement forced US banks to separate research from investment banking, and MiFID II (2018) made European asset managers pay for research separately. For bank risk and compliance teams the lesson is that published recommendations from conflicted desks carry predictable bias; the alpha lies in the unpublished "least favourite child."
Luck. But shorts do not explain everything: in most years Robertson would have beaten the market without them, an edge precisely where theory said none could exist, in ordinary buying. And like Buffett, he headed a village. On one 2008 count, 36 former Tiger employees ran "Tiger cub" funds managing $100 billion, and Robertson had seeded 29 more after restructuring in 2000. Mallaby's own test (Appendix I) shows the cubs beat both the market and other hedge funds over 2000-08, a period when the crowd of long/short funds had long since eliminated easy short-selling profits. Whatever the edge was, it was profitable and transferable.
The real edge: a special-forces unit
Mallaby's explanation starts with an updated Jones innovation. Jones paid for performance, rewarding segment managers by returns and outside analysts by the profits of their tips. Robertson also put rocket fuel in his people, but not only through money: something about him made you want to please him.
Charm and terror
He would melt people with Carolina flattery, telling a junior in an exaggerated drawl that he was utterly paralysed without his powerful assistance. The pleasure was spiked with fear, since his mood could turn ice cold and he might call the same man a petty tin-pot dictator. At the Friday meetings around a long table, the payoffs were extreme: a pitch was either the best idea he had ever seen, leaving a square-shouldered young analyst high-fiving himself inwardly, or the dumbest, leaving six-foot-plus of Wall Street alpha male to shrivel.
Working at Tiger was like joining special forces: the commander made you bigger, brighter and tougher, convinced you could beat the market year after year as part of a team that would out-think and out-hustle everyone. For about the first dozen years he sat at a desk in the open beside his young men, who watched him schmooze and holler down the phone, sucking information from his network.
- The wagon wheels. Two assistants ran a pair of giant Rolodexes almost the size of wagon wheels. If an analyst proposed shorting Boeing, Robertson would recall knowing its former head of international marketing; the assistants found the number, the man came on the speakerphone, and the analyst had to defend his short against him.
- Showing off his stars. Analysts doing well were "big tiger," paraded before his friends: one was introduced to Chicago Bulls owner Jerry Reinsdorf as Robertson's Michael Jordan; a golf-loving lieutenant met Jack Nicklaus and Ely Callaway; a prized employee called Lou was presented to Bill Clinton at the White House as able to do anything. Eighteen months later Lou was gone, joking to a colleague that he had fallen from the White House to the outhouse in record time.
The network of investors
Robertson drew people in by force of personality. He signed up the singer Paul Simon over a shared love of baseball; the writer Tom Wolfe invested too, and Robertson used his stars to recruit.
Robertson wanted Michael Bills, a Goldman Sachs analyst whom Goldman's grandees were promising a great future. He gave Bills the full charm treatment, then wheeled in Lew Lehrman, the red-suspendered financier and philanthropist who had run for New York governor in 1982, to swear he had never met an investor to compare with Robertson. The trump card was Tom Wolfe: Bills's father had been a military pilot, and Wolfe, author of The Right Stuff, called to talk about flying, honour and courage. After a twenty-minute call, Goldman had lost its man.
Tiger's investors included many captains of industry and finance, and Robertson's letters invited them to phone in ideas (or, he added, in the case of ladies, intuition):
- Mentor. Recommended by an investor in the early 1980s; Tiger tripled its money.
- General Instrument and Equitable Life. Among the best picks of the early 1990s, both urged by friends connected to those firms.
- Citicorp. Bought mainly because the bank looked set to rebound once its real-estate losses were cleared, but also because a friend vouched for chief executive John Reed (the book spells it Reid).
Mallaby is explicit that this was not insider trading: contacts offered broad guidance, not secrets about imminent earnings. But Robertson built the network deliberately and cashed in on it brilliantly.
Judgment
Close observers had no doubt about his talent. He could match an analyst's grasp of company detail in a CEO meeting, spot the make-or-break point in a presentation on a firm he knew nothing about, and, seeing a chief executive nudge his ball out of the rough at golf, resolve never to own the man's stock. Jim Chanos, the famous short seller who ran money for all three titans of the 1980s (Robertson, Soros and Steinhardt), found Robertson the most intellectually engaging and said that if he had to give his own money to one of them, it would have been Robertson, because he knew stocks better than anyone.
This edge was hard to define. A May 1986 Institutional Investor profile of "the red-hot world of Julian Robertson" made it sound bland: he did nothing dramatically different, just did it well. And the golf test hardly counted as scientific method. Mallaby's answer is that Robertson's approach, neither formal nor original, was right more often than wrong, which is the definition of success: even builders of state-of-the-art quant systems are delighted to be right six times in ten.
Long-termism
The same six-in-ten rule applied to Tiger's long horizon. Wall Street analysts usually offer a 12-to-18-month view, and hedge funds tend to win by being either shorter or longer term; Robertson was firmly long. His ideal, a "fah-bulous" investment, could plausibly double in three years, and if he thought he had one he would grit his teeth through bad times:
- Oil, 1983. He shorted oil and oil-service stocks; it hurt for a while, but three years later crude had halved.
- Generic drugs, 1984. He shorted generic manufacturers, reasoning that unbranded products could not keep margins; the bet lost money at first, but two years later Zenith Labs fell 45% in one quarter.
In 1987 lieutenant John Griffin persuaded Robertson to short a small appliance maker manufacturing in China. Strong Christmas sales lifted it from $20 to $25, then $35 by spring. In 1988 Griffin left for Stanford business school, begging his boss to hold; Robertson did, even as the stock doubled to $40, a loss of 100% of the original stake. One day a fax arrived at Stanford bearing only: "$50!" Then in April 1989 some 100,000 demonstrators marched into Tiananmen Square, revolution seemed possible, and US firms with Chinese factories collapsed. Griffin ran to a pay phone, ecstatic to be redeemed. Robertson's drawled reply: it had taken a revolution of a billion people for his darn short to work.
Mallaby draws two lessons. Tenacity, like character judgment, does not pay every time, and here luck was decisive. But the story also shows Robertson's genius: Griffin, who later ran his own multibillion-dollar fund (Blue Ridge), had physically left but psychologically never left. In a 2007 speech he said his little birdie had a southern drawl and a bald head, and sometimes chirped at him not to do something.
Hustle
Before the internet and ubiquitous data terminals, hustle counted hugely. To gauge Ford's sales, a Tiger analyst stayed on the phone with customers, competitors, suppliers, dealers, parts makers and Detroit rivals.
- Avon. An analyst weighing a stake in Avon Products became an Avon representative.
- Korean cars. An analyst considering shorting a Korean carmaker with allegedly faulty engines bought two of its cars and hired a mechanic to test them.
- Citicorp and Mexico, 1995. When Mexico's crisis struck at the start of 1995, New York feared for US banks. Tiger's analyst flew down, found Citicorp was not exposed and that Mexicans were eager to bank with Citi now their own banks were weakened; as panicked New Yorkers dumped Citi stock, Tiger bought more.
His gregariousness set him apart from other titans and let him stage a contest between the workaday masses and his own special-forces unit: him against the patsies.
Going global
As the takeover bonanza of the 1980s gave way to the globalization bonanza of the 1990s, Robertson went global too. He was no natural cosmopolitan like Soros, but travel suited his swashbuckling style: he flew to Hong Kong with buyout tycoon Teddy Forstmann, sped through Europe (pronouncing the US embassy in Paris very liveable), regretted the sex ratio on Brazil's beaches but admired São Paulo's business leaders. In each city he would charm his first few hosts into opening their Rolodexes to the top people.
His usual "camp counselor" was John Griffin: smart, tireless, an Ironman triathlete, the ideal lieutenant. They raced between company meetings worldwide with bouts of high-voltage tennis in between; after a lunch at Union Bank of Switzerland in Zurich, Robertson told investors, they grabbed a rental car, turned the music up full and drove to Austria via Liechtenstein to ski.
Germany: from bull to bear
In early November 1989 a fax reached Griffin at Stanford: the Berlin Wall was about to fall and it would be a very big deal. Days later it fell, and two days after that Tiger began loading up on German securities. Robertson knew next to nothing about Germany, but Griffin had studied its market during a London summer, and inexperience was not going to block a historic opportunity. Tiger bought Deutsche Bank (a unification boom), Veba (a utility with power plants along the inner-German border, poised to capture the eastern market) and Felten & Guilleaume (maker of the cables to carry the power east). German stocks went on a euphoric run and the Felten & Guilleaume stake doubled.
The next summer (1990) Robertson and Griffin toured Germany. In East Berlin nobody had heard of hedge funds or of him. On his first company visit he ran a finger along a waiting-room table and held up the dust as a sign of how far these people had to go. By American standards and relative to their assets, German stocks were ludicrously cheap: run the American way they would yield huge returns, and if incumbents were too sluggish, surely Wall Street-style takeovers would fix it. But the more he toured, the cooler he became. Managers asked about return on equity talked about sales, running their firms for employees, not shareholders.
At Bayer he was given a lavish lunch and remarked that it must be good to be chief executive; his hosts replied that all employees got the same meal. Noticing planes flying close by, he learned it was the company flying club, where anyone could train for a pilot's licence. His verdict to Griffin was that these people just did not get it, and by 1994 he wrote off German industry as a "giant flab bag of inefficiency." Mallaby reads the journey from bull to disaffected bear as a warning: branching out in all directions, Tiger was lurching into areas it barely knew, and expansion created new opportunities, including opportunities to be wrong.
Growing large
At the end of 1990 a magazine profile opened by announcing that Robertson had fortune but not fame, a self-cancelling claim, since nothing attracts attention like being called a secret, especially alongside a performance record. Money from the rich and increasingly from endowments flooded in, and in 1991 Tiger became the third hedge fund to manage over $1 billion.
Through the 1980s Robertson had preached that small is beautiful, an edge over lumbering rivals. In the 1990s he threw caution aside, seduced by a bigger staff, a bigger office and grander investors, and became fiercely competitive in the race for assets. He passed $1 billion behind Soros and Steinhardt but by end-1993 ran $7 billion, more than Steinhardt and only slightly less than Soros. Even second place grated: he envied Soros's limelight and could recite their head-to-heads by heart, claiming Tiger had trounced Quantum in 1981, 1983 and 1984 and lost in 1982. Yet Soros still got the attention and the assets.
The cost of size
Expansion forced Robertson beyond his strengths, since US equities offered too few opportunities for a multi-billion-dollar fund. Small caps became virtually off limits: an analyst might find a company that could double in three years, but with only $20 million of stock available it was not worth the bother. Going abroad was one answer, but it bet that American instincts would work in other cultures.
- Germany. Firms that looked cheap by American metrics might be fairly valued given German indifference to profit.
- Japan, the mirror image. With impeccable logic Tiger shorted Japanese bank stocks: mismanaged, lending at negligible margins, weighed down with bad loans, yet trading at huge multiples. In 1992 Robertson told investors Tokyo's bizarre valuations were a rich opportunity; by 1995 his Japan analyst wrote a memo asking why the banks refused to collapse and admitting he did not know.
Into macro
Size also pushed Tiger into more macro risk. Robertson had dabbled early: just over a quarter of the 1985 gain came from shorting the dollar, a smaller version of Soros's Plaza trade, and Tiger profited from a version of Bruce Kovner's carry trade, buying Australian and New Zealand bonds yielding 16% to 24% with money borrowed largely in countries with rates below 10%, locking in the spread. In the early 1990s he doubled down on currencies, hiring Japanese-speaking currency specialist David Gerstenhaber from Morgan Stanley in 1991, followed by currency trader Barry Bausano.
But Robertson was not equipped to be a macro trader. Chanos vouched for his grasp of stocks; no macro trader would have said the same about his grasp of rates or currencies. Mallaby argues that the value-investing mindset almost disqualified him.
Value investor. Little or no leverage, long holding periods; if a stock falls he buys more, since a bargain at $25 is a bigger bargain at $20. Conviction is rock solid because he has torn apart a balance sheet and knows what the company is worth.
Macro trader. Leveraged positions make such averaging down impossibly risky, so he must be ready to jump out when a bet goes wrong. There is no reliable way to establish the objective value of a currency, so no comparable source of conviction.
Tiger's macro men struggled with his style. He could not abide charts, which he had called hocus-pocus mumbo-jumbo (with a ruder last word). Their risk-control instincts rankled: if a trade went against them they assumed they were late or wrong and wanted out, whereas Robertson had conviction and would stay the course to the mountaintop. When they once feared a short-term turn against a European bond position and proposed a temporary hedge, he exploded: hedging just meant making less money if he was right, and that was "dirt under my fingernails." That, Mallaby notes, ended that attempt at risk management.
Robertson's value-investing virtues, conviction and patience, become liabilities in leveraged macro trading, where staying the course on a losing position is how funds blow up. Mallaby plants this as the seed of Tiger's later collapse.
Peak and strain
Through the 1980s and early 1990s the strengths still dwarfed the weaknesses. His special-forces unit parachuted into markets worldwide, stumbling sometimes (as in Germany) but more often succeeding. He often expressed a macro view through stocks:
- After Plaza. With the dollar set for a weak spell, he bought US firms that would gain from strong exports.
- After the 1990 property collapse. He shorted the right banks, then went long financials the moment the bad property debts were cleared.
- Bonds and currencies. The march into macro, though dangerous, paid off spectacularly for a while.
Tiger beat the S&P every year from 1988 through 1992, then returned 64% net in 1993, with Robertson's own earnings estimated at $1 billion.
Absent commander
Then something subtle shifted. In ten weeks at the start of 1993 he spent five long ski weekends in Sun Valley, a long weekend hunting sailfish in Costa Rica, three days at Augusta National and a fortnight in Kenya. He remodelled the office and moved into an elegant corner office away from his lieutenants, and hired a natty psychiatrist, Aaron Stern, to help manage his people and his own mood swings, another buffer. Less present, his intensity grated more: lieutenants went days without seeing him, then he would swoop for facts and fresh scoops. Stern warned recruits that people who bled easily would not be happy there, and many analysts were gone within a year.
The exodus begins
He also lost people he wanted to keep. At the autumn 1992 Tiger party he introduced Gerstenhaber and Bausano as the men who had earned $1 billion last year, a line that made raising money easy; in spring 1993 they quit to found their own fund, the first of a series of departures. Lieutenants felt they should run their own shows, especially as Robertson refused to share control: he toyed with sub-portfolios for lieutenants but never ceded real authority. And leaving was now seductively easy, since wealthy investors were hunting the next stars. Within two months, the pair's Argonaut fund ran $200 million, many times Tiger's $8.5 million start in 1980.
He tried to preserve the culture with mountain trips: hiking up slopes on skins with sixty-pound packs and sleeping in tents below zero; forest scavenger races with bikes; making two quarrelling analysts share a tent. The contrived team building often backfired. Campfire evenings raised expectations of a grand discussion of Tiger's future: would lieutenants get portfolios, what was the succession plan? Robertson had not organized retreats to discuss his own retreat, and he reacted angrily.
On one trip a guide led a group of Tigers to three ropes slung over a ravine several hundred feet deep, in driving rain and wind. Each clipped two safety lines to the rope bridge and stepped out, the bridge wobbling most at the centre. Analyst Tom McCauley slipped and fell about ten feet before a line caught him; laughing, he bounced on it like a baby jumper, in a fluffy white outfit that made him look like the Pillsbury doughboy. Then onlookers saw he was held by only one line, and its carabiner was opening. Amid shouts to stop bouncing, the guide shimmied out and locked it. For the group, Mallaby writes, it was the moment Tiger's reach exceeded its grasp.
Two slips: Citicorp's chief executive was John Reed, not "John Reid"; and Mexico did not formally default in 1995: the peso collapsed in December 1994 and a US-led rescue of about $50 billion prevented default. More broadly, the cub test supporting "transferable skill" is Mallaby's own (Appendix I), on a small sample with possible survivorship bias, since failed cubs drop out of databases; and cubs may share Tiger's network and investors, not only its method. Robertson's Soros scorecard is his own recollection, not an audited comparison.
Julian Robertson died in August 2022, aged 90. Several cubs became giants, notably Chase Coleman's Tiger Global, which itself suffered one of the largest hedge-fund losses on record in 2022, when tech stocks crashed.
Rock-and-Roll Cowboy
Paul Tudor Jones made 200% in 1987 and 80-90% in 1990 not by forecasting supercycles but by writing a script for the market and betting hard when the odds were skewed in his favour. Mallaby isolates three edges: reading how other players are positioned, trading against forced sellers, and, most distinctively, becoming the trigger that starts a stampede. Jones's distinctive strength was self-awareness: he understood how his own trades and reputation changed other traders' behaviour, a power that would grow dangerous as hedge funds grew large enough to move whole markets, and even governments.
The post-crash boom
The late 1980s were a turning point. After the early-1970s bear market the industry had nearly vanished: a few dozen funds operated under the radar with negligible capital. After the 1987 crash, something profound changed. One count found 600 hedge funds by 1990; by 1992 there were over 1,000. Commentators talked knowingly of the "Big Three" (Soros, Robertson and Steinhardt), and 1993 was celebrated as "the year of the hedge fund."
Behind the Big Three a younger pack was expanding fast, and the stars' connections created the sense of a movement:
- Family and partnerships. Robertson was related by marriage to the young trader Louis Bacon; Steinhardt had briefly gone into business with Bruce Kovner of Caxton.
- Social life. The moguls hunted and fished on one another's estates, sat on the same charity boards and met each year at Lyford Cay in the Bahamas; their presence in one room signalled a new force on Wall Street.
The cotton cowboy
The most colourful of the young titans was Paul Tudor Jones II, Soros's friend and Patton fan. Born in 1954 to a Memphis cotton family (his grandfather prospered as a cotton merchant and his uncle Billy Dunavant was a baron of the industry), he studied economics at the University of Virginia, apprenticed with a New Orleans cotton trader, then after two years moved to the New York Cotton Exchange.
On the floor it was hard to believe markets efficiently reflected information. Prices were driven less by data than by the cotton cowboys who downed martinis at lunch and charged back into the pit to gun the market up or down as it felt to them. They reacted to growth and jobs figures, but survival depended less on understanding the news than on foreseeing the pit's reaction. Mallaby retells the story of a floor trader who made $10 million in a burst of trading after an inflation report, then walked out of the ring and asked what the number had been.
Tudor Investment Corporation
Jones left the floor in 1983 to build Tudor Investment Corporation, not yet thirty. Commodities Corporation invested $35,000 and connected him with veterans who validated his view of markets. He quickly emerged as a prodigy who treated trading as psychology and high-speed bluff, a poker of sly subtlety and crazed bravado:
- Read the other hands. Bullish sugar news matters only if the big traders have not already bought their fill; if they are waiting to rush in, the price takes off like a rocket.
- Get inside their heads. Watch rivals long enough and you can lure them in when they want to buy and spook them out when they are fearful.
- Bluff, then reverse. If big traders were nervous, you could shout that you were selling, knowing they would sell too, then turn round and buy heavily.
Loudness as a weapon
Jones called himself a pure cowboy and traded with the violence of the boxer he had been at college. He screamed huge orders to brokers on several exchange floors, often reversing them mid-call, knowing the crazier he sounded the more off balance rivals would be. He varied tactics: sometimes small orders through many brokers to stay hidden, sometimes an ambush, guns blazing, knowing a sudden big buyer would send others scrambling. Perhaps because notoriety was itself a weapon, in 1986 and 1987 he let a documentary crew film his trading tantrums.
Minutes before the open Jones sits calm and preppy: white shirt, conservative tie, signet ring, big glasses, tidy parting, a Memphis drawl counting down. Then he erupts, leaping up, pumping his fists and roaring at the speakerphone that he is tom-tomming on the market, barking an offer, countermanding it, ordering a market sale, while squadrons of floor traders fan out to obey. After placing a multimillion-dollar bet he swaps his shoes for sneakers bought at a charity auction that had belonged to Bruce Willis, claiming they were good for a point rally in bonds and thirty dollars in stocks each time he wore them. Then he sets up an inflatable Godzilla on top of a trading screen.
The explanations that do not explain
What intellectual process lay behind the antics? Mallaby says the answer is subtle, because Jones's own explanations were unconvincing.
- The 1929 overlay. His twenty-something chief economist Peter Borish, who wore old-fashioned suspenders to look older, plotted the 1980s stock chart over the 1920s, found both rising vertiginously, and concluded a 1929-scale crash was coming. In a remarkably candid interview he admitted fudging the start points until the lines fitted.
- A house of anecdote. On that quasi-quantitative base Jones piled signs of excess: Wall Street's exorbitant pay, thinly stretched bank capital, a van Gogh fetching ten times its pre-auction estimate. In the 1987 film he gleefully predicts a bloodcurdling collapse: "It's going to be total rock and roll."
- Kondratiev waves. Like Tony Cilluffo, the autodidact behind Steinhardt's 1970s success, Jones liked the theory of predictable 24-year cycles. It had helped Cilluffo foresee 1973, which would put the next cataclysm in 1997; yet Jones thought it supported an imminent crash in 1987.
- Elliott waves. He was even keener on Robert Prechter, who confidently predicted one last explosion upward and then a plunge of at least 90%, the greatest crash since the South Sea bubble of 1720. Jones apparently sincerely credited much of his success to the Elliott approach, but Prechter's doom was wildly overblown, and Jones conceded it could not time anything.
Jones may have liked the world to think Borish had a crystal ball, since it struck fear into rivals, who after returns of 136% in 1985 and 99% in 1986 might believe anything. But the 1920s work was incidental: most of Wall Street expected a break eventually, the hard part was the date, and Borish's date, spring 1988, was no better than the others that littered the business.
Writing the script
The real source of profit, Mallaby argues, was agile short-term trading, not dubious multi-decade supercycles. Like the Commodities Corporation traders, Jones rode waves, mounting the surfboard when a swell seemed to be building and jumping off fast if the market turned. He admitted: "When you take an initial position, you have no idea if you are right." His method, in candid moments, was to write a script for the market (how it might behave) and test the hypothesis repeatedly with low-risk probes, hoping to catch the moment the script became reality. Years later he described the mental routine: each evening, eyes closed in a quiet room, he pictured himself in the pit, walked through the opening and the day, and imagined each emotional state the market would pass through, so that when extremes came he had already lived them and could exploit them.
- Script. Write out how the market could behave and what would trigger the move.
- Visualize. Rehearse the day's emotional swings in advance, so panic or euphoria finds you prepared.
- Probe. Test the script with small, low-risk positions; cut them fast if wrong.
- Look for asymmetry. Bet heavily only when the potential gain dwarfs the potential loss.
- Press. When reality matches the script, scale up hard.
Black Monday: the double coup
The 1987 crash showed the power of this preparation. The moment the S&P 500 began falling on Friday 16 October, Jones sensed the long-expected break might be arriving. Borish's chart had said the crash was months away, but Jones never took that literally; what mattered was that he had visualized a crash and saw that once the market fell, the odds of a monster fall were high. Investors had awaited a reckoning for months and their confidence could crack; portfolio insurance meant falling prices would trigger more selling. If Friday's dip fizzled, a short would lose modestly and he would close it; if skittishness and insurance cratered the market, the payoff would be enormous. The balance of risk and reward was overwhelmingly attractive.
By Friday evening he had sold armfuls of S&P futures, then flew to his remote Virginia hunting lodge with Louis Bacon and European friends. When too many guests wanted seats on the private plane back to New York, Jones chivalrously offered to stay behind; someone refused, knowing he had a big position. So on Black Monday morning he was at his Manhattan desk, and rode the cascade all the way to the bottom.
Panicking investors jammed brokers' phones; the only calm people were the numb. Jones later compared Wall Street's reaction to being run over by a boat: with his backside chewed by the propeller, his first thought was annoyance at a ruined Sunday, since shock hid the damage until he saw his friends' faces. The crash paralysed people the same way, but Jones had his script and was ready.
The second bet
As he rode stocks down, he saw a second trade. Scripting the Fed's response, he reasoned it would calm nerves by pumping cash into the banks and making borrowing cheaper. That was another asymmetric bet: if the Fed acted, bonds would soar; if it did nothing, there was no reason for bonds to fall. When bonds ticked up late on Black Monday, he took it as the signal his script was coming true and bought the largest bond position he had ever held, which became his most profitable.
- $35,000. Commodities Corporation's seed investment in 1983.
- 136% / 99%. Returns in 1985 and 1986.
- $80-100 million. Reported take from the Black Monday double coup.
- 200%. Tudor's return for 1987.
- 80-90%. His own estimate of the 1990 return, mostly from Tokyo.
Not long after, Jones showed a side at odds with the cowboy act, founding the Robin Hood Foundation, which channelled millions of dollars of new hedge-fund wealth to New York's poorest neighbourhoods.
The Commodities Corporation trio
Jones's triumph was no fluke. The Big Three all lost heavily in 1987, while Kovner and Bacon, also from the Commodities Corporation tradition, did well, if less spectacularly. Both groups had expected a reversal; they discussed it often, and Jones had even tried to persuade Robertson to run a portfolio of stock shorts for him. The difference was speed of response. A stock picker like Robertson was wedded to his stocks, each exhaustively researched by his Tigers and painful to dump. Jones, Kovner and Bacon carried none of that baggage: they did not care about individual stocks, they traded the whole market, and they could turn on a dime.
Equity tradition (Soros, Steinhardt, Robertson). Positions built stock by stock through deep research; emotional and practical attachment; large, sometimes illiquid holdings; slow to exit.
Commodity tradition (Jones, Kovner, Bacon). Positions in index futures and bonds; trend surfing with fast stop-outs; no attachment to names; able to flip short to long and pile into bonds the same week.
After the crash Kovner and Bacon also piled into bonds, following Jones's script, and thrived for the rest of the decade:
- Kovner. Reaped glorious rewards in currencies, building on the carry trade he had developed at Commodities Corporation, and was reportedly Wall Street's highest earner in 1989 and 1990, helped by bets on oil futures.
- Bacon. Did so well trading a small account as a broker at Shearson Lehman Hutton that he left to found Moore Capital in 1989: up 86% that year and 29% in 1990, having correctly foreseen how Iraq's invasion of Kuwait would hit equity and oil markets. His Commodities Corporation seed was soon dwarfed by other money.
- Jones. Unable to absorb all the money pressed on him, he advised clients to invest with Bacon instead.
Tokyo, 1990
Like everyone sophisticated on Wall Street, Jones saw a Japanese bubble forming in the late 1980s. After Plaza the authorities had cut interest rates aggressively to offset the strong yen, and the flood of cheap capital inflated Japanese assets and foreign ones too: Japanese money became the key buyer of everything from California golf courses to impressionist paintings. In 1987 Nippon Telegraph and Telephone floated at a fantastical price-earnings ratio of 250. The market was overvalued on any sane measure and still rose.
As with every bubble, the hard part was timing. Shorting Tokyo hard after the NTT flotation would have been suicide: over the next two years the Nikkei rose 63%, proof that few things are costlier than tilting at a bubble early. Jones, a trend follower, would not bet against bulls with momentum. He waited, and at the start of 1990, when Tokyo fell nearly 4% in a few days, he had his signal.
The 8% script
At a Barron's roundtable in mid-January 1990 Jones listed his reasons. He began with the standard point that Tokyo traded at huge earnings multiples, but, as with Wall Street in 1987, his passion was positioning. Portfolio insurance had created an asymmetry in New York; in Tokyo, financial culture did the same. Japanese savers expected fund managers to deliver 8% a year, and when equities turned down, managers would rush defensively into bonds, where they could lock in 8% risk-free.
The timing of the first fall was everything. A December fall after strong gains would not have mattered: managers above the hurdle would not care, and a fortnight of bond income was too small to bother with. A January fall was different. Managers had no cushion of earlier gains and 50 weeks remained, enough to secure 8% by hiding in bonds. If they behaved defensively, the stampede out of equities could push the market off a precipice.
Trading the rallies
The script played out. The Nikkei 225 fell 7% in February and 13% in March, and by year-end had lost two fifths of its value, crippling what had been the world's biggest stock market. But Jones also anticipated the fluctuations with uncanny accuracy. From earlier bear markets he predicted in January that the fall would be followed by a weak rally; when the Nikkei stabilized in spring he switched from a heavy short to a mild long, and profited again when it rose 8% in May, though convinced the rally was temporary. Mallaby contrasts this flexibility with a value investor like Robertson, who never traded in and out.
In a May 1990 Barron's interview, congratulated on his January call, Jones modestly recalled predicting a Japanese crash prematurely in 1988 and 1989; the essence of his skill was making such predictions without placing losing bets on them. He forecast rallies on the way down, each followed by a steeper dive: investors who had not sold in the first leg would be hoping to recoup, and each rally too feeble to restore their losses would make more of them give up. He was lightly long but planned to be short again by late summer; Tokyo duly fell steeply from July to early October.
Jones's real achievement was spotting when the odds justified a bet without predictive certainty. In October 1987 and January 1990 the market could have recovered, costing him a small loss; but a fall was more likely, and would be far bigger than any rally. Mallaby's image is a doctored roulette wheel with two extra red numbers and a croupier paying five to one on red: you cannot be sure of winning, but you know it is time to shove your chips on the table.
Three edges against efficiency
Many of Jones's successes followed the same formula, and Mallaby breaks it into three edges, each a crack in efficient-market theory.
1. Positioning
From the cotton floor he knew that whether the big players are sitting on cash or fully invested tells you which way a market can break, and so the mix of risk and reward. Pit traders saw rivals' positions as they shouted bids; off the floor, Jones improvised substitutes:
- Calling around. Brokers for big institutional clients, commercial hedgers in the commodity markets, and fellow hedge-fund managers.
- Options data. Whether investors were buying more calls (expecting rises) or puts (fearing a crash).
- Allocation reports. The balance of cash and stocks in pension and insurance portfolios.
- Objectives, not just holdings. What investors wanted to hold and how they would react: knowing Japanese managers' fixation on 8%, you knew they might flee to bonds after a January fall.
2. Institutional distortions and forced sellers
Jones extended the Commodities Corporation insight that behavioural quirks tint the randomness imagined by theorists, adding institutional biases, which sometimes combined with psychological ones:
- Cotton. Farmers psychologically clung to part of the harvest for weeks, hoping for higher prices, until year-end tax rules forced them to sell: a wave of selling every December, recovered every January.
- Cattle. A calf takes nearly a year to gestate, so supply responds sluggishly to higher prices and uptrends in cattle futures last.
- Quarter-end Fridays. Dow stocks tended to rise on the closing Friday of each quarter, when arbitrageurs bought back shorts that hedged expiring futures.
This let him beat the market without prophecy. The theorists showed that future price moves are hard to foresee because today's price contains all information; Jones sidestepped that by getting better than the going price from a forced seller (or buyer). It resembles Steinhardt's block trading, buying at a discount from institutions that needed to unload size and would pay for the privilege. Neither man needed godlike foresight: they were paid for providing liquidity when it was needed.
3. Becoming the trigger
The third element goes back to the documentary's swagger. Jones looked for the trigger that would set off a sudden move, and sometimes became the trigger himself, jump-starting a reversal with a massive trade to make his script come true.
This exploits another weakness in efficient-market thinking. The theory assumes that if, say, Ford is underpriced, a few smart investors will buy until it reaches its efficient level. But their firepower is limited: they may lack the cash to keep buying, and when a whole market is out of line (Japan's equities, later dot-coms or mortgages) they cannot borrow enough to bet against it with deflating force. That is why contrarians' power is limited, why markets trend, and why finance is prone to bubbles. Economists Andrei Shleifer and Robert Vishny named it "the limits to arbitrage." Jones grasped the corollary intuitively: a trend can run far past rationality because speculators lack muscle, so a trader with more ammunition and courage can ambush the market, jolt it out of its sleepwalk, and be first to profit from the new trend.
Pit cowboys had always wrong-footed rivals, but Jones was perhaps the first to do it across many markets. Most "upstairs traders" hid their positions behind many brokers; Jones saw that sometimes it paid to be as loud as possible.
In one documentary sequence OPEC agrees a production cut and oil rallies. Jones doubts the cartel will keep its quotas, so the trend has no basis. First he works by stealth, feeding in small sell orders so as not to disturb the rally while he builds a short. Then he turns cowboy, yelling at his broker to raise the offer and "Show 'em size!", telling him to say there is more behind it. He phones a friend at an oil-trading firm to gauge the mood and learns that the traders there think someone has sold heavily; delighted, he hopes they imagine a reckless Arab insider who knows the deal is about to collapse. The rally reverses: rather than await the trigger for his script, he created it.
Gold had rallied, silver usually followed, and there were rumours of mine disruptions. Early one March morning Jones ran a pincer worthy of Patton: he bought a gutsy position in silver futures from floor traders, leaving them short, then bought physical silver from four dealers. The dealers, expecting silver to follow gold, rushed to replace their inventory, but the floor traders who normally supplied them had already sold to Tudor. Getting urgent buy orders, the traders assumed the supply rumours were true and scrambled to buy back contracts they had sold to Jones. Shorts on both sides chased the price up; Jones left with a handsome profit.
The avalanche and the reputation
Jones claimed no boundless power. He said that by giving a market a little gas at the right moment he could create the illusion of a bull market, but unless it was sound, prices would drop the moment he stopped buying. Mallaby finds the admission more striking than the caveat: that he could move a market at all, even briefly, was remarkable. His image is a boy atop a freshly snowed mountain: he cannot make snow fall uphill against economic forces, but if a mass of powder is ready, one well-aimed stone sets off an avalanche of money at a time of his choosing, and he wins first.
Size and reputation compounded the effect. Seeing the cowboy coming, traders assumed the market would move, and their assumption was self-fulfilling; the more his size and notoriety grew, the greater his power, even in the deepest markets. In the late 1980s Treasury bond futures traded routinely in $50 million lots, and Jones would send a broker into the pit with twice that, panicking traders who wanted to be on the right side of the stampeding elephant. James Elkins, among the biggest S&P futures traders of the early 1990s, recalled that even when his own orders were larger, Jones's arrival threw the whole pit into a scare.
Jones's loud ambushes sit close to lines regulators now police. Since Dodd-Frank (2010) US law bans spoofing (orders meant to be cancelled), and the EU's Market Abuse Regulation (2016) covers trades giving false signals of supply or demand. Bank trade-surveillance teams monitor exactly the pattern of quiet accumulation followed by conspicuous size. Jones's trades were real, not cancelled, which is the legal distinction.
Self-awareness
Mallaby's summing up: from the cotton pit Jones understood that psychology moves markets, anticipating the behavioural finance findings that surfaced from the late 1980s; he saw institutional quirks, anticipating the academic work on tax-driven trading that put kinks in the efficient-market hypothesis. But his most distinctive strength was being self-aware: he knew how his own trading changed other people's calculations and could set off profitable chain reactions.
The bigger hedge funds grew, the more that insight mattered. As billions flowed to the famous titans, it was no longer only oil or silver traders who had to beware: from the 1990s, hedge funds were large enough to move markets of every kind, and even to overpower governments, the subject of the next chapter.
Mallaby credits Jones with anticipating behavioural finance and tax-driven anomalies, but the evidence is anecdotal, drawn largely from a promotional documentary and from interviews Jones gave to shape his image; the 1985-90 return figures are reported or self-estimated, not audited. The "self-fulfilling reputation" argument also cuts against hedge-fund defenders later in the book: a trader who can start avalanches is precisely what critics of speculation fear. Two smaller points: Kondratiev's long waves are usually put at roughly 45-60 years, so the 24-year cycle Mallaby attributes to the theory reflects the traders' own adaptation rather than Kondratiev; and Shleifer and Vishny's "The Limits of Arbitrage" appeared in the Journal of Finance in 1997, so the label is applied to Jones in hindsight.
The Commodities Corporation trio later retreated from outside money: Kovner retired from Caxton in 2011, and in November 2019 Bacon announced Moore Capital would return client capital after three decades. Jones's Tudor still manages outside money.
White Wednesday
In September 1992 Stan Druckenmiller and George Soros sold about $10 billion of sterling short, helped force Britain out of Europe's exchange-rate mechanism (ERM) and made over $1 billion on the pound alone, announcing hedge funds as a force that could overpower governments. The trade fused equity research, macroeconomics, an eye for institutional asymmetries and a close reading of European central-bank politics, with Soros supplying the instinct to bet the maximum. Mallaby's verdict is double-edged: Britain invited the crisis by clinging to an untenable peg, yet once one hedge fund had shown it could break a central bank, the risk that speculators might attack sound currencies became real.
The ninth permanent successor
Druckenmiller agreed to join Soros Fund Management in autumn 1988, against his friends' advice and half expecting to be gone within a year. Soros had wooed him as a genius who would inherit the fund, but his record suggested he could not cede real authority. The day before starting, Druckenmiller visited Soros's weekend house in Southampton, where Soros's son Robert greeted him on the lawn with congratulations on becoming his father's "ninth permanent successor."
The two were stylistic opposites: Soros wrote philosophy for fun, Druckenmiller watched the Pittsburgh Steelers at Three Rivers Stadium. As investors they fitted perfectly, because both came from stock picking and neither was wedded to it.
A combiner of disciplines
- Fast promotion, thin craft. Druckenmiller started as an equity analyst at Pittsburgh National Bank and became research director at 25, too quickly to build an edge in balance-sheet analysis.
- Equities plus economics. His strength was combining disciplines: a feel for currencies and interest rates from an economics PhD he abandoned, joined to a solid sense of stocks. A colleague said he knew the stock market better than economists and economics better than stock pickers.
- Two-way signals. Talking to company executives gave early warning of economic trends, which shaped his view of bonds and currencies; following economies gave early warning for stocks. A falling currency meant buy exporters; rising rates meant short property developers.
- Charts for timing. His first Pittsburgh boss was a chartist, and Druckenmiller found technical analysis, scorned by most stock pickers as voodoo, useful for knowing when an overvalued asset would correct. It let him join Jones's trading agility to Robertson's stock-picking. He came through the 1987 crash and profited richly in the days after, which no manager from a pure equity background managed, Soros included.
Duquesne, Dreyfus and the first lunch
After four years at the bank, Druckenmiller gave a presentation in New York; a listener asked what he was doing at a bank, urged him to go independent, and offered $10,000 a month just to talk to him. In February 1981, aged 28, he launched Duquesne Capital Management, developing a new-school macro style that blended company views, economic views and charts in a freewheeling portfolio. Four years later (1985) Dreyfus asked him to run several mutual funds alongside Duquesne; one rose 40% in three months, making him a Wall Street celebrity. When The Alchemy of Finance appeared in 1987, he recognized his own blend of fundamental and technical trading in Soros's; at their first lunch the two clicked, and Soros tried to hire him before it ended.
"Don't leave. I'll leave."
At first Robert Soros's jibe looked prescient. Quantum's incumbent analysts, including earlier "permanent successors," fought the newcomer (built like the footballers he watched), and Soros treated him as an S&P 500 specialist despite his bond and currency record. One evening in August 1989, visiting Pittsburgh, where he still ran Duquesne, Druckenmiller discovered his Quantum bond position had been sold behind his back. He phoned Soros, shouted that he felt cramped and intimidated by him, and said he wanted to leave.
Soros's reply became, in Mallaby's judgment, one of his best gambles: he would leave instead. Although much of Quantum's money was his own, he moved his family to London, telling Druckenmiller they would now discover whether Soros had been in his hair or Druckenmiller was genuinely inept.
- For Soros. The Berlin Wall fell, and London freed him to concentrate on philanthropy in eastern Europe.
- For Druckenmiller. The wall's collapse brought the turbulence he loved and an astonishing run of returns.
- The new roles. Having learned from the outburst, Soros never undercut him again. Through the 1990s he acted as coach, prodding with questions and advice but leaving Druckenmiller to pull the trigger, while trading a small side pot to satisfy his own urge to bet.
- The credit. Journalists kept attributing Quantum's success to Soros, partly because Soros did little to correct them and partly because Druckenmiller hated the limelight as much as Soros enjoyed it.
- 31.5%, 29.6%, 53.4%, 68.8%, 63.2%. Returns for 1989 through 1993; a colleague likened the run to Bob Dylan in the mid-sixties turning out one hit album after another.
- $1.8bn to $5bn. Growth in Quantum's assets over the period.
- $8.3 billion. Soros Fund Management's total assets by end-1993, after new funds were opened beside the flagship.
- $2 billion. Long Deutsche marks bought within a few days of the Berlin Wall's fall; the mark rose a quarter against the dollar over the next year.
The Deutsche mark rehearsal
From the moment Soros left, Druckenmiller ran an extension of the coach's method. He had absorbed Alchemy and two years of constant conversation, and copied the structure: Quantum's capital in a long/short equity book, with borrowed money used to trade S&P 500 futures, bonds and currencies; regular contact with company executives as early warning of economic trends; and above all the lesson to pile in with everything when the right moment came.
The fall of the Berlin Wall in November 1989 supplied that moment. East Germans poured west expecting jobs and benefits, so Germany seemed bound to run large budget deficits. The standard logic said deficits mean inflation and a weaker currency, so traders sold marks and the mark dipped against the dollar. Druckenmiller recalled the Alchemy passage on Reagan's early deficits, which had strengthened the dollar because the Fed offset loose budgets with tight money, and high rates drew investors into dollars. Germany's hawkish central bankers would do the same: loose budgets would push the Bundesbank to raise rates, and the mark would rally. He bought a $2 billion mark position in a few days.
Mallaby links the mark trade to Jones's insight that understanding the other players reveals bets with superb risk-to-reward ratios, but Druckenmiller played a bigger game: reading governments. Central banks advertise their intentions (the Bundesbank made no secret of its war on inflation) and their actions move markets, so merely seeing that the Bundesbank would tighten was enough to create a trend he could ride.
The mark trade powered Quantum's 1990 return (given here as 29.8%), but Mallaby calls it a dress rehearsal for the coup two years later that shattered Europe's monetary order.
German unification and the ERM
Unification did more than make the Bundesbank predictable: it exposed its conflicting roles as guardian of the mark and anchor of the ERM. Set up in 1979, the mechanism damped currency swings inside Europe so companies could trade and invest without exchange-rate shocks wrecking their plans. For over a decade it worked without going as far as a single currency:
- Narrow bands. Member currencies could move against one another only within set limits.
- Negotiated devaluation. If the band was not flexibility enough, a country could agree a realignment with its partners.
- The balance. These rules left national governments some room to set interest rates for their own cycles, trading off exchange-rate stability against monetary flexibility.
Unification broke the balance. It created inflationary pressure in Germany, so the Bundesbank raised rates just as the rest of Europe sank into recession and needed cuts. Money flowed into marks, and the weakest currencies, the Italian lira and the British pound, sat near the floor of their bands. Europe had two remedies: Germany cut while Italy and Britain raised, or central banks intervened, selling marks and buying lire and pounds. If both failed, devaluation followed.
Setting up the sterling bet
In summer 1992 Druckenmiller focused on Britain. Scott Bessent, a young Quantum portfolio manager, had studied British housing and shorted housebuilders. He pointed out that most British mortgages were floating-rate: any Bank of England rise hit family budgets at once. That transmission mechanism put Britain in a vice. Raising rates to defend the pound would immediately squeeze consumption in an economy already in recession.
Druckenmiller saw a bet that could hardly lose. There was a real chance Britain would refuse higher rates and let sterling fall; there was almost no chance sterling would rise against the mark, since a depressed Britain would never raise rates more than it had to. He bought marks and sold pounds, building $1.5 billion of the position by the end of August.
- Equity research as economic signal. Housing stocks revealed the mortgage channel; like Soros, unlike the Commodities Corporation trio, he treated companies as a forecast of the economy.
- Cross-asset fluency. An equity man equally at home in currencies and rates; like Soros, unlike Robertson.
- Institutional asymmetry. Britain's floating mortgages played the role that Japan's 8% hurdle had played for Jones in 1990.
- Financial politics. The fourth and decisive talent: reading how the Bundesbank would weigh its anti-inflation mandate against its duties to Europe.
Reading the Bundesbank: Bath and Basel
Since the hyperinflation that helped bring Hitler to power, Germans had prized monetary stability. The US Federal Reserve has a dual mandate (low inflation and full employment); the Bundesbank's mission was inflation alone. Its instinct would be to refuse cuts while unification deficits lasted, and the longer it held out, the greater the strain on sterling. Yet politics might soften it: Europe's leaders had just signed the Maastricht Treaty, which envisaged a single currency, and the German government backed the project. Bonn might hesitate to fight inflation so zealously that Europe's monetary order splintered.
At Bath on 4-5 September 1992, European Community finance ministers and central bankers met. British chancellor Norman Lamont, desperate for room to cut British rates and egged on by recession-hit Italian and French colleagues, pressed again and again for German easing, finally banging the table at Bundesbank president Helmut Schlesinger and demanding to know why he would not cut when twelve finance ministers wanted it.
- Schlesinger's reaction. So shaken he nearly walked out, a career defender of the bank's independence who resented political pressure, above all from a foreigner. Once composed, he said he had no plans to cut but saw no reason to raise rates either.
- Lamont's spin. He sold that remark to the press as a concession, though nobody had expected a rise.
- The backlash. Stung, Schlesinger moved to restore the Bundesbank's independence. On 8 September, after a central bankers' meeting in Basel, he said publicly that he could guarantee nothing about future rates, voiced little confidence in Europe's fixed parities, and singled out the lira as unsound.
Soros was in the audience. Afterwards he checked he had heard correctly and asked Schlesinger about the ECU, the notional currency that preceded the euro. Schlesinger said he liked the idea of a European currency but disliked the name; he would rather call it the mark. Soros read this as decisive: the Bundesbank welcomed union only on its own terms, and weaker economies that could not stomach German discipline should devalue. He suspected Schlesinger would happily let his hard line sabotage monetary union, since union meant a European central bank that would supplant the Bundesbank. All bureaucracies seek self-preservation, Soros reasoned, and a lifelong Bundesbanker embodied it. He phoned Druckenmiller, who added a short on the lira to the sterling short.
Back in New York, Soros called Robert Johnson, a currency expert moving from Bankers Trust to Soros Fund Management, to ask about sterling. Because Bankers Trust recorded its traders' calls, Johnson came over by cab to the shabby offices (duct tape on the carpet, a couple of screens by Druckenmiller's desk). Around a small table, Soros and Druckenmiller cross-examined him:
- Downside. Sterling was liquid, so a losing position could always be cut: maximum loss about 0.5%.
- Upside. If the peg broke, about 15-20%.
- Probability. Over three months, roughly 90%.
- Sizing. In his own Bankers Trust fund he would leverage 3-5 times capital. Druckenmiller visibly tensed like a player about to spring.
- The ceiling. Britain had only £22 billion (about $44 billion) of reserves; with private buyers scarce after Schlesinger, the Bank of England was the main buyer, so its reserves capped how much anyone could sell. Druckenmiller mused that Quantum might take $15 billion of that, ten times his existing bet.
- Time. Johnson thought Britain could hold out a few months at most.
Bessent, patched in by phone, went further: the government had no stomach for higher rates and, choosing between deeper recession and devaluation, would devalue, perhaps sooner than anyone expected. Johnson left sensing the two men's coiled energy: when the moment came, they would destroy the pound.
A watershed: the herd turns on Europe
Mallaby treats the following days as a watershed between governments and markets, showing that a huddle of traders in midtown Manhattan could have global consequences. Quantum led, but other currency-trading hedge funds (the Commodities Corporation trio among them), bank trading desks and corporate treasuries joined the attack on Europe's weak currencies.
- The old balance. Through the 1970s and 1980s nobody imagined private players could overwhelm big central banks; the Plaza accord (1985) had confirmed governments' grip on exchange rates.
- The new balance. Since the mid-1980s cross-border flows had roughly tripled, and hedge funds and others now held large war chests.
- The proof. In August 1992 the George H. W. Bush administration organized dollar purchases by eighteen central banks, which failed to budge the dollar.
Thu 3 Sep Britain borrows 10 billion ECUs (£7.25 billion, $14 billion) to defend sterling; the pound jumps.
Fri 4 - Sat 5 Sep Bath meeting; Lamont spins Schlesinger's "no rise" as a concession.
Tue 8 Sep Schlesinger's Basel warning; Finland abandons its ECU peg and the markka falls nearly 15% in a day.
Wed 9 Sep Run on Sweden, repelled with overnight rates of 75%; the herd turns on the lira despite Italian rates of 15%.
Fri 11 Sep The lira breaks through its ERM floor despite DM 24 billion ($15.4 billion) of Bundesbank support.
Sat 12 Sep Italy negotiates devaluation; Lamont sings "Rule Britannia" at the Last Night of the Proms.
Sun 13 Sep "Cock-a-hoop" Treasury breakfast; a quarter-point Bundesbank cut looks helpful for sterling.
Mon 14 Sep A $700 million intervention lifts sterling slightly.
Tue 15 Sep Sterling battered; Schlesinger's interview hits the wires in the evening.
Wed 16 Sep Two rate rises fail; Britain leaves the ERM at 7:30 p.m.
Finland, Sweden, Italy
On Tuesday 8 September, the day of Schlesinger's Basel remarks, speculative selling overpowered Finland's central bank and forced it off its ECU peg; the instant profits whetted traders' appetite. On Wednesday speculators ran at Sweden, which won the capital back only by pushing overnight rates to 75%, then stampeded the lira. Italy's 15% rates would once have held money at home, but the swollen currency market changed the game: the lira offered an asymmetric bet, its trend invited technical traders to surf, and the wave of money was plainly larger than any official response. By Friday 11 September the lira was through its band floor, and Italy negotiated a formal devaluation that weekend.
The lira was the more sobering case. Finland was outside the ERM and could expect no help; Italy was a member, entitled to Bundesbank support, and the Bundesbank bought an unprecedented DM 24 billion of lire in the week before devaluation. The markets still won: it was the first time an ERM member had been so bloodied.
Britain's false dawn
London failed to grasp the new order. On Saturday 12 September, while Italy negotiated with visiting German officials, Lamont kept his diary and sang "Rule Britannia" lustily at the Last Night of the Proms. At a croissant breakfast with Treasury advisers next morning, the team still saw no immediate crisis; one press account called Lamont "cock-a-hoop," because the lira deal included a quarter-point Bundesbank cut that might even lift sterling.
The optimism matched the conventional wisdom:
- Not comparable. Analysts argued that Italy was Europe's most shambolic rich economy while Britain had a Conservative government that had transformed its economic performance.
- A track record. The Bank of England had fended off pressure since August.
- A new weapon. On 3 September Britain had copied the hedge funds by borrowing 10 billion ECUs to defend the pound, and traders believed it now had the firepower. Soros and Druckenmiller were amused: the sum equalled what Quantum alone hoped to sell. Nobody outside their office could imagine a fund of fewer than fifty people matching a government's war chest.
On Monday 14 September the Bank spent $700 million, and with the German cut that modest support lifted sterling a little. Mallaby sees this as the day Britain's fate was sealed: the small rise confirmed the speculators' premise that bets against shaky pegs could be leveraged heavily, because the worst case was a slight move against you. On Tuesday the pound took a beating; asked by Spain's finance minister how things were, Lamont said they were awful.
The Schlesinger interview
That evening Lamont met his Treasury team and Bank governor Robin Leigh-Pemberton. They agreed to intervene heavily next morning and, failing that, consider a rate rise. As the meeting closed, Leigh-Pemberton read a press-office note: Schlesinger had told the Wall Street Journal and Handelsblatt that a broader realignment of Europe's currencies would have been better than the lira adjustment alone.
- Lamont's reading. Close to a call for sterling to devalue; Schlesinger's post-Bath remarks had already set off the lira attack, and now he was targeting Britain.
- The phone calls. Lamont made Leigh-Pemberton ring Schlesinger at once, over his worry about interrupting the punctilious German's dinner. Schlesinger said he had given the interview on condition of checking his quotes and had not yet done so.
- The non-denial. Lamont protested that New York and Asia would trade on the wires overnight. The Bundesbank press office called the quotes "unauthorized" because unapproved, and Schlesinger would issue a statement from his office next morning. Mallaby's point: Germany's monetary master was in no hurry to adapt to 24-hour trading.
Lamont later wrote in his memoir that the idea of Britain being forced out the next day never crossed his mind.
Tuesday in New York: "Go for the jugular"
Druckenmiller read Schlesinger's comments on Tuesday afternoon and, caring nothing for whether they were authorized, acted at once. The Bundesbank was plainly happy to see sterling ejected and would not indulge weak neighbours with more cuts; given Britain's recession, devaluation was all but inevitable. He had held $1.5 billion short since August and added since the Johnson meeting. Now he told Soros he had his trigger and would build steadily.
Soros objected that steady building made no sense: if the story was accurate and the downside almost nil, why not go straight to $15 billion? "Go for the jugular," he said. Druckenmiller accepted that this was Soros's genius: he had done the analysis, read the politics and spotted the trigger, but Soros sensed the moment to go nuclear. When you knew you were right, Mallaby summarizes the creed, no bet was too big.
For the rest of Tuesday the two men sold sterling to anyone who would buy, working the phones themselves instead of leaving execution to traders.
- The overnight scramble. ERM rules obliged the Bank of England to buy sterling at the floor of DM 2.7780, but only during London trading hours. With the Bank closed, buyers were scarce, especially once word spread that Soros and Druckenmiller were selling wildly.
- The ripple. Each bank that took a giant Quantum order tipped its own currency traders, who sold too; as the calls spread around the world, everyone saw an avalanche beginning. The pound was soon knocked out of its band and buyers almost vanished.
- Bacon's call. Late in the day Louis Bacon rang Druckenmiller and mentioned he was still finding ways to dump sterling. Druckenmiller put him on hold and fetched Soros, who furiously demanded to know where Bacon had found the market.
The night shift
The bosses went home, leaving traders hunting for more buyers. Quantum's head trader beeped Johnson at home; he slipped out of bed to return the call quietly, because his wife was an official at the New York Fed. Around 2 a.m. Druckenmiller was back at his desk to meet the London open, when the Bank of England would again have to buy. Bessent, arriving soon after, found the boss in his darkened office against the night skyline, lit only by the telephone: Soros, on speakerphone, urging him to lever up and redouble the selling.
Black Wednesday in London
When London opened, the promise of Bank of England support lifted sterling back into its band, but flat on the floor. Following Lamont's plan, the Bank bought £300 million twice before 8:30 a.m., to no effect; Druckenmiller was selling by the billion across the Atlantic and a legion of imitators followed. By 8:40 it had bought £1 billion and sterling had not moved. Ten minutes later Lamont told Prime Minister John Major that intervention was failing and rates must rise.
- Major's refusal. Major had taken Britain into the ERM; if the policy visibly failed his credibility could collapse and a cabinet rival might challenge him. He cited economic data due later that day and told Lamont to hang on in hope that the markets would calm.
- The Fed calls. Central banks worldwide were on alert. The New York Fed called Johnson's flat again, and his wife spent the rest of the night monitoring a crisis her husband had helped cause.
- Liquidity for speculators. Obliged by ERM rules to keep buying, the Bank no longer hoped to lift the pound; it was simply providing liquidity to Druckenmiller and his imitators, loading up on a currency sure to be devalued. Mallaby calls it a vast transfer from taxpayers to traders. At 10:30 Lamont again urged Major to raise rates.
While Lamont pleaded, officials projected calm. Bank deputy governor Eddie George kept a long-scheduled meeting at Threadneedle Street with David Smick, a consultant who supplied political intelligence to Druckenmiller and Soros. In checked shirt and striped tie, George breezily said the Bank had everything under control and in the unlikely extreme would raise rates a full point. Smick asked whether the Bank had fallen too far behind the curve. Before George could answer, the phone rang; after a tense minute he said quietly that rates had just gone up 200 basis points, shook Smick's hand and left the room running.
The rise (taking the base rate from 10% to 12%) was announced at 11:00 a.m. Lamont watched the Reuters screen in his outer office: the sterling line stayed flat. He felt like a surgeon watching a heart monitor after the patient has died.
Hands in the blood
A negotiated realignment needed lengthy European coordination, and every minute cost taxpayers more. Italy had been lucky to crack on a Friday, just before a weekend's respite; Britain hit the cliff on a Wednesday. Lamont's only option was unilateral exit, which needed the prime minister. Major kept him waiting despite repeated calls; at Admiralty House, the Georgian building serving as Major's temporary residence, Lamont's team cooled their heels at least another quarter of an hour while, by his reckoning, the country lost hundreds of millions of pounds every few minutes.
- The meandering meeting. Major looked relaxed, wondered aloud about more diplomacy with Germany, and summoned other ministers to give their views: could Britain leave without offending its partners, would there be calls for resignations?
- The purpose. Major wanted to share responsibility. One minister said they were there to put our hands in the blood. Mallaby calls it shrewd: it neutralized potential rivals for Major's job. Meanwhile Druckenmiller and Soros kept adding to their positions.
- The last throw. Instead of exit, Major insisted on a further 3-point rise (to 15%), effective the next day. Again the Reuters screen did not move. In New York the hikes looked like a dying man's gestures, a signal for one final push.
Lamont warned his European counterparts. Italy's Piero Barucci suggested suspending the markets to buy time for a realignment; Lamont had to explain that no modern finance minister can close currency markets that trade continuously and globally. At 7:30 p.m., before cameras from around the world in the Treasury's central courtyard, he announced Britain's exit from the ERM. The markets had won.
- $27 billion. Reserves the Bank of England spent defending sterling in the first half of September, much of it on the final day.
- ~14% / ~$3.8 billion. Sterling's fall against the mark after exit, and the taxpayers' implied loss on the pounds the Bank had bought.
- $10 billion. Quantum's sterling short at the break, short of the $15 billion target but monumental.
- Over $1 billion. Soros Fund Management's profit on sterling.
- $300 million / $250 million / $800 million. Estimated takes of Bruce Kovner, Paul Tudor Jones, and the top seven US bank currency desks combined.
- $650 million. Soros's reported personal gain in 1992, surpassing Michael Milken's much-gawped-at $550 million; one magazine said he earned a median US family's annual income in five minutes.
The man who broke the Bank of England
The size of Soros's payout was unknown at first. In October the Italian industrialist Gianni Agnelli let slip that his Quantum investment would earn him more that year than Fiat. The next day, Saturday 24 October, the Daily Mail ran a photo of a smiling Soros, drink in hand, under a headline about making a billion as the pound crashed; reporters thronged his front door. Soros became the man who broke the Bank of England, and hedge funds began to replace the 1980s buyout kings as objects of popular envy.
Building out from the coup
The funds' real profits were larger than outsiders guessed. Just as Jones paired his 1987 equity short with a bond bet, Druckenmiller built outward from sterling:
- Gilts and British stocks. Most traders thought capital flight would hurt all British assets. Druckenmiller reasoned that exit would let Britain cut rates, lifting government bonds, and that a weaker pound plus lower rates would help equities. He bought both as he sold sterling on Tuesday and Wednesday; both rose steeply over the next two months.
- The French franc. When speculators turned on the franc, he judged the central bank would win: French households had no floating mortgages and the state had many ways to cushion its people, so temporary rate rises were bearable. He bought French bonds, which soared in 1993, part of why Quantum's 69% in the sterling year was followed by 63%.
- The Swedish krona. Quantum's biggest post-sterling coup was its quietest. Through Johnson, now full-time, it shorted the krona ahead of its November 1992 devaluation and made upward of $1 billion. Burned by the sterling publicity, Soros and Druckenmiller made sure nobody spoke of it.
Two lessons for economics
Mallaby draws two conclusions from the macro triumph.
- Efficiency fails when players do not maximize profit. The efficient-market hypothesis might hold in a market of rational profit-seekers, but governments and central banks have other agendas. At the climax, Major in effect bought sterling from Druckenmiller at a price both knew was absurd, for a reason found in no finance textbook: to make political rivals share the blame for devaluation.
- Pegs are fragile once capital is free. Fixed rates worked in the 1950s and 1960s because capital controls limited cross-border money. With controls gone, governments had to choose: use interest rates to steady the exchange rate or to steady the domestic economy, not both. "Flexible pegs" like the ERM tried to have it both ways and backfired. The floating dollar's failed August rescue caused no break; the pegs of Finland, Italy, Britain and Sweden offered irresistible targets and wrenching dislocations. Europe had made a promise it could not keep, bottling up currency pressure until a greater force blew the cork back in its face.
The episode is the textbook case of the impossible trinity: a fixed exchange rate, free capital flows and an independent monetary policy cannot coexist. For bank treasuries and FX risk teams it also shows why a currency peg is a short option written by the defending central bank: speculators risk about 0.5% to make 15-20%, and a defence resting on finite reserves invites a test. Stress scenarios for pegged currencies (Hong Kong, the Gulf) still model this payoff, and a small, failed intervention, like Britain's on 14 September, is read by the market as confirmation.
Black or White Wednesday
Policy makers did not absorb these lessons at once. As after every crisis, the first instinct was to vilify markets rather than accept that pegs were dangerous.
- Michel Sapin. The week after sterling's fall, with the franc under attack, the French finance minister suggested troublesome traders deserved the guillotine.
- Edouard Balladur. After further ERM turmoil the following summer (1993), the French premier called curbing speculators an economic and moral duty.
- Willy Claes. Belgium's foreign minister claimed Anglo-Saxon financiers were plotting to divide Europe.
Soros: speculation is dangerous. Characteristically, Soros accepted much of the attack. He had been more ruthless than anyone (bank desks, living with regulators, hesitated to assail governments), yet he saw markets as inherently unstable. Soon after the devaluation he told Jean-Claude Trichet, governor of the Banque de France, that he would not attack the franc for fear of destabilizing it; Mallaby finds the selflessness rich, since Quantum had already concluded the franc would hold and was set to profit from that. The greatest speculator would not defend speculation.
Druckenmiller: "White Wednesday." The British press said Black Wednesday; Druckenmiller thought White fitted better. Exit freed Britain from Bundesbank rates to fight its recession, and the FTSE rose almost a fifth in two months. Unification-driven German rates meant sterling needed to leave; the government failed to see it until a speculator saw it for them. If anyone fleeced taxpayers of $1 billion-plus, it was the prime minister, not the trader.
The academic tipping point
The Soros-Druckenmiller split foreshadowed a shift in economics. Before 1992 the consensus blamed currency crises on bad policy, not speculation. During the 1990s it moved from Druckenmiller's view to Soros's: traders might attack decently managed currencies, and the attacks could be self-fulfilling. Sterling's collapse was a tipping point. Druckenmiller was right that Britain had invited the crisis with untenably high rates, but his own success revealed speculators' power over governments and the temptation to abuse it: where traders once attacked only currencies doomed by fundamentals, they might now have a go at stable ones. French complaints about excessive hedge-fund power, Mallaby concedes, were not baseless.
Why little was done
- Capital controls rejected. Taming speculators meant taming the cross-border flows they surfed, a return to Bretton Woods. Most policy makers recoiled: if free trade in goods lets each country specialize (cars made where they are made best), free capital should steer scarce savings to their most productive uses.
- Controls leak. In the week after the sterling crisis Spain and Ireland restricted banks' currency trading to blunt attacks; the controls were quickly circumvented.
- Abolish the target. The remaining fix was to abolish national currencies. Soros told an interviewer that speculation could be very harmful and that a single European currency would put speculators like him out of business, a sacrifice he claimed he would welcome.
The attacks of 1992 at a glance
| Currency | Defence | Outcome |
|---|---|---|
| Finnish markka | ECU peg, outside ERM, no partner help | Peg dropped 8 Sep; fell nearly 15% |
| Swedish krona | Overnight rates 75% | Survived September; devalued Nov, Quantum made $1bn+ |
| Italian lira | Rates 15%; DM 24bn Bundesbank support | Broke band 11 Sep; devalued that weekend |
| British pound | $27bn reserves; rates to 12%, then 15% | Left ERM 16 Sep; fell ~14% vs mark |
| French franc | Fixed mortgages, state levers, rate hikes | Held; Quantum's long French bonds soared in 1993 |
| Spanish peseta, Irish punt | Curbs on banks' currency trading | Controls quickly circumvented |
Europe unified its currencies in 1999. But emerging economies in Asia and Latin America kept their pegs, creating the immense hedge-fund opportunities of the late 1990s.
- Inconsistent figure. Mallaby gives Quantum's 1990 return as 29.6% in the returns series and 29.8% for the mark-trade year.
- The cost to Britain. The $3.8 billion taxpayer loss is a rough estimate; HM Treasury papers released in 2005 put the cost at about £3.3 billion.
- Rate levels and Sweden. Mallaby gives the size of the hikes, not the levels (10% to 12%, then a 15% announced but never applied, as the exit came first). He also omits that Sweden's Riksbank pushed its marginal rate to 500% on 16 September, yet still floated the krona in November, which strengthens his point about pegs.
- Sources and theory. The narrative leans on participants' recollections (Johnson, Bessent, Smick, Lamont's memoir). The academic shift he describes is the move from Paul Krugman's "first-generation" crisis model (1979) to Maurice Obstfeld's "second-generation" self-fulfilling models of the late 1980s and 1990s.
Scott Bessent, the young Quantum manager who spotted the floating-mortgage mechanism, was confirmed as US Treasury Secretary in January 2025.
Hurricane Greenspan
In the early 1990s hedge funds turned themselves into shadow banks, borrowing short from brokers and lending long by buying bonds, often on as little as $1 of capital per $100 of Treasuries. When Alan Greenspan raised rates by a quarter point on 4 February 1994, leveraged selling turned a gentle tightening into the steepest bond rout in more than a decade, sinking Michael Steinhardt and David Askin and linking yen trade talks to Mexican assassinations. Mallaby's central lesson is that leverage makes overreaction mandatory, and that regulators, warned in 1994, chose to look away, so that the failure of Long-Term Capital Management four years later made 1994 look trivial.
Steinhardt on the beach
In December 1993 Michael Steinhardt was at his vacation home in Anguilla when his staff called to say his funds had made more than $100 million in a single day. He marvelled at earning so much while sitting on the sand; a lieutenant told him this was simply how things were meant to be. Steinhardt Partners was running about $4.5 billion, employed more than a hundred people, had returned 47% in 1991 and 48% in 1992 after fees, and was heading for a similar year in 1993. Magazines reported his purchase of a Picasso drawing for nearly $1 million; at New York parties the portly, mustachioed investor was mobbed by people begging to give him money. He had no inkling that humiliation was weeks away.
How a stock picker became a shadow bank
After his sabbatical Steinhardt had set up his own firm in 1979 and resumed his blend of stock picking and block trading. But he also began making money in bonds, and by the early 1990s he was pioneering an early form of what later became known as the shadow banking system. As with Druckenmiller's currency trades, the strategy exploited a central bank's policy, this time the Federal Reserve's.
- The setting. In 1990-91 the US was in recession after the savings-and-loan crisis, and the Fed held short-term rates low to stimulate it.
- The trade. Steinhardt borrowed short-term money very cheaply and loaded up on longer-term bonds yielding considerably more, pocketing the spread.
- The risk. If long rates spiked, his bonds would crater; but a sluggish economy kept the demand for capital low, so its price (the interest rate) was unlikely to rise.
- The bonus. Long rates in fact fell in the early 1990s, adding capital gains on top of the carry.
Mallaby draws the parallel with sterling: just as John Major enriched Druckenmiller while trying to neutralise political rivals, the Fed rewarded Steinhardt while trying to rescue the banks. Widening the gap between short and long rates made it more lucrative for banks to do their normal business, borrowing short and lending long, and so sharpened their incentive to push money into the economy. Steinhardt crashed this party: he skipped the tellers who gather deposits and the credit officers who lend them, borrowing instead from brokerages such as Goldman Sachs and Salomon Brothers and "lending" by buying bonds. With no branch infrastructure, he could charge in and out of the banks' business as Fed policy shifted.
Regulated bank. Funds itself with customer deposits; a capital adequacy rule makes it hold about $10 of capital per $100 of loans so depositors are repaid even if loans sour; carries the costs of tellers, branches and credit officers.
Steinhardt's hedge fund. Funds itself with broker credit; has no depositors and therefore no capital rule; can borrow whatever brokers will lend, which often meant only $1 of capital per $100 of US Treasuries (100-to-1 leverage); can enter and exit the business at will.
The early-1990s boom
In 1993 Steinhardt and fellow speculators took the trade to Europe. With the exchange-rate-mechanism crises over, the continent was heading for monetary union, which forced interest rates to converge. Spain and Italy, with weak inflation records and hence high rates, now faced strict inflation-fighting rules, and their rates began falling toward Germany's. Traders bought truckloads of Spanish and Italian bonds and reaped capital gains as yields dropped. Again, Mallaby notes, governments had practically invited the bet: European statesmen made no secret of their plans or of the convergence they implied.
Because hedge funds were built to seize such openings in a way most rivals were not, they profited hugely. Senior investment bankers quit Goldman and Salomon to join in; one Wall Street law firm claimed to be launching new partnerships at two a month. The Helmsley Building at the foot of Park Avenue, known for its Art Deco lobby and brass elevators, became notorious as a "hedge-fund hotel", its unmarked upper suites full of small trading bands investing every which way for rich clients. The number of hedge funds leapt from a bit over 1,000 in 1992 to perhaps 3,000 in 1993, and fees rose almost as fast.
- A. W. Jones (first generation). No management fee; a 20% share of investment profits.
- Second generation (Steinhardt and peers). A 1% management fee plus the 20% profit share.
- Early-1990s hot funds. "2 and 20": 2% of assets plus 20% of profits. Forbes wondered whether so few had ever made so much so fast.
Cornering the Treasury auction
The fast money brought controversy. Steinhardt's shadow banking, mirrored by most big funds, led him to buy vast amounts of newly issued government bonds, at one point amounting to an effective takeover of the market.
At the April 1991 Treasury auction, Steinhardt and Bruce Kovner between them bid for $6.5 billion of the $12 billion of paper on offer. They then lent the bonds to short sellers and bought them back, ending with $16 billion of bonds, well over 100% of the issue. As prices rose the shorts tried to cover but could not, because the pair had cornered the market and were not selling. The victims, Goldman Sachs, Salomon Brothers and Bear Stearns, were hardly innocents, but someone sued. After three years in court the two settled without admitting guilt: Steinhardt paid $40 million to the short sellers and Kovner $36 million. They were also sued over irregularities in the May auction and again paid settlements.
The quieter danger: monetary transmission
Mallaby argues that manipulation should not have been the only worry. Largely unnoticed, hedge funds were changing monetary policy. By aggressively buying longer Treasuries whenever the Fed held short rates low, they tightened the link between short and long rates: a Fed cut that once took weeks to reach long rates now did so far faster. That helped when the Fed wanted stimulus. But the real economy lags the bond market, so if the Fed held rates down until Main Street felt the benefit, Wall Street was likely to inflate a giant bond bubble.
Such bubbles were newly frightening because they rested on breathtaking leverage. The system was pyramiding debt upon debt: the government borrowed from hedge funds, which borrowed from brokers, which borrowed from someone else. If one link failed, the others could lose their funding and be forced to dump assets fast; a bubble could burst instantly. Nobody had this in mind at the start of 1994, least of all Steinhardt. In January he let a lieutenant buy a staggering position in Canadian bonds, on top of bets in the US, Japan and Europe, and then left with his wife and friends for a holiday in China.
A quarter point that cracked the bond market
On 21 January 1994 Alan Greenspan went to the White House. Fuelled by low rates, the economy had grown smoothly for 34 consecutive months; inflation was quiet, but Greenspan wanted a small hike now to pre-empt it, avoiding a later need to slam on the brakes and steering for a "soft landing." Vice President Al Gore objected. In 1988-89 the Fed's short rate had climbed from 6.5% to nearly 10% in a dozen small steps; if markets expected a repeat, long rates could shoot up in anticipation, the bond market would crash, and the hard landing Greenspan wanted to avoid might arrive.
Greenspan, a former New York economic consultant who understood markets better than most Washington figures (and, Mallaby suggests, better than the British officials who lost the sterling battle), relished the question. He had considered a Wall Street backlash, but he was focused on stocks: the S&P 500 had hit records in 1993 and looked ripe for a correction. Bonds seemed secondary. He told Gore that long rates were governed mainly by inflation expectations: a hike would signal vigilance, lower those expectations and so lower long rates. The move should be bullish for bonds.
On 4 February 1994 the Fed's rate committee raised the rate by 25 basis points, from 3% to 3.25%. Greenspan warned colleagues that a 50-point move carried a very high probability of cracking these markets, a remark that later looked a touch too confident. On stocks he was right: the S&P dipped gently, as intended. On bonds he was wrong: the ten-year Treasury yield rose by the same 25 basis points.
Greenspan's model assumed bond investors who price inflation. Leveraged shadow banks price risk: the first hike in half a decade created uncertainty, and because even a small fall could wipe out their thin capital, the mere possibility forced them to sell, pushing long rates up, the opposite of what the Fed expected. The logic of leverage changed the central-banking game.
The contagion of February and March
A week later, on Friday 11 February, US-Japan trade talks collapsed and the Clinton administration signalled that a stronger yen would suit it. Within a week the yen rose 7% against the dollar. Stan Druckenmiller had an $8 billion bet against the yen, almost as large as his sterling wager, and lost $650 million in two days. Traditional central-banking logic said this should not touch inflation expectations or bonds, but funds forced to raise capital dumped assets, including bonds, and over the next two weeks the ten-year Treasury yield rose more than a quarter of a percentage point. A world in which hedge funds traded everything, Mallaby observes, was a world of unpredictable connections.
Europe was next. In February the central banks of Germany, Britain, France and Belgium had pushed short rates down, signalling no inflation risk. Yet after the yen shock long rates spiked: within a fortnight the ten-year yield rose 37 basis points in Germany, 58 in Italy and 62 in Spain. Hedge funds and banks' proprietary trading desks, having lost on Treasuries and the yen, were dumping European bonds regardless of fundamentals.
Once the flight began it fed on itself. Brokers that had lent freely reversed: instead of $1 million of margin per $100 million of bonds, they demanded $3 million or $5 million. At 100-to-1, an extra $4 million margin call forces the sale of $400 million of bonds, quickly; those sales push down remaining holdings and trigger further calls. The pyramid of debt that inflated the bubble now accelerated its implosion.
On 1 March new data suggested US inflation was a bigger threat than feared, and the ten-year Treasury yield jumped 15 basis points, a reaction Greenspan's logic could explain. What followed it could not. Japan, grappling with deflation, saw its ten-year rate jump 17 basis points on 2 March; as brokers issued more margin calls, funds dumped an estimated $60 billion of European bonds and long rates there spiked too.
| Date (1994) | Shock | Leveraged transmission |
|---|---|---|
| 4 Feb | Fed hikes 25bp to 3.25% | US ten-year yield up 25bp instead of falling |
| 11 Feb | US-Japan trade talks collapse; yen up 7% in a week | Druckenmiller loses $650m in two days on an $8bn yen short; funds sell bonds, ten-year up 25bp+ in two weeks |
| Late Feb | European central banks cut short rates | Yields still rise in a fortnight: Germany 37bp, Italy 58bp, Spain 62bp |
| 1-2 Mar | US inflation data; ten-year up 15bp | Japanese ten-year up 17bp despite deflation; about $60bn of European bonds dumped |
| 23 Mar | Mexican presidential candidate assassinated | Selling spreads from Mexican to all emerging, then rich-world bonds |
| 28-31 Mar | Askin margin calls; brokers seize collateral | Brokers hedge seized mortgage bonds by dumping Treasuries |
The casualty list
- Commodities Corporation trio. Paul Tudor Jones, whose great skill was sensing how others were positioned, missed the danger in Europe and was down sharply that spring; so were Bruce Kovner and Louis Bacon.
- David Gerstenhaber. The macro trader Julian Robertson hired in 1991, now running his own hot fund, blew up spectacularly.
- Proprietary desks. Goldman Sachs had its worst year in a decade in 1994.
- Insurers. They were reckoned to have lost as much on bonds as they paid out after Hurricane Andrew, prompting one analyst's quip: "Hurricane Greenspan."
- Bankers Trust. For a few hours on 2 March it teetered on the brink of bankruptcy; trading was suspended on the New York Stock Exchange, and New York Fed president William McDonough phoned top bankers to rally confidence. It survived, pushed to the edge but not over, this time.
Steinhardt drowns
The greatest casualty was Steinhardt. Back from China, he found the US bond market had sold off and he had suffered modest losses; his biggest bets were in Japan, Canada and especially Europe, where he held an astonishing $30 billion bond portfolio, and he assumed the Fed's move would not hit foreign markets hard. Then deleveraging whipsawed every market. Each basis-point rise in European yields cost him $10 million; by the end of February he had lost $900 million, almost 20%, and the worst was still ahead.
His deeper error was liquidity. The supposed beauty of macro was that bond and currency markets could absorb far more capital than individual stocks. In the boom, hedge funds and other foreigners had bought about half of Germany's government and government-guaranteed bonds without moving prices, because they built positions gradually and could wait days if a bond was hard to find. But under margin calls they all had to sell at once, to each other. The liquidity vanished; nobody was buying.
In equity block trading Steinhardt had known his brokers personally, and as a major client he could rely on them in a crisis. In Europe the brokers were anonymous voices in another time zone; they did not know him and would not help him out. At his bow-shaped desk facing seven screens, he felt as though he were drowning; he ate little, slept less and, known for explosive rages, now whispered with lieutenants behind closed doors. His traders kept hoping for a better moment to sell while prices fell like stones. Veteran lieutenant John Lattanzio strode to the bond desk and barked at them to just sell; the trader replied flatly that he could not.
Steinhardt resolved to dump bonds at any price, conceding the sort of discounts he once extracted from block sellers of stocks. A four-day burst in early March offloaded $1 billion of European bonds, while the struggle to exit Japan and Canada went on; the Canadian central bank kept phoning nervously to ask whether his traders were done.
Steinhardt fell into the trap that would define leveraged markets: he failed to see that the trade had become crowded. When a broker's margin call can force fire sales, the danger lies wherever leveraged players are concentrated. By the end of March his funds were down 30%, with about $1.3 billion of capital vaporised; he later admitted the crowding had totally passed him by.
Askin Capital: the fraud inside the rout
Steinhardt's humiliation was quickly followed by a second collapse. David Askin launched Askin Capital Management in the glory days of 1993 with a $2.5 billion portfolio of mortgage securities. He had little experience of actually managing money; he was an analyst and salesman, and his pitch was that, as former head of mortgage research at Drexel Burnham Lambert, he could profit from mortgages in any climate.
His claimed edge was forecasting prepayment: a mortgage likely to be paid off early is worth less, because investors lose the income stream they counted on. Wall Street was meanwhile slicing mortgages into exotic instruments.
- Interest-only strips (IOs). Receive the interest payments; they lose when homeowners prepay fast, because the interest stream stops.
- Principal-only strips (POs). Receive the principal; they gain from fast prepayment, since the money arrives sooner.
- Slow prepayment. Reverses both: IOs gain, POs lose.
- Askin's menagerie. He traded IOs and POs, inverse IOs and inverse POs, and even "forward inverse IOs"; for anyone who truly nailed prepayment, the opportunities looked endless.
The pitch won over respected names, including the insurer AIG and the Rockefeller Foundation. But, not for the last time in hedge-fund history, it was fraudulent. Askin called his funds "market neutral", profitable whether bonds rose or fell, but had little idea how his positions would behave in a crisis. He said a sophisticated proprietary prepayment model lived on his own computer; no such model existed.
When the Fed tightened, his funds fell. Higher long rates raised mortgage costs and removed the incentive to refinance, so prepayments slowed and durations stretched out like chewing gum. For any given rise in long rates, his portfolio fell five times as much as an ordinary bond. Askin first hid this by misreporting results; a revolt inside his firm forced him to confess on Friday 25 March that the funds had lost 20% in February. Two days earlier, on 23 March, the assassination of a Mexican presidential candidate had renewed the panic: investors dumped Mexican bonds, then all emerging-market bonds, then rich-world bonds as they rebalanced. The new jump in rates guaranteed that Askin's March numbers would be awful.
- Monday 28 March. Margin calls begin: Bear Stearns wants $20 million, Kidder Peabody $41 million; by afternoon Bear raises its demand to $50 million.
- Tuesday 29 March. Salomon Brothers, Lehman Brothers and others pile on. A Kidder team of mortgage traders arrives uninvited and works through the night to price a chunk of the funds, but the holdings have been pledged as collateral to dozens of creditors whose claims cannot easily be unscrambled, so no clean bid is possible.
- Wednesday 30 March. Brokers issue ever larger calls, each racing to seize cash before rivals. Swiss manager Unigestion, owner of one sub-fund, orders liquidation; Askin canvasses 18 dealers, and most bids are near worthless. Two positions in forward inverse IOs draw no bids at all: nobody can value them.
- 3:45 pm that day. Bear Stearns offers a paltry $5 million for a chunk of the portfolio and demands an answer in 15 minutes; Askin asks for 15 more and refuses.
- Aftermath. Brokers seize their collateral and hedge it by dumping Treasuries, rocking the bond market again. Kidder, Bear and DLJ lose about $500 million between them.
Mallaby pulls two lessons from Askin that would haunt later decades. First, leverage that makes a fund crash also complicates its burial, because pledged collateral cannot be sold cleanly. Second, complexity makes assets unsellable in a panic, a premonition of 2007-08.
Clinton's plea and the new bond market
The panic now reached the White House. On the morning of 31 March, as Askin's brokers hammered Treasuries, Bill Clinton interrupted his California vacation. His advisers had taught him to revere the bond market: in 1993 he had raised taxes on the theory that a smaller deficit would reduce government demand for capital and lower bond rates. He disliked the argument at first, cursing the undue power of bond traders, but his advisers prevailed, and when rates duly fell in the 1993 bond bubble they seemed vindicated. Now rates were soaring and Clintonomics looked less brilliant.
He phoned Robert Rubin, the ex-Goldman chief and White House top economic adviser who more than anyone had promised that the tax rise would be rewarded. For about half an hour Rubin struggled to explain. Growth and inflation could not account for it, he admitted; Wall Street had changed faster than anyone had noticed. The Fed had raised rates only twice, 25 basis points in early February and 25 again in March, yet bonds were suffering their steepest decline in more than a decade. Neither Greenspan, nor Steinhardt, nor Rubin understood it. Clinton then told reporters the underlying economy was sound and there was no reason for people to overreact.
Mallaby treats Clinton's plea as the template for a generation of official responses to panic, and as a misreading of the hedge-fund challenge. Leverage by definition makes investors feel economic ripples in magnified form and keeps their fingers on a hair trigger. In a leveraged world, overreaction is not a choice: it is mandatory.
Rubin was right that Wall Street had been transformed. Per Securities Data Company, new public issues of non-Treasury bonds and notes grew thirteenfold, from $96 billion in 1981 to $1.27 trillion in 1993. The market was also growing more complex: Askin exaggerated his analytics, but other firms were hiring physicists with supercomputers to design ever more fanciful securities. Just as deep currency markets had destroyed central banks' ability to intervene, as Britain learned in 1992, deep bond markets had weakened governments' ability to anticipate long rates, let alone control them.
The responses that might have been
Track one: monetary policy
The Fed knew it had been surprised. On a 28 February conference call of the rate committee, Greenspan admitted the bond reaction had been a shock. Alan Blinder, the Princeton professor who joined the Fed's leadership in June, complained about the outsized power of 27-year-olds in yellow suspenders. The question was what to conclude.
The Fed could have redefined its mandate. It targeted consumer prices, but 1994 was about asset prices. A bond market at record highs, as in 1993, might signal that credit was too cheap and that rates should rise to deflate a bubble. Because the Fed watched inflation rather than bonds, it let the bubble grow; then came the blow-ups, more than $600 billion knocked off US securities and about $900 billion of wealth destroyed in foreign bond markets. An earlier hike might have reduced the carnage.
Greenspan's eventual answer (over the next decade). No. Bubbles are hard to identify until they pop; it is easier to clean up afterward than to prick them pre-emptively; and dampening asset prices would require rate hikes so aggressive that lost growth would exceed the benefit of calmer markets.
Mallaby's verdict. For several years Greenspan's judgment looked right, but after the mid-2000s credit bubble the clean-up approach proved disastrously costly. The case for weighing asset bubbles in monetary policy, and for making financiers restrain leverage when markets look frothy, was belatedly vindicated.
Track two: regulation
The second road not taken was regulatory. Storied investment banks were entrusting billions to cowboys, inside hedge funds and on their own proprietary desks. If an obscure, relatively small fund like Askin could leave its three main brokers (Kidder Peabody, Bear Stearns and Donaldson, Lufkin & Jenrette) with $500 million of losses, what might a big one do?
Plenty of officials asked. On 7-8 March hedge funds were the main topic at a gathering of top central bankers in Basel. In Washington the President's Working Group on Financial Markets began studying hedge funds and leverage. Representative Henry González, the banking committee chairman, growled that hedge funds deserved extra scrutiny. Business Week called them "rogue elephants", overleveraged, undersupervised and disruptive, citing a commentary titled "The Hedge Fund Crisis." González announced hearings, which the Washington Post predicted would light a fire under regulators.
Unseen, a counterattack was forming. Robert Johnson, the Soros economist who had helped shape the sterling coup, had once worked on the Senate banking committee, and his research assistant from that time was now González's staff director. Johnson reasoned that lawmakers were bad-mouthing an industry they did not understand; the way to cool Washington was to call their bluff and make them say concretely what they would do. He phoned his former researcher and urged him to hold the hearings. He then worked on Soros, warning that without an effort to educate lawmakers a backlash was certain: ordinary Americans pictured a jack-in-the-box that would leap from the ceiling at night and eat their net worth, and its name was George Soros. The industry had to be demystified; after all, Warren Buffett was a folk hero.
Panel one: the regulators. With no concrete plan to stop imploding funds from hurting creditors, they fell back on asserting that no action was needed. Comptroller of the Currency Eugene Ludwig said only eight national banks lent to hedge funds, mostly against collateral, with his examiners on site full time at each. Fed governor John LaWare said supervisors were not losing sleep. When Representative Melvin Watt asked what would happen if a big fund blew up, the panel assured him banks and brokers would be fine: they would not be foolish enough to over-lend.
Panel two: Soros alone. Johnson's bet paid off: preparing questions had taught the staff they had no firm ground for an attack, and the tone was warm. Tom Lantos introduced his fellow Hungarian American, and González, pleased with himself, practised pronouncing the name as "Shurush." Soros lectured on reflexivity, plugged his book, and explained that leverage could destabilise markets but that any limits should apply to all financial actors, starting with the big brokerages. Askin's fund, he said, had as little in common with his as hedgehogs with hedge cutters.
Mallaby stresses what was lost. Had Soros's advice been taken, system-wide leverage might have been reined in and the next decade's financial history might have been different; but lawmakers, fixated on hedge funds as a novel threat, ignored it, and some were still struggling to grasp what a hedge fund was. Future Treasury secretaries Robert Rubin and Lawrence Summers joined the debates but pushed nothing. McDonough, who had manned the phones for Bankers Trust, did nothing afterward to curb the leverage behind the panic. The Askin postmortem showed brokers had lent carelessly, contrary to the assurances given to Watt, yet little was done to make banks cautious. Mallaby calls this perhaps the clearest policy error specific to hedge funds, and its cost came due when Long-Term Capital Management failed four years later.
Askin and Steinhardt are templates for the risks prime-brokerage and counterparty-credit teams now manage: procyclical margins, crowded trades, and collateral pledged across many lenders. Post-2008 rules (the Basel III leverage ratio, haircut floors on securities financing) answer Soros's 1994 point by limiting leverage at the lenders. Archegos (2021), which cost banks over $10 billion, showed the lesson still needed relearning.
Steinhardt's exit
For Steinhardt's funds, 1994 was terminal. He had flirted with retirement for years, telling Institutional Investor in 1987 that making wealthy people wealthier did not touch his soul. After the bond rout he resolved to leave. Staff saw that his nerve was shattered: he feared risk, and his temper went from bad to horrible. He would fire someone at lunch and invite him back the next day, or scream with the intercom on so his curses carried through the office.
He hung on through 1995, returning a surprisingly good 27% and recouping $700 million, then, honour restored, announced his long-rumoured retirement. It was a watershed: after 27 years, one of the Big Three was leaving. A dollar invested with him in 1967 was worth $480 at the close, 26 times the $18 it would have become in the S&P 500. The 1994 debacle cost his funds $1.5 billion, but he had earned more than twice that in 1991-93, and his record otherwise was excellent. He reflected that he had made his investors and himself more money than he had ever imagined as a child; retirement meant more time to dance with Martha, the blue crane on his country estate that courted him with a graceful gavotte.
Two verdicts on hedge funds
Clients were harsher than regulators. By September 1994 investors had withdrawn about $900 million, with more leaving. Financial magazines noted indignantly that hedge funds were not hedged; Forbes declared, not for the first time, that the hedge fund party was over. Macro managers, who had badly overestimated the liquidity of currencies and bonds, handed money back unasked: Paul Tudor Jones returned a third of his capital and Bruce Kovner, in June 1995, two thirds, both citing the difficulty of moving in and out of markets with too much money. A month later Soros told Quantum's investors that size explained recent disappointments. Macro was now derided as "leveraged directional speculating."
Yet unlike the early-1970s backlash that buried hedge funds for a generation, the industry proved resilient. Apart from Steinhardt, the famous funds shrank but survived, and an ecosystem of smaller players had evolved styles that held up in a crisis: merger bets, lending to bankrupt companies, and computer models arbitraging gaps between similar instruments. One survey put the average hedge fund's loss in the first quarter of 1994 at only 2.2%, against 3.3% for the average equity mutual fund; another found hedge funds had beaten the S&P 500 over five years with considerably less volatility. University endowments and other sophisticated investors realised that hedge-fund returns could diversify stock and bond risk.
So 1994 produced two verdicts. Regulators confronted worrying questions but, lacking a theory of how to tame the industry, looked the other way; institutional investors decided that hedge funds' risk-adjusted returns were irresistible. Mallaby's point is that the two are one: because markets are not perfectly efficient, hedge funds are part of an unstable game that can wreak havoc, and the same inefficiency is what lets them prosper, so investors would line up to get in.
Several slips and gaps. Mallaby calls González chairman of the House Financial Services Committee, but in 1994 it was the House Banking Committee (renamed in 2001). He later calls McDonough New York Fed "governor"; he was its president, as Mallaby says earlier. Blinder became the Fed's vice chairman in June 1994. Steinhardt's 1994 loss is given as $1.3 billion by March and $1.5 billion for the whole debacle without reconciling the two. Finally, Mallaby leans hard on leverage as the cause; many economists stress that markets were simply repricing a long tightening cycle (the Fed took its rate from 3% to 6% within a year), with leverage amplifying rather than creating the move.
In December 2021 Steinhardt surrendered 180 looted antiquities worth about $70 million to the Manhattan district attorney and accepted a lifetime ban on acquiring antiquities, avoiding prosecution.
Soros versus Soros
In the emerging-market crises of 1997-98, the Soros funds made about $750 million shorting the Thai baht, then lost about $800 million in Indonesia, missed a fortune in South Korea, and lost $1-2 billion in Russia. Mallaby's explanation is a split personality: Soros the speculator kept colliding with Soros the statesman-philanthropist, who wanted to save countries rather than break them. The episode demolishes the image of hedge funds as all-powerful superpredators: in Asia and Russia they were often cautious, frequently wrong and hemmed in by political risk, while the deeper causes lay in pegged currencies, fragile banks and hot foreign capital.
Palladium in the Arctic
Stalin built the mining city of Noril'sk with slave labour: his secret police arrested hundreds of engineers on invented charges and shipped them to a treeless corner of the Siberian Arctic, where perhaps 200,000 political prisoners (zeks) died building the city and mine, their bones still washing out with each spring thaw. By the mid-1990s about 260,000 people lived there amid temperatures of -40°C, sulphurous smelter smog and five months of polar night. The mine held more than half the world's palladium.
In 1990 Paul Tudor Jones and a few fellow hedge funders vacationed at a salmon camp on the nearby Ponoi River and, as the rich sometimes do, decided to buy it. Negotiating fishing rights at a meeting in Helsinki, they heard a Russian official lament that the creaking Soviet infrastructure at Noril'sk was close to collapse. Thorpe McKenzie, Julian Robertson's former Tiger partner who was negotiating for the anglers, took note: a breakdown would have global consequences.
McKenzie returned annually, hosting Russian officials at the camp and pumping them for information, and studied the market.
- Dentistry. Roughly stable demand.
- Catalytic converters. Growing demand, driven by environmental regulation of cars.
- Cell phones. A new market with promise.
- Supply. Apart from Noril'sk, the main producers were in Africa, with their own infrastructure problems.
He concluded that demand would outgrow uncertain supply, bought some metal himself and tipped off Robertson. By 1994 Tiger owned $40 million of palladium and sent a young commodities specialist, Dwight Anderson, to investigate.
Anderson flew to Moscow, then took an ancient Aeroflot plane north, landing in November half-light beside wrecked aircraft cannibalised for parts. The next day he found the mine working surprisingly well. Unlike cramped Appalachian coal mines, Western open pits or deep South African shafts, Noril'sk was Soviet in scale: you drove underground in a vast truck through a warren of caverns far bigger than needed. That excess was a blessing: if one tunnel or trolley line failed there were alternatives, and every broken drill had two defunct ones nearby to strip for spares. Anderson concluded the collapse stories had been deliberately exaggerated by local officials, who all had stakes in sales and knew that rumours of breakdown were bullish for prices.
Something else was going to collapse, though. Anderson confirmed that Russia was selling more palladium than it mined, ripping it out of disused Soviet military equipment. Around 2000, he calculated, the missiles to scrap would run out and prices would skyrocket. So Tiger held on even though the original logic had changed; the position lost money for three years, before things got interesting.
Hot money and a fund too big
Tiger's Siberian foray was part of a wider opening. Before 1990 Westerners were barred from the Kola Peninsula, and East Asia and Eastern Europe let only a trickle of foreign money into small equity and bond markets. Around the turn of the decade emerging markets came into their own: developing-country bonds outstanding jumped from about $40 billion to $100 billion in a year, and growth continued all decade. Traditional managers warmed slowly; hedge funds moved fast, buying debt such as Peru's. Soros and Druckenmiller launched Quantum Emerging Growth in 1992; Louis Bacon's Moore Capital followed in 1993.
The benefits. They supplied capital when others would not; other Western managers eventually followed their lead; and because foreigners chose companies shrewdly, they often raised the quality of capital as well as its quantity.
The risks. They could juice growth for a few years and then yank money out and crater it. Britain in 1992 and Greenspan in 1994 had learned that deep, fast capital markets could force wrenching moves in currencies and rates; frail emerging economies were more exposed still. Mexico's currency crisis at the end of 1994 was a dress rehearsal for a far bigger drama, much of it conceived at Soros Fund Management.
The size quarrel
After sterling, Stan Druckenmiller kept Quantum winning: +63% in 1993, the golden bond year; +4% in the 1994 crash; +39% in 1995. But he began a long argument with Soros over fund size. In 1992 they had judged they should sell $15 billion of sterling but managed only $10 billion; a fund a third smaller would have aimed at $10 billion, sold it all, and made a percentage gain a third higher. For Druckenmiller, whose pay and ego rode on the performance number, that settled it.
For Soros the sums differed. His ego was invested as much in the size of his empire as in returns: a global firm made him a global player and amplified his writings and philanthropy. He was known as the only private citizen with his own foreign policy and enjoyed statesmen beating a path to his door. Offered a word with visiting Henry Kissinger, Druckenmiller asked whether he knew anything; Soros replied that he did not like Kissinger but the man did know things, vaunting the connection while seeming unimpressed.
The compromise: the firm kept gathering assets, but not all into Quantum. By 1996 just over half of Soros Fund Management's $10 billion sat in three other funds.
- Quantum Emerging Growth. The emerging-market vehicle.
- Quasar. Invested with outside hedge-fund managers.
- Quota. Run by Nick Roditi, a secretive London macro trader.
To staff them Soros hired, among others, Princeton-trained Brazilian economist Arminio Fraga, just out as a deputy governor of Brazil's central bank; within days of meeting him in early 1993 Soros offered a partnership and then simply informed Druckenmiller. Druckenmiller, who might have wondered how he was meant to run a place where he could not choose his staff, said only that they would see if Fraga was any good.
Thailand: signal or sledgehammer
For four years Fraga performed the benign function of hedge funds, financing economies others shunned: bonds of Brazil and Venezuela, exotic Moroccan loans, absurdly cheap Brazilian utility shares. In late 1996 he heard IMF number two Stanley Fischer speak. The mood was upbeat, but asked who might be the next Mexico, Fischer said he saw none for now, though imbalances in Asia might repay a look. That lit a spark. Weeks later Fraga read a joint IMF-Federal Reserve paper, "The Twin Crises," on how a currency collapse and a banking collapse feed each other, and asked Druckenmiller if he could go and look at Asia. Druckenmiller told him to go.
Fraga's diagnosis
Landing in January 1997, Fraga found Thailand fitted the double-crisis model exactly.
- Lost competitiveness. China's rise as a low-cost rival had hammered Thai exports, and the baht was tied to a currency basket dominated by a strong dollar.
- Refusal to adjust. Instead of letting the baht fall, Thailand ran a big trade deficit, consuming more than it produced and borrowing from foreigners to pay.
- Crumbling finance. A major Thai bank collapsed in 1996; the central bank cut rates to protect banks, which lowered returns to foreign creditors and gave them another reason to leave.
- The toxic link. Much foreign lending was in dollars but funded property and baht-earning projects; a devaluation would make those debts unpayable and tip fragile banks into ruin.
Fraga, Soros's Asian equities expert David Kowitz and Hong Kong economist Rodney Jones met a senior Bank of Thailand official. Drawing on his Brazilian central-bank experience, the mild-mannered Fraga, who seemed more an emerging-market peer than a Wall Street predator, laid out the dilemma: defending the peg meant high rates, while a trade deficit and wobbly banks argued for devaluation and lower rates. The official answered honestly and naively that Thailand had so far accepted whatever rates the band required, but with the banks in trouble, getting rates down might now matter more than defending the currency. In effect he had conceded the peg was unsustainable, handing over a metaphorical suitcase of cash. To avoid alerting him (he could raise rates or crack down on speculators), Fraga nodded and moved on; a little later Kowitz, feigning that he was out of his depth, asked him to repeat the point, and he did.
After a stop in South Korea, Fraga briefed Druckenmiller, who approved at once. Over a few days in late January the team sold short about $2 billion of baht. The selling was both forecast and trigger: Thailand spent reserves and raised rates by 3 percentage points, punishing for tottering banks. But the Soros team had borrowed baht for six months at the old low rates and could wait until end-July.
Why not go for the jugular?
Mallaby inverts the usual question. In 1992 the team sold $10 billion of sterling, about 2.5 times capital; the baht trade was only a fifth of capital. Repeating the sterling leverage would have meant selling roughly $23 billion more, wiping out most of Thailand's reserves and probably forcing capitulation within days. A quick, magnificent profit was there for the taking. Three reasons held them back.
- Soros's ambivalence. He liked to say markets could swing not like a pendulum but like a wrecking ball, laying waste to economies.
- Borrowed conviction. Druckenmiller was following Fraga's idea, so lacked the intensity he felt when a trade was wholly his.
- Moral qualms below. Rodney Jones challenged Fraga and Kowitz on the ethics of speculating against poor countries, where crashes push millions into poverty. The Soros who had once urged going for the jugular now had a squeamish team.
Jones settled on a justification that Soros later repeated in a book: speculation helps poor societies if it is a signal, not a sledgehammer. The virtuous speculator alerts a government that it must change course (here, devalue); the longer it procrastinates, the more reserves it burns and the more brutal the eventual collapse. Mallaby accepts the logic but notes it justifies only mild speculation; an all-out attack would precipitate a crisis rather than prevent one.
Thailand digs in
Thailand exposed the flaw: signals help only if governments heed them. Instead the government threw away reserves, buying baht from Druckenmiller at a rate certain to lose once devaluation came. When Jones returned in late April, he counted 100 cranes on Bangkok's skyline with almost none working; property firms had stopped paying; 87 of 90 finance companies were said to be insolvent. Thailand's late-1970s crisis had hit when loans were 40% of GDP; now, thanks to global capital flows, they were 140%, so a banking bust would hurt the real economy far more. The one hope was that officials still had time to devalue before panic forced them.
Jones's visit itself helped precipitate the crisis: the Thai press reported it, local financiers and companies dumped baht for dollars, and the central bank intervened again. On the evening of Sunday 11 May, Prime Minister Chavalit Yongchaiyudh vowed on television to defend the baht, then added that he could not promise to succeed. Like Britain in 1992, Thailand's leaders lacked the courage to accept an untenable position, and the Soros response was predictable.
- 14 May. Druckenmiller raises his short from $2 billion to $3.5 billion, still only about a third of capital; the central bank spends at least $6 billion defending the baht that day.
- The pack. Thai investors lead the flight; Paul Tudor Jones, who spoke with Druckenmiller several times daily, and other macro funds pile in; Tiger builds a short that reaches $2 billion.
- 15 May. Thailand bans banks from lending baht to anyone offshore; Tiger, rolling short-term baht loans, sees carrying costs hit $10 million a day in early June.
- Three weeks later. The baht has risen 10%; hedge funds have lost perhaps $500 million; the PM promises central bankers a victory party, and one paper says the bank aims to inflict $4 billion of losses on Soros.
- Druckenmiller trims from $3.5 billion to $3 billion.
- Hidden reserves. Official reserves still showed over $30 billion in late June, but by phoning the banks that executed the central bank's forward-market sales, Jones reckoned $21 billion had been used in May alone, two thirds of the war chest, fooling even IMF inspectors and members of the Thai government.
A visit from Stanley Fischer failed to persuade Thailand to devalue; it counterattacked instead. This was the second reason Druckenmiller never went all out: traders face political risk as well as market risk. Leveraging to the maximum might have forced devaluation, but was more than the political system seemed likely to tolerate. Yet he would not walk away, because he knew from Jones's sleuthing that Thailand was far closer to collapse than it admitted.
Jones took no pleasure in the finding; his moral defence of speculation was battered. With high rates and currency uncertainty, banks charged exorbitant rates to lend to each other, and money stopped flowing. On 25 June he wrote a memo, "The Economics of Deflation: What Keynes Would Say to Thailand," calling the choice of sky-high rates over devaluation sheer lunacy and citing Keynes on Emperor Aurelian, whose zeal for sound coinage in AD 274 caused deflationary misery and a rebellion that killed 7,000 soldiers.
The lending ban worked like plugging one hole in a showerhead: capital simply rushed out elsewhere. Harassing shorts prompted foreign dollar lenders to demand repayment, so Thai firms dumped baht, and offshore borrowing, though costly, remained possible for anyone sure of a collapse. On Tuesday 1 July, by at least one account, Tiger unleashed a barrage of selling, adding $1 billion to its short until the Bank of Thailand ran out of reserves and no buyers remained; Robertson played the part Druckenmiller had played against sterling. At about 4:30 am Bangkok time officials conceded they had no ammunition left. The baht fell 32% against the dollar over three months; Soros gained about $750 million, Tiger perhaps $300 million, and Thai output collapsed 17% from peak, plunging millions into poverty. Mallaby notes that 1 July 1997 was also the day Britain handed back Hong Kong, as a new kind of imperialism stamped Southeast Asia.
Mallaby's verdict on the blame game: the vilification of hedge funds had some fairness, since Soros led the January selling and Tiger delivered the final push (then almost unknown). But the roots lay in Thailand's refusal to devalue gradually in 1995-96 as China rose; speculators merely forced an inevitable adjustment. They triggered but did not cause the crisis, and Druckenmiller, fearing political backlash, actually cut his position in the endgame. That caution was a prophecy: as the crisis spread, the superpredator image grew ever less accurate.
Hong Kong: two Soroses on one stage
In late September 1997 Soros flew to Hong Kong for the annual World Bank and IMF meetings, and his reception mirrored his dual persona. The speculator was reviled: Malaysia's Prime Minister Mahathir Mohamad, who had already called him a "criminal" and a "moron," demanded a ban on currency trading, which he called unnecessary, unproductive and immoral. The statesman was lionised: Soros addressed a packed auditorium on stabilising the world economy, declaring that the main enemy of the open society was now the capitalist threat rather than the communist one, and that the laissez-faire faith in leaving markets alone was a dangerous idea, a stance the press loved coming from a capitalist billionaire.
By now Soros knew which persona he wanted on top: thinker, statesman, public figure, not neo-imperialist smasher of small currencies. Mallaby suspects this coloured his trading. Talking with Kowitz and Rodney Jones during the meetings, Soros declared the time for shorting Asian currencies had passed, even though Mahathir's outburst had triggered a fresh regional sell-off. The lieutenants doubted him, but he would not listen. He did not micromanage, leaving decisions to Druckenmiller and the team, but his optimistic bias was plain.
Indonesia: the wrong way
The bias showed first in Indonesia. Before the meetings the team bought about $300 million of rupiah, judging that Thai turmoil had spilled over without justification; unlike Thailand, Indonesia had not defended a doomed peg but let the currency fall 11% in August, so a rebound looked possible. After Fraga and Rodney Jones visited in October, the bet grew to about $1 billion.
Not everyone agreed: Tiger, also long rupiah, dumped its holdings at the end of October. The Soros team followed the technocratic consensus that the fall was temporary. On 2 November the IMF announced a $33 billion credit line, the rupiah rallied the next day, and Soros showed a small profit. Then the plan unravelled. The IMF money came with two commitments: close 16 corrupt banks and run a responsible monetary policy. The cronies around the ailing President Suharto owned those banks, so they could not be shut, and they hammered the central bank for loans, inflating the money supply. By late November the rupiah was in free fall; in early December rumours of Suharto's serious illness raised fears of a power vacuum. By 15 December the currency was down 44% from its early-November high and the trade had cost $400 million.
Strangely, the team stuck with it. After the 1987 crash Soros had dumped losing positions immediately to cap losses; now the team seemed paralysed. On 10 December Jones obtained Bank of Indonesia data confirming it had been printing money that flowed to crony-run lenders, and sent the details to New York. The position stayed.
Finance minister Mar'ie Muhammad, a respected technocrat, was sent to reassure foreign investors about a programme in which money was being printed for the Suharto family's friends. In New York he met Soros, Fraga and Druckenmiller at the Plaza Hotel; they fired question after question, and he mumbled evasive answers, unable to meet their eyes. Striding back toward the Soros offices at Columbus Circle, Druckenmiller exclaimed that he could not believe they were long, that he believed nothing the man said, and that he doubted he was even from Indonesia. Trapped in a trade with almost no rupiah buyers, Kowitz suggested converting the rupiah into a physical commodity such as iron for later barter. Soros called it an interesting idea; nobody acted, not even when the government, unable to print fast enough, issued plastic souvenir banknotes as legal tender.
The team rode the rupiah to the bottom and lost about $800 million, wiping out the entire baht profit. The fiasco dented the superpredator image, and it coincided with an extraordinary missed opportunity.
South Korea: no way
In mid-November 1997 Rodney Jones visited South Korea. One bank's boardroom was decorated with triumphant notices of financings for Thai companies that Jones knew had since gone bust; more than 50 of its Thai borrowers were behind on payments. Digging further, he found Thailand's collapse had left Korean financial firms short of dollars they would never recover, and that the central bank was covering up by depositing its foreign reserves in wounded Korean banks. The bombshell: like Thailand's, Korea's central bank was misleading markets. Official reserves were $57 billion; net of sums promised to banks or committed in forward markets, the real figure was closer to $20 billion.
This was a second Thai suitcase of cash. The won had dipped only 4% in the previous fortnight; nobody imagined two thirds of reserves were gone, and the IMF's annual review a month earlier had declared Korea immune from regional turmoil. Jones's memo of 17 November said Korea was in the late stage of the crisis, that official figures understated corporate dollar debts by about $60 billion, and that much of it matured within weeks.
| Market | Position | What happened | Outcome for Soros |
|---|---|---|---|
| Thailand | Short baht: $2bn (Jan), $3.5bn (May), trimmed to $3bn | Lending ban and intervention hurt shorts; reserves secretly drained; peg snaps 1-2 July 1997; baht -32% in three months | Gain ~$750m (Tiger ~$300m) |
| Indonesia | Long rupiah: $300m, raised to ~$1bn after an October visit | Tiger sold out late October; IMF $33bn credit line 2 Nov; cronies blocked closing 16 banks and forced money printing; rupiah -44% by 15 Dec | Loss $400m by mid-December, ~$800m in all |
| South Korea | None taken | Jones's 17 Nov memo: true reserves ~$20bn not $57bn; debts understated by $60bn; IMF finds $9bn left; record $55bn package 3 Dec; won -60% | Zero; Quantum invested only in October 1998 |
| Russia | Long: Svyazinvest $980m plus bonds and equities | Default and devaluation 17 Aug 1998 | Loss 15% of capital, $1-2bn |
The memo proved prescient. Nine days later the IMF's top Asia official flew to Seoul and found reserves falling $1 billion a day, down to $9 billion, with short-maturity dollar debt draining the country. On 3 December the IMF announced a record $55 billion package, still inadequate against private dollar obligations more than twice that size. By end-December the won was 60% below its level on the day of the memo.
Yet the call earned the Soros team nothing: despite the memo and a follow-up the next day, nobody shorted the won. Mallaby is unsure why. Top IMF officials still thought Korea would escape, which may have diverted attention; but he suspects the dual persona. Soros wanted to be a statesman, not a wrecker of nations; in Korea he would be saviour, not scourge.
In the first days of January 1998 Soros visited as guest of president-elect Kim Dae-jung, met by camera crews. He dined at Kim's home, calling him "DJ," toured top industrialists and breakfasted with Michael Jackson, who was plotting to take over a theme park from a bust Korean underwear maker. Soros criticised the IMF prescription of piling on more debt and called for radical restructuring of industry and finance, including cleaner accounting and freedom for managers to fire workers; if Korea complied, he said, Quantum would invest heavily and other Western money would follow. The KOSPI jumped a quarter in ten days.
The trip did his fortune few favours: having missed the won's fall, his funds did not buy into Korea's rebound until October 1998. Its payoff was reputational. The press contrasted Hong Kong, where one Asian leader had called him a criminal and moron, with Seoul, where another rolled out the red carpet. Asked about Mahathir versus Kim, Soros smiled and said one of them must be wrong.
Russia: the messiah's descent
The clash of personas was sharpest in Russia. In 1987, before the Soviet collapse, Soros had opened a Moscow branch of his Open Society Institute; in the 1990s it backed educational reform, textbooks free of Marxism and millions of dollars in grants for scientists. Hungarian historian István Rév, on his philanthropic boards, thought Russia drew Soros as it had drawn Napoleon: its vastness, backwardness and perpetually unfulfilled promise. So that his charity would not look like a Trojan horse for his money, Soros made it a rule not to invest personally in Russia, though Druckenmiller's team could. In spring 1997 he cracked.
He was not alone. After President Boris Yeltsin re-energised reform at the end of 1996, hedge-fund managers jetted to Moscow for the Bolshoi and walks in the Novodevichy Convent gardens. Foreign portfolio investment jumped from $8.9 billion in 1996 to $45.6 billion in 1997, about 10% of GDP, and the stock index nearly tripled in nine months. Property rights and the rule of law were vague, but portfolio investors reasoned they could dump and exit if reformers lost out.
Svyazinvest and the secret loan
Soros, a creature of markets who knew the value of an exit strategy, now staked $980 million on something almost totally illiquid. Going over Druckenmiller's head, he joined a consortium bidding for 25% of Svyazinvest, the sprawling state telephone utility. Long-term potential was real: Russia had 19 phone lines per 100 people against 58 in the US. But a roughly $1 billion stake in a state company could not be dumped if politics turned, and even amid the portfolio flood, foreign direct investment remained a trickle.
What made it crazier was what Soros knew. In June 1997, shortly before the auction closed, the government secretly asked him for emergency money: Yeltsin had sworn to start paying the backlog of unpaid wages and pensions by 1 July. Without the markets or the IMF (which was monitoring Russia's debt) knowing, Soros lent several hundred million dollars; without it, Yeltsin's brittle legitimacy might have cracked and workers might have rioted.
Mallaby reads the loan two ways. As statesman, Soros was going behind the IMF's back even while urging Russia to become a responsible member of the international monetary system. As investor, the loan screamed that stability was tenuous, yet he was about to place a hard-to-exit $1 billion bet that Russia had turned the corner. Against sterling and the baht his team had profited by reading governments' frailties; here he had a privileged window onto them and invested as though blind.
Mallaby's explanation is Soros's messiah complex, which Soros himself described. Pure philanthropy, he wrote, felt too good to be true and fed a godlike self-image; Russians could not understand what he was about. Appearing as a robber capitalist with civic values would be more credible than a disembodied intellect preaching the open society, and would offer a role model for Russia's budding robber capitalists; by entering as an investor he would descend from Mount Olympus and become flesh and blood.
He hoped Svyazinvest would mark a turning point. Until 1997 state assets had gone to oligarchs at knocked-down prices with foreigners excluded; this time foreigners could bid and the highest bidder would win. In July 1997 Soros's consortium did win with the most money, though his presence was hardly necessary for that, and the victory was Pyrrhic. The losing oligarchs owned newspapers and TV stations, and their smear campaign dragged on for weeks, forcing three officials to resign and derailing Yeltsin's reform agenda. Far from ushering in cleaner capitalism, the deal plunged the government into chaos.
Asia's shock waves reach Moscow
Then the Asian contagion arrived. Banks hurt in Thailand, Indonesia and Korea pulled loans from Russia; Russian financiers, reading the Svyazinvest war as the end of reform, moved money out; and Asia's slump drove down the price of oil, Russia's main export. Squeezed between falling export income and capital flight, the government had to offer ever higher yields. Short-term ruble bonds (GKOs) yielded 30% annualised in April 1998 and 70% in May.
That made Soros's illiquid stake look crazy, but 70% on three-month paper set half of New York's hedge funds salivating. The argument ran that the West would never let a nuclear power default and descend into chaos; at worst the US would make the IMF increase support. In June Goldman Sachs underwrote a $1.25 billion Russian bond issue that sold out within an hour; every macro investor from Soros to Tiger wanted in.
Tiger's palladium gambit
The most exotic Russia play had nothing to do with GKOs. Since 1994 Anderson had kept building Tiger's palladium position and had learned Russia's palladium bazaar: each year parliament and the state export company haggled over how much to sell, then the central bank, finance ministry and Noril'sk officials fought over who sold how much. Anderson cultivated the key sellers over long meals and whiskies and bought whenever he could. In summer 1998 Russia's desperation offered a way to break the pattern.
- The offer. Tiger would buy the entire stock of non-gold precious metals held by the central bank and finance ministry: all the palladium, rhodium and silver.
- The logistics. Some metal would come from Siberia to Moscow by armoured train; the consignment would be flown to a Swiss vault in Tiger's name.
- The money. Tiger would pay $4 billion on delivery, easing the budget at once, then sell gradually and share profits; Anderson reckoned Moscow's eventual take at about $8 billion.
- July 1998. With GKO yields above 100%, Anderson met a senior central banker in Moscow; they got on well, and Yeltsin apparently knew and approved. Back on Park Avenue, Anderson thought one New York fund was about to buy a country's treasure.
- The catch. Moscow demanded a commission, via two contracts: an official one omitting it and a private one directing payment to a bank in Cyprus, so parliament and public would never ask who kept it.
- The refusal. Anderson said Tiger paid commissions to brokers all the time, but there had to be one contract, disclosed. Moscow came back offering Gibraltar instead. Anderson explained that the haven did not matter: both sides' lawyers had to see the same paper, able to withstand public scrutiny in Russia and the West. No agreement was possible, and the deal died in late July.
The default
By then the crisis was intensifying. An IMF loan announced on 13 July calmed markets only briefly; the following week a second Goldman-backed issue seeking $6.4 billion drew bids for only $4.4 billion. Bonds tumbled, then stocks; even risk-hungry hedge funds grew leery. Yeltsin's team reopened secret talks with Soros, proposing to auction another 25% of Svyazinvest while the first auction's winners provided an immediate bridge loan, as Soros had done a year earlier.
This time Soros would not play: a few hundred million had sufficed in 1997, but now billions were needed. On Friday 7 August he phoned Anatoly Chubais and Yegor Gaidar, Yeltsin's top economic officials, and told them to get real: nobody would roll over the maturing GKOs. Gaidar put the funding need at $7 billion; Soros said $10 billion, with $500 million from the Svyazinvest consortium and the rest from Western banks and governments.
He then worked Washington. He urged David Lipton, the Treasury's top international official, to use the facility employed in the Mexico bailout; Lipton replied that Congress had no appetite, and the July IMF package had been Russia's last chance. On Monday 10 August Soros spoke to Lipton again, then to Deputy Secretary Lawrence Summers, and on Friday to Secretary Robert Rubin; he even had Republican Senator Mitch McConnell call Rubin to offer his party's backing. In Korea he had played the IMF, prescribing policies; now he aspired to broker a full government rescue.
The dual persona backfired. Officials could not tell whether Soros spoke for the world or his portfolio. When he went public with a long Financial Times letter on 13 August, proposing a 15-25% devaluation inside a Western financing package, the sensible prescription read to every trader as an announcement that Soros was shorting the ruble. Investors fled; GKO yields hit 165% the next day, and on Monday 17 August 1998 Russia devalued and defaulted on its foreign debts.
In truth Soros had not shorted the ruble. He had seen devaluation coming and even hastened it, but instead of selling a currency he knew politicians would not save, he had tried to make politicians behave differently. Far from short, the funds were vastly long: beyond the $1 billion Svyazinvest exposure, they held all kinds of Russian bonds and equities. Quantum and its sister funds lost 15% of capital, between $1 billion and $2 billion. For Druckenmiller, whose lifework was Quantum's performance, it was bitter; for anyone who knew the size and manner of the loss, it made nonsense of the superpredator label. Soros processed it on another level, writing that he had no regrets about trying to move Russia toward an open society: he had failed, but at least he had tried. Mallaby's closing image: the messiah came down from Olympus to save sinners and was crucified.
Thailand's and Korea's hidden forward commitments and reserves parked in failing banks are why the IMF introduced the Reserves Data Template (1999), requiring central banks to disclose forward positions and encumbered reserves. For bank country-risk analysts, usable reserves net of forwards and short-term external debt (the Greenspan-Guidotti ratio) matter more than the headline figure. Tiger's insistence on a single disclosed contract is also, in modern terms, basic anti-bribery compliance.
Mallaby's "uncanny coincidence" with the 1 July Hong Kong handover holds only on New York time: Thailand announced the baht float on the morning of 2 July in Bangkok (his own "4:30 a.m. Bangkok time" implies this). The claim that Tiger's selling on 1 July pushed Thailand over rests on "at least one account," and his explanations for the Korea miss and the Indonesia paralysis (Soros's wish to be a saviour) are admitted speculation. The $1-2 billion Russian loss range is wide for a figure Mallaby also gives precisely as 15% of capital.
Soros's Russian philanthropy ended in hostility: in November 2015 Russia's prosecutor general declared the Open Society Foundations an "undesirable" organization and banned its work.
The Enemy Is Us
Long-Term Capital Management turned tiny bond-price anomalies into returns of over 40% a year through leverage, then lost 44% of its capital in August 1998 and needed a $3.625 billion Fed-brokered rescue in September. Mallaby rejects the cartoon of reckless gamblers: the partners kept their own savings in the fund and ran value-at-risk, stress tests and liquidity safeguards, and still failed. The real lesson of LTCM is that all attempts to be precise about risk are unavoidably brittle, because the correlations that matter in a crisis are created by the traders themselves.
Two Bank of China buildings
For years Hong Kong's Bank of China building was squat and heavy, the granite solidity of old finance. In 1990 the bank emerged from a makeover by I. M. Pei as a slim, daring tower whose triangular, tapering shafts of aluminium and glass were meant to evoke bamboo. Mallaby pushes the image further than Pei intended: thanks to leverage, financial institutions were growing as fast as bamboo, sucking money out of the world's bond markets and pumping it back, but they were also thin-sided and hollow, vulnerable to an unexpected sideways blow.
In September 1997, during the IMF/World Bank meetings, LTCM, the epitome of the new financial engineering, threw a party in the art gallery of the squat old Bank of China building. Thailand and Indonesia were in crisis and Mahathir was campaigning against speculators nearby, but LTCM seemed to float above the storm: its small team of economists had earned $2.1 billion the previous year with a formula that ignored Asia's turmoil, and its guests were an A-list of bankers and officials. Back at their hotel the partners found a fax of that morning's Wall Street Journal front page reporting that they would return two fifths of the fund's capital, $2.7 billion, to outside investors. By borrowing more, LTCM could keep its towering portfolio on a thinner foundation, slender like a Pei creation.
Salomon without the bullshit
Founder John Meriwether was among the first on Wall Street to see the potential of financial engineering. As a rising star at Salomon Brothers in the mid-1980s he set out to make his small trading group a "quasi-university": he hired PhD stars and kept them on the academic circuit, visiting finance faculties and conferences.
- John Meriwether. Founder; built Salomon's arbitrage group, rose to vice chairman, resigned in 1991.
- Eric Rosenfeld. Former Harvard Business School professor; later named trader-driven correlation as LTCM's central error.
- Larry Hilibrand. Holder of two MIT degrees; compared markets to Slinkies.
- Robert Merton and Myron Scholes. Options-pricing pioneers on the team by 1990; Nobel laureates in October 1997.
- Greg Hawkins. Partner who kept thoroughbred horses.
- Jim Rickards. LTCM's general counsel.
- Peter Fisher. Number two at the New York Fed, who brokered the rescue.
- Jon Corzine. Head of Goldman Sachs.
In the mid-1980s most Salomon partners had not gone to college, let alone done a PhD. The firm's trading culture was personified by government-bond chief Craig Coats Jr., a tall, charismatic figure thought by many to be the model for the hero of Tom Wolfe's The Bonfire of the Vanities, whose main tool was faith in his own instincts while Meriwether's professors debated whether bond relationships were out of range or volatility likely to slow. The professors' superiority soon showed: by the end of the 1980s the small quasi-faculty produced 90% of Salomon's profits. Coats left after a big trading loss and Meriwether became vice chairman. Quantitative precision had triumphed; Mallaby calls it the end of Wall Street's anti-intellectualism.
The victory raised a question: if a hundred-strong department generated tens of millions, what were Salomon's 6,000 other employees for? Meriwether's lieutenants complained their brilliance was subsidising undeserving corners. Hilibrand campaigned to shut the bankers' fancy catering service and even suggested closing investment banking entirely. The result was a secret deal granting the group a fixed 15% of its profits: a guaranteed performance fee that made it a hedge fund inside a bank, one step from a hedge fund instead of a bank.
The step was forced on Meriwether. In 1991 the Treasury auction scandal that embarrassed Steinhardt and Kovner engulfed Salomon: its government-bond trader Paul Mozer had cheated repeatedly, and Meriwether, his supervisor, resigned and paid a $50,000 SEC fine. He resolved to reassemble his rocket scientists without a big bank's trappings, outsourcing marketing, clearing, settlement and operations so profits would not leak to back offices. A world in which a small brotherhood of academics could out-earn a large bank needed a new institution: "Salomon without the bullshit."
The convergence machine
Meriwether launched LTCM in February 1994 with eight members of his Salomon brain trust, including Rosenfeld, Hilibrand, Merton and Scholes. They leased space at 600 Steamboat Road, Greenwich, Connecticut, the four-storey building on Long Island Sound to which Paul Tudor Jones had recently moved, and came to work in golf shirts and chinos; when bluefish jumped in the sound at lunchtime, quants grabbed rods and chased them.
Mallaby frames the method as A. W. Jones's hedge transplanted to bonds. A Jones manager might buy Ford and short Chrysler on a judgment of management, car designs and sales-team gossip, which is only an opinion. Bonds are simpler: a loan, a rate and a promise to repay on a set date. Two bonds can be near twins, with the same issuer, same repayment year and word-for-word identical documents; if one trades cheaper, you buy it and short its pair. That is arbitrage, promising almost certain profit, and the hedge works better than with stocks.
The signature trades
- On-the-run versus off-the-run Treasuries. Newly issued "on-the-run" bonds trade constantly, so traders pay a liquidity premium over slightly older "off-the-run" ones; but a 30-year and a 29½-year bond must converge by maturity. LTCM sold the dear new bond, bought the cheap old one and waited. In normal times the profit barely covered costs, but in panics the premium ballooned, and LTCM struck then. Hilibrand likened markets to Slinkies: they always spring back once panic subsides.
- Italy. Cumbersome tax rules deterred foreigners, so Italian bonds were cheap: a foreigner could earn about 10% and hedge by borrowing lire at about 9%, keeping 1 point. The solution was a partnership with an Italian-registered bank untouched by the tax. From the Salomon days into LTCM's first years the team pursued it more keenly than anyone; Italy produced about $600 million of LTCM's $1.6 billion profits in its first two years.
- Italy, round two. Others crowded in and the government scrapped the tax obstacle, so LTCM found a new angle: retail savers bought a particular government bond from corner-shop banks, leaving it overvalued. LTCM shorted it and bought higher-yielding lire bonds; as savers drifted into mutual funds, demand for the shorted bond fell further, adding profit.
- Everything else. Convergences between bonds of the same maturity, a bond and its futures contract, Treasuries and mortgage-backed bonds, and bonds in different currencies.
The common theme was that anomalies arise when investor behaviour is distorted by tax rules, regulation or institutions' idiosyncratic needs. French insurers had to buy French government bonds of particular maturities to match their policy liabilities, bargain or not. An October 1996 Federal Reserve ruling led US banks to issue bonds and swap the payments into floating rates, flooding the market and depressing the fixed swap rate. LTCM took the other side of such forced trades. As the flexible player mirroring inflexible ones, it supplied liquidity, let insurers and banks trade at better prices than they otherwise would, and reaped fabulous profits.
- 19.9% net in ten months of 1994, 42.8% in 1995, 40.8% in 1996, after a 2% management fee and a hefty 25% performance fee.
- Uncorrelated with any stock or bond index.
- 41 staff at launch, 165 by the autumn 1997 party, with offices added in London and Tokyo.
- 2.45% return on assets in 1995 became 42.8% on capital through leverage.
- $2 billion on/off-the-run position soon after launch on a $1 billion fund, judged 1/25 as risky as an outright bond, so equivalent to an $80 million unhedged bet.
Capital poured in, and the trappings followed: Rosenfeld's 10,000-bottle wine cellar stocked from France, Hawkins's thoroughbreds, and Meriwether's purchase of Waterville, a golf course in County Kerry, Ireland, where he courted Wall Street heavyweights. A Merrill Lynch salesman recalled that everyone was enamoured of their intellect, likening them to Kennedy's Camelot. Within a year of the Bank of China party the fund had blown up, showing once more that newfangled finance could carry large, unappreciated risks for the world economy.
Not reckless, just brittle
Roger Lowenstein's best-selling account casts LTCM's demise as punishment for hubris. Mallaby agrees that this is ultimately right, but insists the firm was not crass about risk. The partners were not gambling OPM (Wall Street's sneer for other people's money): most reinvested nearly all their earnings year after year, and after ejecting $2.7 billion of outside money, nearly a third of the fund was their own savings. Unlike many who ruin their institutions, they had every incentive to be prudent.
Leverage was the essence of the firm, since the anomalies were too small to matter without borrowed money. Meriwether reasoned it was safe because LTCM hedged out nearly all risk in each trade. And LTCM was among the first hedge funds to quantify risk mathematically.
Value at risk and the magic of diversification
Macro traders like Druckenmiller kept their exposures in their heads. LTCM used value at risk (VaR), a 1980s technique that formalised those mental sums: work out each position's volatility and turn it into a dollar loss possible in normal times.
- Position VaR. If history showed that on 99 days in 100 the worst move in an Italian bond pair was a 10bp widening, and each basis point cost $10 million, the position's VaR was $100 million.
- Perfect correlation adds. $100m at risk in California bonds plus $100m in New York bonds moving in lockstep is $200m.
- No correlation dampens. Uncorrelated risks combine as total = $100m × √n: two bets (say mortgage prepayment and Italian retail bonds) give $141m; the third adds only $32m, the fifth $24m, the tenth $16m.
- The consequence. Through diversification risk could almost disappear, so trades that looked mad to others looked richly rewarding to LTCM.
After the 2007-09 credit bust VaR fell from favour, and Warren Buffett warned against geeks bearing formulas. But Mallaby argues VaR was an advance on the leverage ratio, which ignores swaps and options, treats a hedged convergence bet as no less risky than an outright one, and, most fundamentally, compares capital to assets rather than to potential losses. Since capital exists to absorb losses, LTCM's yardstick was the right one; the old approach was like sizing the US armed forces against the number of foreigners on the planet, when there was no prospect that all would attack at once.
The objections LTCM had already answered
- The worst day in a hundred. VaR by definition ignores the bad tail; critics later likened it to an airbag that works except in a collision. LTCM relied on markets' self-correction: when a relationship broke, other arbitrageurs would see profit and buy, pulling prices back, the Slinky effect.
- Unprecedented shocks. Models built on history cannot foresee the unprecedented, so the partners brainstormed surprises and stress-tested repeatedly, for example against a derailment of European Monetary Union, cutting positions whenever losses could threaten survival.
- Liquidity risk. Many trades paired an illiquid instrument with a liquid hedge (off-the-run bonds were cheap precisely because illiquid). In a panic, liquidity premiums would rise everywhere at once, apparently uncorrelated bets would lose together, and brokers and investors might pull money, forcing fire sales: arbitrage's corrective power could prove an ambulance parked too far away.
- The liquidity defences. Hence the name Long-Term: instead of short broker funding, Meriwether sought term loans and standby credit lines, and instead of monthly or quarterly redemptions he required a three-year lock-up, later allowing withdrawals of up to a third a year.
After the collapse, LTCM was seen as a victim of naive faith in models, implying a little more caution would prevent a repeat. Mallaby's truth is harder: the decade's prescriptions (add stress tests to VaR, watch liquidity risk) were things LTCM already did. It failed because stress-testing means imagining the inconceivable and VaR means guessing how correlations change in extremes; the approach was not too simple, precision itself was the problem.
Overreach: equities and the central bank of volatility
In October 1997, soon after the party, Merton and Scholes won the Nobel Prize in economics. The award was hailed as a vindication of the new finance: a formula to price risk had let it be sliced, bundled and traded a thousand ways, taming with an equation the fear of loss that had braked human endeavour for centuries. LTCM was celebrated as the professors' vision fulfilled, holding exactly the capital it needed and no more.
Yet it stood at a fateful crossroads. The upstarts of the 1980s now faced fixed-income arbitrage desks at nearly every investment bank, and in the first half of 1997 profits slowed. One answer was to shrink by returning money, which LTCM did only in part; to keep the machine going it moved into equities, seeking trades with the mathematical clarity of bonds rather than picking stocks in Robertson's manner.
- Dual-listed pairs. The favourite was Royal Dutch/Shell, owned by Royal Dutch Petroleum (Netherlands) and Shell Transport (Britain). Both shares claimed the same profits, yet the British traded at a discount. LTCM bought London and shorted Amsterdam, collecting more in dividends than it paid on the short and indifferent to the share price. It played the trade in vast size; many on Wall Street felt the Greenwich professors were overreaching.
- Selling volatility. In autumn 1997 jittery investors were overpaying to insure against sharp moves, buying calls in case of a spike and puts against a crash. Five-year S&P 500 options implied volatility of 19% a year, against 10-13% typical in the 1990s. LTCM, betting on how far markets would move rather than which way, sold calls and puts and pocketed the premiums. Morgan Stanley dubbed it "the central bank of volatility."
The partners kept their prudential rules. VaR said the most they were likely to lose on 99 days in 100 was $116 million, and over 21 trading days $532 million. The worst month ever had been -2.9%, and ruin would take a ten-sigma event, one in 10^24. At twice-weekly risk meetings they gamed a Wall Street crash or a Japanese earthquake. Rosenfeld later said they had thought they were being conservative.
The first warning: May-June 1998
In May 1998 Indonesia's IMF bailout faltered and Suharto fell; Japan entered its first recession (two quarters of falling output) since 1974; Russia tripled interest rates to halt capital flight; equity volatility jumped worldwide. Investors fled into US Treasuries, driving the 30-year yield to its lowest since the bond was first issued in 1977. In other words, they panicked, and LTCM's shared premise, that calm arbitrage would correct irrational panic, broke.
- Swap spreads. LTCM was short Treasuries and long the swap rate, betting that an abnormally wide gap reflected temporary risk aversion; extreme risk aversion widened it further.
- Volatility. Each point of S&P implied volatility cost $40 million; it rose from 19% on 1 May to 26% by mid-June.
- The damage. The fund lost 6% in May and 10% in June, far beyond its VaR. Partners cancelled holidays and debated whether something fundamental had gone wrong; then July brought a recovery, they trimmed positions, relaxed and wrote the spring off as a freak.
LTCM's book on the eve of the crisis
| Trade | Long / short | Premise | What happened in 1998 |
|---|---|---|---|
| On/off-the-run | Long old Treasuries, short new | Liquidity premium must vanish by maturity | Flight to the most liquid paper widened the premium |
| US swap spread | Long swap rate, short Treasuries | Wide gap is passing risk aversion | Gap moved 8bp in a day (normal under 1bp) |
| UK swap spread | Long market rates, short gilts | Same as US | Widened sharply in August, then in September as LTCM became the target |
| Emerging markets | Long risky bonds, short stable ones | Credit spreads mean-revert | Lost badly after the Russian default |
| Royal Dutch/Shell | Long London, short Amsterdam | Same profits, so the discount should close | Premium jumped 8% to 17% |
| Equity volatility | Sold S&P calls and puts | 19% implied vol versus 10-13% normal | Rose to 26% by June; by 21 Sep prices implied a monthly crash |
| Merger arbitrage | Long targets of announced deals | Deals mostly close | $5 billion book offered to Buffett, who declined |
August 1998: the Slinky stops springing
On Monday 17 August Russia defaulted on its foreign debts. Unlike Soros, LTCM had little direct exposure, but the indirect effects were appalling. The default bankrupted a hedge fund aptly named High-Risk Opportunities, threatening firms it owed; by Friday 21 August rumours said Lehman Brothers might fail, triggering another flight to quality. Safe government paper rose and everything risky fell, more viciously than in spring.
- US swap spread. The Treasury-swap gap, which usually moved less than a basis point a day, widened by 8 basis points.
- Britain. A similar bet on gilts versus market rates blew out too.
- Emerging markets. The same structure (short stable bonds, long risky ones) lost badly.
- The tally. By the end of that Friday LTCM had lost $550 million, 15% of its capital.
Most partners were on deferred holidays: Meriwether in China, Rosenfeld in Idaho, Rickards in North Carolina. The skeleton crew in Greenwich watched what Mallaby calls a bonfire of the fund's vanities. Credit relationships assumed stable were stormy; strategies assumed uncorrelated fell in lockstep; a one-day VaR of $116 million was out by more than $400 million; and the market did not spring back. Rosenfeld phoned at midday from an Idaho golf course, his 9 a.m. tee time delayed, to check all was fine, and knew his holiday was over. Before flying home, Meriwether called Jon Corzine of Goldman to say LTCM had had a bad day but there was no cause to worry.
The scramble for capital
That Sunday evening the partners planned how to stem losses, knowing that selling into a panic would be hard.
- Buffett, round one. Rosenfeld asked Warren Buffett to buy LTCM's $5 billion merger-arbitrage book; Buffett graciously declined.
- Soros. Next morning Meriwether breakfasted with Soros and Druckenmiller at Soros's Fifth Avenue apartment, pitching concessionary terms on bargain positions. They seemed intrigued, but the most he got was a conditional offer: $500 million at the end of August if he raised another $500 million elsewhere.
- Everyone else. He tried old colleagues at Salomon, sent Hilibrand to Omaha with a better offer, and approached Saudi Prince Alwaleed and computer magnate Michael Dell.
- Merrill. Herb Allison, despite the Irish golf, warned that raising money might make it look like a problem: infuriating but true.
Each pitch spread alarm. Buffett, while rebuffing Hilibrand, ordered lieutenants to check Berkshire Hathaway's hedge-fund exposure; Salomon's bosses told their brokerage to stop lending to almost all hedge funds. Mallaby likens the partners to runners on a moving carpet carrying them away from home: their fund-raising was destroying the capital it was meant to rebuild.
The Pogo moment
Arbitrageurs faced what Mallaby calls a Pogo moment, after Walt Kelly's comic-strip line: "We have met the enemy and he is us." Asia and then Russia had hurt every big arbitrage player; short of capital and hit by margin calls, they could not correct anomalies and were forced to dump positions, pushing prices away from equilibrium. Arbitrage became reverse arbitrage, each unwinding portfolio damaging the others in a self-reinforcing cycle. Meriwether's own legacy, Salomon's Arbitrage Group, had done so badly that its bosses were closing it, and its selling hammered LTCM.
Every LTCM trade went wrong in a correlated way, not because the trades were economically alike but because the same kind of funds held them. Rosenfeld now regards failing to anticipate this trader-driven correlation as the fund's central error: had they foreseen it, diversification would not have seemed to magic risk away, and every trade would have been sized more cautiously.
Crowded by its own disciples
The second underestimated risk came from LTCM's own success. Although it spread trades across many brokers so none could see the whole book, an army of imitators pieced together its strategy, building a shadow superportfolio larger than LTCM's own. By spring 1998 everyone who might buy LTCM's positions already owned them; if it had to retreat, there would be nobody to sell to. Canny traders saw a one-way bet: the positions could not rise, but a shock forcing arbitrageurs to sell would make them collapse. The "central bank of volatility" was like a real central bank defending a peg, and once predators noticed, the game was up.
The irony was painful. For four years LTCM had profited from counterparties acting on institutional imperatives rather than market judgment; now, with banks pulling loans, arbitrageurs were themselves forced sellers, and the biggest prize was to short whatever LTCM owned. Meriwether noticed that trades unknown to the Street recovered after the post-Russia shock, while known ones kept bleeding. Spreads between risky bonds and Treasuries kept widening, and the Royal Dutch/Shell premium jumped from 8% to 17% without any fundamental reason. One team member recalled the moment it stopped feeling as if others were selling positions similar to theirs: they were selling LTCM's positions.
- 3 in 4 hedge funds lost money by Labor Day.
- -44%: LTCM's loss for the month, $1.9 billion, an event its models said should occur less than once in the lifetime of the universe.
- -$550 million on Friday 21 August alone, 15% of capital.
- 8bp daily move in a spread that normally moved under 1bp.
- 8% → 17%: the Royal Dutch/Shell premium.
At month's end Meriwether called Vinny Mattone, an old friend at Bear Stearns, and said they were down by half. Mattone told him he was finished. Meriwether protested that they still had $2 billion and Soros; Mattone replied that once you are down by half, people assume you can go to zero and push the market against you. Calls to Merrill Lynch, UBS and Deutsche Bank produced no $500 million, and the Soros offer lapsed.
September: hyenas at the carcass
Meriwether reported the losses in an investor letter on 2 September, reaffirming his faith in arbitrage and calling the environment unusually attractive. Nobody was fooled. It leaked before the last copy was faxed, and the Journal ran it on the front page the next day. Now the whole Street traded against LTCM. Most junk bonds rallied in early September, but LTCM's stayed dead. Its small position in hurricane bonds (which let insurers sell hurricane risk) fell 20% the day the letter leaked, though hurricane odds were unchanged. In Europe the gap between government bonds and market rates widened in Britain and narrowed in Germany for no reason except that LTCM was positioned the opposite way. Rickards told Meriwether that it was no longer about the market but about him: whatever LTCM was long, others were short, and vice versa.
On 13 September Meriwether asked Corzine for help raising the $2 billion now needed to avoid bankruptcy. Goldman agreed, at an extraordinary price: $1 billion of its own plus a promise to raise another, in exchange for half the management company, full access to the strategies and the right to limit positions, conditional on raising the money and passing a detailed inspection. Goldman won either way: half of LTCM at a bargain, or a free look at books showing exactly which trades would crater in a fire sale. Rickards asked the inspectors to sign a nondisclosure agreement and was told Goldman would sign nothing. He says they plugged an oversized laptop into LTCM's network and downloaded its positions; Goldman denied it, but its proprietary desk was soon selling similar positions, feeding like a hyena on a trapped antelope. A Goldman trader in London said that if a gorilla has to sell, you want to sell first, and that this was not illegal. Corzine conceded Goldman may have hurt LTCM while protecting its own positions and was not apologetic. Goldman cited a Chinese wall between inspectors and traders; there was no proof otherwise, but some suspected it was porous.
The Fed steps in
The partners told the Federal Reserve that failure was possible. With president William McDonough leaving town, it fell to his number two, Peter Fisher, an unflappable expert in financial plumbing who cut through complexity with simple propositions. On Sunday 20 September he rode in an assistant's Jeep from New Jersey to Greenwich, and no neat solution came to mind.
- Scale. Leverage had built a portfolio of $120 billion.
- Concentration. LTCM might hold half the open interest in gilt futures, an outlandish position in Danish mortgages and an enormous equity options book. Fisher feared some markets might stop trading altogether.
- Counterparty losses. By LTCM's reckoning its 17 biggest counterparties faced about $3 billion of losses on an immediate wind-down, more in a rushed liquidation, plus hits to their own copycat books.
- No assets to seize. Unlike a failed bank or brokerage, a hedge fund's assets sit with its brokers as collateral. One default would trigger cross-default clauses in every other agreement; close-out notices would flutter from a thousand fax machines and brokers would dump the positions fast. Sitting among LTCM's empty workstations with a Treasury colleague, Fisher faced Askin's problem from 1994, one that would haunt regulators again a decade later: firms entangled with dozens of partners are too intertwined to bury easily.
The context made it worse: East Asia in recession, Russia heading there, Wall Street already bruised and its immune system compromised. A $3 billion-plus hit could push a teetering Lehman over, with further consequences, and weak balance sheets made a clean private takeover less likely. That evening Corzine told Fisher that Goldman's search for a buyer was failing. LTCM was losing hundreds of millions a day and would be lucky to see another weekend.
On Monday 21 September Asian and European markets tumbled. TV pundits blamed the release of President Clinton's videotaped Lewinsky deposition, which Fisher found grimly amusing; in truth each rescue attempt had alerted another predator. Salomon said Goldman's Tokyo desk was pounding LTCM; Goldman said Salomon was doing the same in Europe. By midday attacks on LTCM's equity options were so extreme that prices implied a crash every month. LTCM lost $550 million that day, a third of its equity, and for the first time its capital fell below $1 billion.
A lone buyer would not work: it would have to resell some holdings, so others would keep selling ahead of it. A consortium, each member holding a digestible piece, would stop the hyenas feeding, but banking rivalries ruled that out unless the Fed brokered it.
Tue 22 Sep, breakfast Fisher hosts the heads of Goldman, Merrill Lynch and J.P. Morgan; all favour joint action. The choice is between creditors buying pieces of the portfolio or a consortium injecting capital; after two hours they go away to draft a plan all three can back, for a wider creditors' meeting.
Tue, end of day The banks conclude capital must be injected: lifting pieces out of LTCM's cat's cradle of trades is impossible.
The price tag $1 billion at the August Soros breakfast, $2 billion at Goldman three weeks later, $4 billion now; every bank tempted to free-ride.
Tue evening 16 counterparties at the Fed are asked for $250 million each; smaller banks object; Fisher adjourns around 11 pm until 10 the next morning.
Wed 23 Sep Lehman pleads, with some reason, that it is too fragile to pay $250 million; two French banks offer only $125 million each; lead broker Bear Stearns's Jimmy Cayne refuses a single dollar despite years of fees. Morgan Stanley's Philip Purcell calls it unacceptable and Merrill's David Komansky explodes at Cayne. Fisher prepares to order phones down and draft a "no deal" statement, so insiders could not sell ahead of ordinary investors.
Outcome Eleven banks raise their share to $300 million each; smaller sums from Lehman and the French bring the total to $3.625 billion, just enough. The partners' fortunes are reduced to almost nothing.
Rickards later said the partners had believed in themselves without hypocrisy, joking that they had lost money the old-fashioned way, through honesty and hard work.
What the rescue meant
At House hearings on 1 October, lawmakers vented about secretive traders. Pennsylvania Democrat Paul Kanjorski suggested America's enemies no longer needed weapons of mass destruction: they could simply partner with hedge funds. For all the outrage, Mallaby finds the debate an echo of 1994: regulators admitted hedge funds posed risks and did little.
It is tempting to be harsh. In 1994 regulators had said banks were too savvy to let hedge-fund clients take wild bets; now Wall Street's finest had needed the Fed to save them from their exposure. Yet Alan Greenspan again told Congress that lenders have a strong interest in getting their money back, relying on private creditors to police hedge funds. Not until 2009 did he concede that lender monitoring was a flimsy defence, and even then he presented the 2007-09 crisis as a bolt from the blue, as though LTCM had never happened.
The case for Greenspan's restraint
Mallaby sets out good reasons for the diffidence.
- Recklessness was not confined to hedge funds. LTCM's errors (too much leverage, ignoring copycat unwinds, overestimating the precision of risk measurement) could happen anywhere; LTCM frightened officials partly because Lehman was wobbling too. The real issue was system-wide leverage, exactly what Soros had told Congress in 1994. In 2007-09 the culprits were banks' off-balance-sheet conduits and SIVs, leveraged broker-dealers and a leveraged insurer: hedge funds were not the villains.
- No good tool. A cap on the leverage ratio was nearly meaningless, since it ignores hedging and derivatives; a cap on VaR had just been shown misleading. Real risk depended on shifting conditions: copycats, market liquidity, brokers' health. Peter Fisher saw the theoretical case for controls but doubted government would get the details right, so never pushed.
- The Basel II warning. From 1998 global regulators designed new risk rules for banks alone (excluding insurers, broker-dealers and hedge funds); it took six years and ended in abject failure, deferring to banks' own models, effectively self-regulation. European banks on Basel II proved hopelessly fragile in 2007. Controls on shadow banks might have fared no better.
- The cost of safety. As in 1994, regulators fell back on plan B: voice concern while assuring the nation the dangers were tolerable. Greenspan warned about models outrunning judgment and reminded creditors that hedge funds can blow up in sudden volatility, but noted that requiring banks to hold capital of 40% of assets, as after the Civil War, would mean less turbulence but costlier capital and lower living standards: you cannot have the system's benefits without its costs.
Fair at the time. In the context of 1998, the benefits of modern finance outweighed its risks, and regulatory fixes looked unlikely to work, as Basel II later confirmed.
Wrong in hindsight. Banks learned one lesson and cut hedge-fund leverage, as they should have after Askin. Others went unheeded: AIG later repeated LTCM's error of insuring the world against volatility without capital; investors repeated the diversification fallacy with bundles of supposedly diversified mortgage securities. Most fundamentally, the world answered LTCM's lesson about leveraged finance by tolerating a vast expansion of it.
LTCM is still the textbook case for three modern controls: stressed and liquidity-horizon measures rather than plain 99% VaR (the Basel FRTB framework replaced VaR with 97.5% expected shortfall for market-risk capital); crowding and wrong-way risk analysis of hedge-fund counterparties, where collateral falls just as exposure rises; and close-out planning for cross-default cascades. A counterparty's book is only as liquid as the least crowded exit.
Some details do not quite hold. The rescue arithmetic: eleven banks at $300 million ($3.3 billion) plus two French banks at $125 million each and Lehman would exceed $3.625 billion; Lowenstein's account has Paribas at $100 million, Société Générale at $125 million and Lehman at $100 million. The 1998 hearing was before the House Banking and Financial Services Committee, not "Financial Services" (renamed in 2001). Mallaby also omits that the consortium took 90% of the fund, wound it down by early 2000 and recovered its money, which bears on his judgment that the rescue was needed to avert broad losses. His sympathetic portrait of LTCM's risk management leans on partners' later recollections, notably Rosenfeld's.
The Dot-Com Double
Within a month of each other in March-April 2000, the two largest hedge funds of the era shut down: Julian Robertson closed Tiger after refusing to buy a bubble he rightly judged absurd, and Stan Druckenmiller left Soros's Quantum after riding the bubble and jumping back aboard one turn too late. Mallaby uses the dot-com mania as a test of whether hedge funds enforce efficient prices. They mostly did not: fighting the bubble was ruinous, so hedge funds were no more contrarian than other investors, and the fund that profited most did so by surfing momentum.
- Julian Robertson. Founder of Tiger, the great long/short value investor; lost on the yen in 1998, ignored tech in 1999, closed in March 2000.
- Michael Bills. Tiger lieutenant, son of a military aviator, who urged attacking the yen squeeze on 8 October 1998.
- Dan Morehead. Tiger's currency trader who executed the counterattack.
- Stan Druckenmiller. Ran Soros's Quantum for twelve years; shorted tech, then rode it, then left in April 2000.
- Carson Levit and Diane Hakala. Quantum's in-house tech enthusiasts (Hakala an aerobatic pilot in her spare time).
- David Einhorn. Founder of Greenlight Capital; lost shorting Chemdex.
- Mary Meeker. Morgan Stanley internet analyst, the "Queen of the Net."
- Philip Duff and John Griffin. Tiger's chief operating officer, and a former Robertson lieutenant then running his own fund.
Robertson's right stuff
Mallaby opens with Tom Wolfe's The Right Stuff, where military pilots explain away each comrade's fiery death as a mistake they would never make: one let his speed drop before lowering the flaps, another was inexperienced when his controls failed, a third forgot to check his oxygen hose and nosed into Chesapeake Bay. Each survivor is sure the lesson does not apply to him.
The week after LTCM's rescue, Robertson played the surviving pilot. His investor letter admitted Tiger had lost almost 10% in September 1998 but insisted LTCM's fate was irrelevant: Tiger had an eighteen-year record, sought big payoffs without dangerous leverage, and had been up a striking 70% before fees just a year earlier on stock picks and Asian currency trades. He compared LTCM's model to a high-volume, low-margin supermarket, nothing like Tiger, and asked what Tiger would change after the uproar over leverage. His answer was "nothing."
The yen trade
Within days Robertson had his own oxygen-hose problem. September's -10% was followed by -17% in October, and the pattern echoed LTCM: reasonable bets in unreasonable sizes, in this case a near-suicidal wager that the yen would fall against the dollar.
- The thesis. In July 1998, with Tiger up 29% for the half year, Robertson came back from a meeting of Tiger's external advisers in Europe (attended by Margaret Thatcher, Bob Dole and financiers from Japan and Mexico) convinced that Japan's financial deregulation would let savers move money abroad. With yen rates just over 1%, Japanese capital would flood out and the yen would fall. His letter told investors plainly that he was short the yen.
- The blind spot. As LTCM underrated deleveraging, so did Robertson. Precisely because yen rates were low, traders worldwide borrowed yen to fund their positions (the carry trade); if they dumped those positions and repaid the loans, the yen would rise.
- The trigger. Russia's default forced leveraged traders to sell, hitting LTCM and Tiger together. Friday 21 August, the day LTCM lost over half a billion, was the day Tiger's yen bet turned. Over the next month the yen rose 7% and Tiger lost over $1 billion.
- The exposure. As with LTCM, rivals knew too much. Robertson's monthly letters were faxed around Wall Street, so every trader knew he was short and expected him to be forced to buy yen back; on 7 October the yen jumped sharply and traders pushed harder, sensing Tiger would crack.
By about 10 a.m. on 8 October 1998 the yen had risen 12% since the previous morning, from 130 to 114 per dollar. Tiger had lost more than $2 billion in just over a day, two hundred times the $8.8 million it was founded with (Mallaby compares this to Morgan Stanley's $578 million loss over a whole quarter in 2009, front-page news then). Robertson summoned his top lieutenants to his corner office. Michael Bills, the aviator's son who had joined Tiger after Wolfe's call about fighter-pilot culture, argued that the market was crazed because it thought Tiger was finished; Tiger should add to its short to prove it was not easy prey. During the thirty-minute meeting the yen went to 112, costing another $500 million. Trader Dan Morehead then called a big bank for a two-way price, hiding his direction. Normally the quote took seconds; this time there was half a minute of silence, because sellers had vanished in expectation of Tiger's forced buying. The dealer quoted 113.5/111.5, a two-yen spread perhaps 40 times normal. Morehead said he was a buyer (of dollars, adding $50 million to the short). The bank instantly understood Tiger would not be squeezed out; its proprietary desk began dumping yen, the news spread to every currency desk, and the yen fell as fast as it had risen. Like LTCM, Tiger had learned that liquidity dries up exactly when it is needed.
Robertson was vindicated in one respect: Tiger's debt-to-equity ratio of about 5 to 1 gave it the strength to hold its position instead of being forced out, which LTCM could not do. That was little comfort. For October Tiger lost $3.1 billion on currencies, mostly the yen. Robertson told investors the yen had gone from liquid as water to dry as the Sahara, without admitting that his own recklessness had drained it: Tiger had been short about $18 billion of yen, almost twice Druckenmiller's famous sterling position. By out-trading Quantum in currencies, Mallaby says, Robertson baked his own Sahara. He promised to scale back currency trading, but the yen proved only a foretaste: the real damage came from a technology bull market.
The bubble as a test of hedge funds
Mallaby treats the late-1990s tech bubble as a test between two views of hedge funds.
Optimistic: efficiency enforcers. Sophisticated traders analyse prices and push them toward fair value. Long/short investors from A. W. Jones in equities to John Meriwether in bonds buy the cheap and sell the dear, driving prices to equilibrium. Faced with a bubble, they should short it down.
Pessimistic: trend riders. Traders know they lack the muscle to enforce efficiency, so they ride trends rather than fight them. Trend followers like Paul Tudor Jones claim no view of fundamental value, buying what rises and dumping what falls. Faced with a bubble, they buy in and reinforce it.
If any market cried out for arbitrage, this one did. Analysts normally value companies by earnings and growth, the cash that will eventually reach investors; the start-ups of the late 1990s had no earnings, so by traditional yardsticks their intrinsic worth was zero. Yet investors scrambled for anything that looked internet-related.
- Books-A-Million. In November 1998 this plodding bookseller announced an improved website and its shares rose tenfold in three days.
- Priceline.com. In March 1999 it gained 425% on its first trading day, making an untested airline-ticket website worth more than United, Continental and Northwest combined. The airlines owned terminals, landing slots and fleets; Priceline owned some software, a few computers and a stake in its spokesman, Star Trek actor William Shatner.
The stakes, for Mallaby, are social. The justification for highly paid money managers is that by enforcing efficient prices they steer scarce capital to companies that will use it best. Bubbles do the opposite: they fund managers who cannot manage and business plans nobody wants, then leave ordinary savers with the losses when they burst. If hedge funds were the enforcers optimists imagine, this was the moment to earn their keep.
No single investor can deflate a bubble. Tiger and Quantum, the largest hedge funds of the late 1990s, each peaked at just over $20 billion, while the tech-heavy NASDAQ was worth more than $5 trillion. A short against the bubble pays only if others join it; meanwhile losses mount, and if they frighten investors into withdrawing, the fund must unwind before the payoff. Keynes's warning applies: markets can stay irrational longer than you can stay solvent, and early and right equals wrong.
Robertson ignores the locomotive
Robertson had no doubt tech stocks were absurdly high, but he had lost on tech shorts in 1998 and concluded there was no safe way to fight the bubble. His method relied on hedging: short one company, go long similar ones, and the general market risk cancels. When an entire sector is overvalued there is no equivalent bucket to go long, so a short becomes a naked bet on timing. He likened the NASDAQ to a locomotive hurtling down the tracks: certain to derail eventually, but only a fool would stand in front of it.
So he chose to ignore the bubble rather than fight it, which created a new problem. Having already sworn off big macro bets after the yen, he now ruled out tech too, and with two frontiers closed, Tiger's size problem turned critical: it was hard to find appealing companies where a fund that large could take meaningful stakes. He crammed capital into old-economy value stocks such as Federal-Mogul (auto parts) and Niagara Mohawk (a power utility).
US Airways: from investor to owner
The airline summed up the trap.
- The good call. In early 1996 US Airways appointed a cost-cutting turnaround chief and Robertson wisely bought a large stake; by summer 1998 the stock had quintupled.
- The refusal to sell. The stake was now worth about $1.5 billion, roughly one fifth of the company. When the airline ran a share buyback in early 1998, Tiger's airline analyst urged selling part of it into the buyback; Robertson refused, calling it one of his best ideas. With few opportunities big enough for a supersized value fund, he held on long after the thesis had paid off and the position had become illiquid.
- The slide. At the start of 1999 disappointing profits sent the stock down 29% in three weeks. Robertson predicted to investors it could triple by year-end as old-economy companies regained respect. Instead machinists rejected a pay contract in July and work stoppages forced cancellations in August. Unable to sell without crashing the price, he had become an owner rather than an investor, reduced to hunting for a strategic buyer to take the block.
- The rest of the book. Federal-Mogul fell 30% in the first half of 1999 and Niagara Mohawk was flat. In April Robertson told investors that fundamentals were widely being ignored.
The summer of withdrawals
By mid-1999 the strategy had become a crisis at Tiger's Park Avenue office. Tiger lost 7.3% in the first half while tech-heavy mutual funds were up a quarter or more and kitchen-table day traders were beating Robertson's special forces. On top of the yen losses, investors' patience snapped: a net $3 billion withdrawn in the six months to March, another $760 million at the end of June. Every withdrawal forced more selling, and once Wall Street saw Tiger as a forced seller the predators returned. Robertson listed his top ten positions in each monthly letter, so rivals knew exactly what to sell ahead of him. One day in June, a false rumour of $3 billion in month-end redemptions set off a predatory wave that cost Tiger $72 million in fifteen minutes while the broader market was steady.
Mallaby draws the parallel. LTCM's death spiral began when brokers called in loans; Tiger's began when investors called in equity. In both cases public knowledge of the holdings made it worse, a point he aims at commentators who assume that hedge-fund transparency would make markets more stable.
Druckenmiller rides the tiger
A mile north, near Central Park, Quantum fought a parallel battle with the opposite tactics. At the start of 1999 Druckenmiller shared Robertson's view that tech was overpriced but, undeterred by momentum, placed an unhedged outright short: $200 million spread across a dozen of the most overvalued start-ups. They shot up so violently there was no escape; he recalled stocks closing at 100 one day and opening at 140 the next. Within weeks the position cost $600 million, and by May 1999 Quantum was down 18%, facing his first seriously negative year.
The reversal of fortunes was stark. In 1998 Quantum had made 21% while Tiger lost 4%; now, having stood in front of the locomotive, Druckenmiller was losing twice as much as Robertson. In May the big feeder fund Haussmann Holdings cut its investment in Soros Fund Management by more than half, the press tracked a stream of departing analysts, commentators asked whether the Druckenmiller legend was finished, and rumours of a three-month medical leave gave way, once denied, to rumours that Soros was tiring of his alter ego.
Yet Druckenmiller was better built for the bubble than Robertson. Tiger's value tradition made buying into the mania almost unthinkable, whereas Druckenmiller blended fundamental analysis with charts and could follow fundamentals in one phase and surf the trend in the next.
- Switching sides. In May 1999 he handed capital to a new hire, Carson Levit, who loaded up on dot-com stocks: the sceptic who had shorted the bubble climbed aboard.
- The Sun Valley conversion. In July he attended the annual tech and media conference in Sun Valley, Idaho, a festival of new-economy buzz where Hollywood moguls, presidential hopefuls and Silicon Valley founders mingled and even Warren Buffett, battered by the bubble, strolled in a polo shirt chatting with Bill Gates and Michael Bloomberg. Druckenmiller came back a zealous convert, gave Levit more money and hired a second enthusiast, Diane Hakala.
- The radioactive book. The pair bought exactly the kind of stocks Druckenmiller had shorted earlier in the year, names he later admitted he could not even spell. VeriSign, Qualcomm and Gemstar became heroes of the new era and Quantum's results took off.
- Feeding on Tiger. Robertson's redemptions forced him to dump a large South Korea Telecom stake bought before the bubble; his selling knocked it down a third in July and August. Levit bought the liquidation cheaply and the stock tripled.
In the last months of 1999 Druckenmiller made more from surfing tech than from shorting sterling in 1992. Quantum went from -18% after five months to +35% for the year, one of the great comebacks in hedge-fund history. Mallaby's pointed observation: it had nothing to do with pushing markets toward efficiency.
The short sellers lose their shirts
Other hedge funds that tried to fight the mania were crushed, even when their analysis was right.
- America Online, early 1999. An army of short sellers attacked AOL for capitalising its marketing costs: $1 million spent winning subscribers was treated as an investment and expensed gradually, like plant and machinery. The shorts were correct that this was a low trick, and AOL eventually dropped it, but the stock kept climbing: they won the accounting argument and lost their money. Mary Meeker of Morgan Stanley allowed that investors raised on Graham and Dodd might struggle to value a company that would show losses under proper accounting; a banker at one new-economy gathering crowed that the Graham and Dodd crowd had shorted AOL and been crushed, and were now learning the new model.
- Chemdex, September 1999. Greenlight Capital's David Einhorn shorted this business-to-business chemicals exchange, which earned only a commission on trades but booked the full value of goods exchanged as revenue. He went short at $26. In December Meeker declared that Chemdex had what it takes, and by February the stock was $164, more than six times his entry. Einhorn was forced to cover with enormous losses; by late 2000 Chemdex traded at $2.
Despite recurring suspicion that hedge-fund short sellers manipulate markets, the 1999 market brushed them off like flies on an elephant. Mallaby locates the manipulation elsewhere: companies cooking their accounts, auditors looking away, and investment banks hyping the companies they took public. Seeing no hope of bucking the mania, most hedge funds jumped aboard, for three kinds of reason.
- Belief or style. Some believed the hype, or their method was simply to milk trends.
- Loose money. Some reasoned that with the Fed's monetary policy extraordinarily loose, the hot asset of the moment would reach stratospheric valuations.
- Insider IPO access. Some were privileged buyers of initial public offerings: banks stoked demand by allocating new issues cheaply, which virtually printed money for funds that got allocations.
- Mallaby's verdict. Hedge funds were no more prone to ride the bubble than other money managers, but no more contrarian either; only the bravest value investors were foolhardy enough to fight it.
The end of Tiger
The bravest and most foolhardy was Robertson. Tiger lost 17% in the third quarter of 1999 while momentum surfers made fortunes, and at the end-September redemption window a net $1.3 billion left. Assets fell from a peak of $21 billion in August 1998 to $9.5 billion a little over a year later, about $5 billion of the decline from clients voting with their wallets. The loss was personal: Tiger had been Robertson's family and network of friends and power brokers, his special-forces unit with team-building trips out west, and he was unused to defeat. In low moments he phoned his former lieutenant John Griffin, by then running his own fund, lamenting letters from investors in since 1982 who wanted their money back to buy a house, having lost faith in him.
- The lockup that wasn't. To slow the exodus Robertson cut redemption dates from four a year to two, a mild step next to later industry lockups. Loyal investors objected that after staying through a terrible year they should not now be chained to the Titanic; he conceded and backed down. The next generation learned the lesson and allowed withdrawals ever less often, to reduce the mismatch between redeemable capital and illiquid holdings.
- A sitting duck. Unable to lock in money, Tiger had to sell liquid or not, and rivals shorted in anticipation. Between August and October short interest in US Airways rose from 1.6 million to 3.8 million shares, knocking the battered stock down another tenth; between September and October short sales of Federal-Mogul jumped from 3.7 million to 6.5 million shares, and the stock fell 80%.
- The scale of the unwind. Over the year Tiger sold about $40 billion of stocks and $60 billion of other positions. COO Philip Duff asked how many institutions could survive a $100 billion balance-sheet downsizing; few, he thought.
By late 1999 Robertson's letters signalled the end. He argued that a great technology can change lives without enriching investors; noted that managers of his old-economy companies were buying back their shares while tech managers were eagerly selling theirs; and in December called the market a runaway tech frenzy amid a collapse in other stocks, a split he had never seen, insisting a correction would come. In January he blamed an irrational public that had driven unprofitable NASDAQ companies up about 38% in December alone.
The reckoning came almost on cue. The NASDAQ peaked on 10 March 2000, but by then Robertson was too battered to change course. On 30 March, with the index already 15% off its high, he told investors he was closing Tiger: rational valuation had given way to "mouse clicks and momentum," and he was tired of waiting for light at the end of the tunnel. Through redemptions, the market had stayed irrational longer than Tiger stayed solvent, exactly Keynes's warning.
| Tiger (Robertson) | Quantum (Druckenmiller) | |
|---|---|---|
| 1998 | Yen loss $3.1bn in October; -4% for year | +21% for year |
| Tech stance | Ignored tech, bought old-economy value | Shorted tech, then rode it |
| 1999 | -7.3% in H1, -17% in Q3 | -18% by May, +35% for year |
| Assets | $21bn (Aug 1998) to $9.5bn | $22bn (Aug 1998), down $7.6bn |
| Fatal flaw | Size, illiquid stakes, redeemable capital | Flipped direction once too often |
| End | Closed 30 March 2000 | He left, announced 28 April 2000 |
The end of Druckenmiller's Quantum
Quantum's late-1999 tech profits had spared Druckenmiller a losing year, but the victory was a nervous one. At weekly research meetings he fretted that the bubble could burst at any moment and the talk turned on how to anticipate it. The NASDAQ fell sharply in early January 2000, then vaulted to new highs, a hair-raising volatility. In February, watching biotech firm Celera Genomics rocket, he told a trader the market was insane, told Soros he was selling everything, and Soros said he was glad, having been uneasy himself. Druckenmiller dumped tech and returned to currencies, apparently out before the bust.
Then the markets taunted him. He got stuck on the wrong side of the new euro; the tech stocks he had sold kept soaring; and Levit and Hakala, still running their tech sub-portfolios, were making him look foolish. His fear shifted from being blown up to losing face, and, not for the first time, he turned on a dime: he bought all his tech back and gave the two room to run. Briefly the good times returned.
When the NASDAQ turned on 10 March, many of Quantum's stocks fell faster than the market. Druckenmiller had personally bought heavily into VeriSign, which lost almost half its value in a month and became hard to sell; he later said he knew he was dead. By the end of March Quantum had lost about a tenth of its capital. By pivoting aggressively one time too many, he failed to get out before the party ended.
He went on vacation to Florida. A year earlier he had led Quantum back from behind; now, drained like Robertson, he lacked the will, and he found himself envying Robertson's freedom despite the bad timing of Tiger's closure. He told his wife that money was meant to be enjoyed, and if he could not enjoy two weeks with his children there was no point. After twelve years carrying Quantum, he told Soros he was exhausted and could not fight out of another hole.
Soros recognised the desperation. Years earlier, running Quantum nearly alone, he had come to hate the job and likened the fund to a parasite swelling inside his body. Druckenmiller had long said Quantum was too large and he needed an exit; in the end, Soros reflected, he had freed himself by blowing up the fund, an expensive but effective escape.
Jul 1998 Robertson's letter announces his yen short; Tiger up 29% for H1.
21 Aug 1998 Yen bet turns as LTCM is hit; 7% yen rise and $1bn+ loss over the next month.
8 Oct 1998 Yen 130 to 112; Tiger counterattacks and the squeeze breaks.
May 1999 Quantum -18%; Druckenmiller funds Levit's dot-com book.
Jul 1999 Sun Valley conversion; Hakala hired; SK Telecom bought from Tiger's forced sales.
Sep 1999 Tiger loses 17% in Q3 and $1.3bn in redemptions.
Feb 2000 Watching Celera, Druckenmiller sells all tech; then buys it back.
10 Mar 2000 NASDAQ peaks; VeriSign nearly halves in a month.
30 Mar 2000 Robertson closes Tiger, NASDAQ 15% off its peak.
28 Apr 2000 Soros announces Quantum -21% for the year and Druckenmiller's exit.
On 28 April 2000 Soros held a press conference. Quantum was down 21% for the year; assets at Soros Fund Management had fallen $7.6 billion from their high-water mark of $22 billion in August 1998. Druckenmiller was leaving after a dozen years, and Quantum would become a sedate, low-risk endowment. Within a single month the two largest and most storied hedge funds had shut their doors. Soros declared that a large hedge fund like Quantum was "no longer the best way to manage money," because markets had become too unstable. He seemed to be writing the epitaph of an industry he had helped invent; Mallaby's closing judgment is that he could not have been more wrong.
Mallaby's own Appendix I puts Tiger's net return at about 31.7% a year from 1980 to its August 1998 peak, so the closure ended an outstanding record rather than exposing a weak one. His case that hedge funds were "no more contrarian" rests on a few vivid cases (Tiger, Quantum, the AOL shorts, Greenlight); it is supported by Brunnermeier and Nagel (Journal of Finance, 2004), who found hedge funds as a group overweighted tech during 1998-2000 and cut exposure before the price peak, riding rather than fighting the bubble. His transparency argument concerns public disclosure of holdings; it does not settle the case for confidential reporting to regulators.
Druckenmiller kept running his own Duquesne Capital and in August 2010 returned outside capital, turning it into a family office after about three decades without a losing year. Julian Robertson died on 23 August 2022, aged 90, having seeded dozens of "Tiger cub" funds that dominated long/short equity investing. David Einhorn went on to make his name as a short seller with Allied Capital and Lehman Brothers.
The Yale Men
Soros's epitaph for hedge funds was premature because a new kind of money arrived: university endowments, led by David Swensen at Yale, who backed Tom Steyer's Farallon in January 1990 and built the "absolute return" allocation that most endowments copied. Hedge-fund assets rose from $490 billion in early 2000 to $1.1 trillion by end-2005, and event-driven funds that offered explicable, steady, uncorrelated returns became the industry's new face. Mallaby's warning is that those smooth returns came partly from buying illiquid assets, whose risk is understated in good times and concentrated in a crisis.
Robin Hood and the real philanthropy
On 1 June 2001, 2,920 guests filled the Jacob Javits Convention Center for the annual gala of the Robin Hood Foundation, the charity Paul Tudor Jones had conceived after the 1987 crash. In thirteen years it had given away over $90 million to New York groups fighting poverty, teen pregnancy and illiteracy. The cocktail area mixed disco with rain forest (green poles nearly twenty feet tall, projected treetops, a roving green light hunting celebrities such as Meg Ryan, Keith Hernandez and Tom Brokaw), and caterers were marshalled by traffic cops waving fluorescent batons. Above all, the room was full of hedge-fund managers.
The evening proved, Mallaby says, that Soros's epitaph had come too early. Stan Druckenmiller, back in the markets almost at once after leaving Quantum and now running his own Duquesne Capital Management, was a sponsor. Tickets cost $5,000 and the auction was lavish:
- Lunch with a mogul. One guest paid $540,000 to lunch with a leading financier.
- "Be a Star." Another paid $260,000 for walk-on parts in A Beautiful Mind and The Duplex plus dinner with Catherine Zeta-Jones and Michael Douglas.
- Yoga. Sonia Jones, Paul's wife, bid $420,000 for a yoga class taught by Madonna and Gwyneth Paltrow, auctioned by Jerry Seinfeld.
- The finale. A wall dropped away, cannons fired confetti, and Robert Plant of Led Zeppelin took the dance floor.
The night raised $13.5 million, proof that hedge-fund wealth had become a social force. Soros's Open Society Institute, the oldest and largest hedge-fund philanthropy, was disbursing $450 million a year, and in 2002 Arki Busson, who channelled European money to Jones and others, founded ARK (Absolute Return for Kids) as London's Robin Hood. Yet Mallaby's point is that the bigger philanthropic effect came not from galas but from profits accruing to endowments that invested in hedge funds, funding research and scholarships. The traffic ran both ways: institutional money made hedge funds larger, slicker and more methodical, turning them into a real industry.
Swensen: the moralist who chose Yale
David Swensen, tall, angular, ascetic and cerebral (a "stiff-backed midwesterner" to one friend), was driven by moral purpose. Raised in River Falls, Wisconsin, he started a church-group recycling club; his mother and sister were Lutheran ministers; he hoped to enter the Senate in the state's progressive tradition. In Yale's economics PhD programme he befriended future Nobel laureate James Tobin and got to know Salomon Brothers, which supplied data for his dissertation, and he fell for finance and for Yale.
1980 Completes his PhD; hired at once by Salomon Brothers.
1981 Helps create the first currency swap, between IBM and the World Bank, letting IBM hedge its Swiss franc and deutsche mark exposure.
1982 Lured to Lehman Brothers to run its fledgling swaps desk.
1985 Accepts his old professors' plea to run Yale's troubled endowment, taking an 80% pay cut for a book-lined campus office.
1987 Turns away Steyer's first pitch.
January 1990 Yale invests $300 million in Farallon.
Asked years later by a Wall Street admirer why he had not become a hedge-fund billionaire, Swensen blamed "a genetic defect."
Diversification, the free lunch
Swensen inherited a conventional portfolio: over four fifths in US stocks, bonds and cash and one tenth in alternatives. To a man fresh from Wall Street innovation it seemed tame, and it slighted Tobin's teaching that diversification is the one free lunch in economics. In a modern system, Swensen reasoned, diversification should reach beyond US bonds and equities to foreign equities, real estate, private equity, oil, gas and timber, each offering equity-like returns with different risks. He added a further class he called "absolute return", a term that entered the investment vocabulary as a synonym for hedge funds.
- Performance fees over asset fees. Returns get harder to earn as a fund grows, so pay tied to asset size is perverse; he preferred the performance fees that supplied most hedge-fund revenue.
- Skin in the game. Performance fees alone invite recklessness (managers take a fifth of the upside with no matching penalty), so he sought managers with their own savings in the fund, and found many.
- High-water mark. A manager who loses money takes no further fees until investors are made whole.
- Uncorrelated equity-like returns. The real prize: returns as large as stocks' but independent of the index, the free lunch of diversification.
- A screw loose. Swensen later wrote that great investors play "for sport" rather than profit; he wanted integrity plus obsession.
The moralist in him had no wish to enrich managers; the economist was impressed by the incentive design. Still, it took time for him to see the potential. In 1987 a Yale alumnus who had heard of his appointment at a homecoming football game came to pitch a small West Coast fund. Swensen and his deputy Dean Takahashi told him Yale was not interested and never would be.
Tom Steyer and Farallon
The supplicant was Tom Steyer, founder of Farallon. He had a sweeping red-blond parting, sideburns and woven wristbands; New York-raised but steeped in San Francisco's vibe; ebullient, funny, a former college soccer star who ran with ideas. He also had an acute sense of right and wrong that showed in his habits: even after Farallon became one of the world's biggest funds he drove a beat-up car, flew commercial, ignored fashion, and sat at a desk in an open-plan hallway with a breathtaking city panorama behind him and the blinds down.
He founded Farallon in 1985, the year Swensen took over at Yale, partly to escape Wall Street and partly from that sense of justice. Mallaby traces a three-step logic in his career:
- Morgan Stanley. As a young analyst he was upset that bank advisers could be paid for being wrong: sounding convincing mattered more than being right.
- Goldman Sachs. After Stanford business school he joined the merger-arbitrage unit of Robert Rubin, the future Treasury secretary, where the firm was paid only when right, though individual credit was sometimes disputed.
- Farallon. A freestanding fund removed every buffer between the quality of his calls and his rewards.
How merger arbitrage works
He rented cheap space downtown with two desks (one for a partner) and rode the elevator up at 5:30 a.m. with coffee and a doughnut to catch the New York open. His method was Goldman's. When a bid is announced, the target moves most of the way to the offer.
A target trading at $30 gets a $40 bid and jumps to $38. Buying risks $8 (if the deal collapses and the stock falls back) to make $2 (if it completes). The skill lies in judging whether antitrust regulators will block it, whether shareholders will revolt, and whether a rival suitor may push the price above $40. Steyer bought targets in deals likely to close and shorted the acquirers to hedge out general market moves, eking out profits month by month.
His intensity could seem excessive: after losses in the 1987 crash he started arriving at 3 a.m., accompanied by his wife, who feared for his stability. But he did extremely well. An arbitrageur could analyze deals as well from San Francisco as from Goldman, Mallaby notes, especially with body and soul tied to performance.
From mergers to distressed debt
Late in the decade Steyer broadened out, partly to survive (the takeover boom halted when junk bonds collapsed in 1989, leaving few mergers) and partly on offense. Junk issuers were going bankrupt one by one, and an investor who could tell which busted debt to buy stood to profit. Better still, pension funds and mutual funds were forced sellers: their rules barred holding bonds of defaulted companies, so they conceded bargains to nimble buyers. When Drexel Burnham Lambert, kingpin of junk, went bankrupt in 1990, Steyer bought a large slice of its debt at cents on the dollar and sold in 1993 for a 35% profit: a double, having profited first from Drexel's mergers and then from its implosion.
Steyer had built what would be called an event-driven fund. Before a takeover bid, a share price is efficient, analyzed to death; the bid scrambles the consensus, forcing fresh judgments of the premium, the time to completion and the discount rate. A bankruptcy does the same to bond values. Profits come where settled prices suddenly stop being right, often at the expense of noneconomic sellers bound by rules.
Winning Yale's money
Even before Drexel, word of Steyer's results reached Swensen. Farallon made money whether markets rose or fell, offering diversification, and its profits came from moments when prices were scrambled, so to an economist aware of the limits of the efficient-market hypothesis they did not look like luck. Yale's due diligence then confirmed the man:
- No craving for luxury. One look at the office settled that.
- Pure incentives. Steyer insisted that employees keep their liquid savings in the fund, so they felt any loss.
- High-water mark. If the fund fell, he would take no fees until he earned the money back.
In the fall of 1989 Swensen flew to San Francisco, approved of the scruffy office, and over a cheap lunch with Steyer and partner Fleur Fairman still said no. Hedge funds, he argued, would stiff clients after losses: rather than work for free to recover, they would close and reopen under a new name with new investors. Fairman beat Steyer to the reply, furious, calling the claim rubbish and telling him that if he thought that of them they did not want his money anyway. It was the best possible answer: someone who took her decency that seriously had the integrity Swensen sought.
In January 1990 Yale put $300 million into Farallon, lifting it to $900 million and starting a gradual change in the social impact of hedge funds.
The endowment model takes over
The deal repositioned Yale and, in time, nearly every endowment. Before Farallon, Yale held a smattering of private equity and real assets but nothing in hedge funds.
- 0% to 21%. Yale's hedge-fund allocation, 1990 to 1995, plus 31% in private equity and real assets.
- 0% to 7%. A typical university endowment's hedge-fund allocation, 1990 to 2000.
- July 2000 to June 2003. S&P 500 -33%, HFR hedge-fund index +10%, Yale +20%.
- $7.8bn of $14bn. The share of Yale's endowment credited, at the twentieth anniversary of Swensen's arrival, to his outperformance of the average university fund.
- About half. Share of hedge-fund capital coming from institutions rather than individuals by 2009.
- $490bn to $1.1trn. Hedge-fund assets from the start of 2000, when Soros pronounced the era over, to the end of 2005.
With that $7.8 billion, Mallaby writes, Swensen eclipsed storied education philanthropists such as Harkness and Mellon, and hedge funds became more than vehicles for the rich to get richer. Soros's epitaph did fit his own kind of trader: the early 2000s were lean for macro funds. But event-driven funds more than made up for it.
The Rubin three
Farallon grew from $8 billion in 2002 to $16 billion in 2006, and imitators from Rubin's Goldman arbitrage group crowded in. Together they overtook the Commodities Corporation lineage (Chapter 3).
| Group | Funds | Assets |
|---|---|---|
| Rubin three (2002 to 2006) | Farallon $8bn to $16bn; Och-Ziff $6bn to $14bn; Perry Capital $4bn to $11bn | $41bn in 2006 |
| Commodities Corp three (2006) | Caxton, Tudor, Moore | $35bn, $6bn less |
| Other Rubin alumni | Frank Brosens (Taconic Capital), Eric Mindich (Eton Park), Edward Lampert (ESL Investments) | Flourishing |
In the hedge-fund family tree, Mallaby judges, perhaps only Julian Robertson had more offspring than Rubin's arbitrage desk.
Why committees loved event-driven funds
Returns alone did not explain the rush. Endowment managers answer to oversight committees that ask skeptical questions, and event-driven returns were pleasingly explicable.
Macro. Paul Tudor Jones might talk of Kondratiev waves and breakout points, which to the average committee sounded like hocus-pocus. A rich individual may bet his fortune on a mysterious genius if he likes.
Event-driven. Farallon's people studied legal labyrinths: the odds a merger would close, how a slice of subordinated debt would fare before a particular bankruptcy judge in a particular court. Committees could also see that profits came from others being hobbled by rules. A committee needs to protect its back with a PowerPoint.
The more endowments displaced rich individuals as the main investors, the more it mattered that strategies could be understood. Event-driven funds also promised consistency. They used very little leverage, a selling point after LTCM, and their returns were almost miraculously steady: Farallon had no losing month between 1990 and 1997, giving it a Sharpe ratio (return adjusted for risk) roughly three times the stock market's. Through the dot-com madness Steyer sailed serenely: he neither rode the bubble like Druckenmiller nor was run over by it like Robertson, but analyzed the era's epic takeover battles with market risk hedged out, which looked very good when the crash sank both men.
Understandable, low-volatility, uncorrelated returns reflecting pure skill: this was the holy grail that endowment consultants called alpha, and capital flooded in. Yet even stars like Farallon, Mallaby warns, had vulnerabilities few suspected.
Off piste: London, Argentina, Indonesia
By the late 1990s Farallon worked from a fashionable downtown San Francisco skyscraper (Steyer's space still modest, a Henry Moore sculpture outside, beautiful people eating organic sandwiches on the lawn) and ranged ever further:
- London, 1998. A merger-arb operation, opened within months of rivals Och-Ziff and Perry Capital.
- Alpargatas. A stake in the bankrupt Argentine textile and shoe maker; Farallon installed new managers, restructured the debt, and some 2,000 idled workers were re-employed, proof it could do well by doing good in a frontier economy.
- Indonesia, November 2001. Farallon set out to buy the country's biggest bank.
Liem's crony empire
Bank Central Asia (BCA) had been founded by Liem Sioe Liong, once Indonesia's richest man and a firm friend of the dictator Suharto. His empire was said to account for 5% of national output, and flour showed how. The government sold Liem imported wheat at a subsidized price and bought it back from his mills at a markup. Free of competition, his Jakarta mill became the world's largest and his Surabaya mill the second-largest, and much flour meant for the state found its way to his Indofood, which held 90% of the instant-noodle market. He prospered likewise in coffee, sugar, rubber, cement, rice and cloves, and naturally owned the nation's largest bank.
Liem had hedged political risk by giving Suharto relatives big stakes in his firms. But the 1997 currency crisis, the one that cost Soros and Druckenmiller a fortune, set off a slow revolution that toppled Suharto, and Liem's insurance became a target on his chest. Rioters broke into his compound, burned his cars and smashed his Chinese vases. His businesses went bust; since many had borrowed from BCA, they threatened to drag it down, and the government rescued the bank to stop a depositor panic.
The noneconomic seller
Suharto's fall was more extreme than any takeover bid: millions pushed into poverty, thousands of demonstrators killed, hundreds of businesses looted, and many Indonesians blaming Western hedge funds, with some American financiers receiving death threats. Yet the more Farallon looked, the better the opportunity seemed. The government was the classic noneconomic seller: the IMF was pressing it to offload rescued companies at almost any price, and because most investors would not set foot in the country, competition would be thin. Mallaby's symmetry: in 1997 hedge funds profited by betting against governments that set illogically high prices for their currencies; in the hangover they profited from governments setting illogically low prices for their economies' broken jewels.
By fall 2001 Farallon held about $1 billion in Indonesia: stakes in PT Semen Cibinong (the third-largest cement maker) and PT Astra International (the largest automaker), and the Jakarta Container Port Terminal, bought and sold on to Hong Kong's Hutchison. Then a government contact told Farallon's point man Ray Zage that BCA would soon be reprivatized: would Farallon like to bid?
The mismatch was astonishing: a few dozen employees taking over a bank with eight million accounts and eight hundred branches in the world's largest Muslim country, Goldman arb culture plus California cool buying the embodiment of crony capitalism. 9/11 made it riskier still, since a country torn by collapse and revolution seemed open to Islamist extremism. CalPERS was about to announce it would not invest in Indonesia at all, and even Goldman tightened its Indonesian exposure limits. Farallon's team changed how it travelled, avoiding clumps of foreigners; Zage regarded the airport stretch between customs and the taxi rank as a natural kill zone. Andrew Spokes, a dapper English banker recruited from Goldman Sachs Hong Kong, later conceded while inspecting his cuffs that Farallon had been "a little off piste", sounding to Mallaby like a vintage James Bond skiing an avalanche in a tuxedo.
Spokes's valuation
Event-driven discipline means ignoring chatter and panic and, when market prices stop being a guide, valuing an asset by its cash flows. Spokes looked past BCA's crony reputation to three facts, plus a political argument.
- A bond fund. Since nationalization, the rotten Liem loans had been swapped for recapitalization bonds, so BCA depended on the Indonesian state, not busted crony firms.
- With cheap leverage. Unlike other bond funds, it came with low-cost retail deposits.
- Plus an option. If the economy revived, BCA could start making profitable business loans: a free option on recovery.
- Political risk. Precisely because the world saw Indonesia as scary, the post-Suharto leadership could not afford to cheat a foreign investor in a high-profile deal; its reputation would be ruined.
After spirited debate the partners agreed. With barely a year's track record in the country, Farallon bid $531 million in late 2001 against the Standard Chartered consortium, and in March 2002 the government announced it had won: a hedge fund from latte land controlled the nation's top bank. Conspiracy theories bloomed: was it a front for the US government, or a Trojan horse for Liem hoping to revive his empire?
Creating an event
Mallaby calls the deal a blessing for Indonesia. Farallon installed a new chairman, brought in consultants and patiently coaxed BCA out of the Suharto era. By 2006, when it sold most of the stake to an Indonesian partner, BCA's share price was up 550%. The spectacle also made institutions in New York and London look again: net portfolio inflows rose from $286 million in the year before the purchase to almost $1 billion the year after and more than $4 billion the year after that. Farallon had scrambled the settled view on all Indonesian assets. An event-driven fund had created an event, helping turn the tide for a nation of 240 million.
Water in the San Luis Valley
Meanwhile a venture in southern Colorado went badly. Rancher Gary Boyce proposed buying land in the San Luis Valley and pumping the aquifer to supply Boulder, Colorado Springs and even Denver; Farallon, alert to real assets through its Yale ties, invested. The valley was as remote as Jakarta's back alleys: a flight to Denver, four hours' drive to Alamosa, then scrub plains to Boyce's hollow-walled adobe ranch house, where he wore mother-of-pearl shirt studs and kept ornate pistols on his desk.
He looked the perfect local partner. Raised poor in the valley, he had won the Colorado dirt-biking championship three times, fought an earlier aquifer scheme by founding a newspaper called the Needle, and grown rich training horses for upper-crust Virginians and richer by marrying an MGM heiress. Steyer formed a partnership: half the capital from Farallon, half from Yale, with Yale playing no management role. The project then unfolded step by step:
- 1994. Boyce buys a valley ranch, outbidding the Nature Conservancy, which wanted a national park.
- $3 million. An environmental study showing water could be pumped without harming local soils; lobbyists worked the state legislature.
- $500,000. Signature-gathering for two ballot measures: meters on farmers' wells and user fees for some water, both essential to price what Boyce would sell.
- $400,000. Advertising for the initiatives, which Boyce assured his partners would pass.
- November 1998. Both measures defeated by a large margin, after four years and more than $20 million.
Farmers had revolted against the fees and claimed sandy soil would clog the meters. Steyer, who had visited with his mother (she bonded with Boyce and tried elk hunting), began to doubt his partner: being born local is not being respected locally, and the shirt studs and pistols may have marked Boyce as a poseur.
A blocked exit and a senator
Farallon invited the Nature Conservancy to revive its park plan and signed a deal at the end of 2001, but Boyce sued, claiming the scheme was still viable and the early exit hurt his stake. The delay invited mischief. Colorado senator Wayne Allard accused Yale of profiting at taxpayers' expense (they would fund the Conservancy's purchase), demanded it cut its $31.3 million asking price although Yale was a passive investor, and suggested Farallon had misled Yale on environmental costs, though the referenda had failed over fees and meters, not any flaw in the study. A fund that had always avoided headlines was now being abused by a senator.
The unFarallon campaign
Farallon won the lawsuit early in 2004, and Yale pledged $1.5 million to ease the cost to Colorado taxpayers. Then came a strange coalition on several campuses, the "unFarallon campaign", demanding endowments pull their money. A protest soccer game at the University of Texas featured players dressed as crony capitalists; a "transparency fairy" in a feathered mask waved her wand outside Swensen's office. Its website listed charges:
- Coal. A coal-fired power project in Indonesia.
- Argentina. Alleged suffering of Alpargatas workers.
- Salamanders. The tiger salamander on a California golf course Farallon had invested in, threatened unless ponds were dug.
- Iraq. A $3 million stake in Halliburton, Dick Cheney's old firm.
Activists demanded a meeting on "the ethics of Farallon's investment practices." Steyer wrote to them about his business values and told his investors the site was factually inaccurate, but protests went on. In April students rallied before the Yale president's office, staged a mock aquifer extraction and broke ground for a mock coal plant, then turned up at the Advisory Committee on Investment Responsibility. Swensen, in a fleece vest, confronted them afterwards: investors must guard proprietary secrets, and to reap hedge-fund benefits Yale had to be "the highest-quality limited partner possible." A student replied that Yale should be the highest-quality global citizen.
Mallaby finds the targets ill chosen: Steyer was an open-handed backer of Democrat John Kerry, and Swensen had forgone billions through his "genetic defect." But having grown on college money, hedge funds could not expect immunity from college politics.
The hidden risk of illiquidity
Farallon closed the sale to the Nature Conservancy in September 2004; the water plan had failed, but the land had appreciated, leaving a small profit. Mallaby draws three lessons for bargain-hunting funds.
- Expertise lags capital. Bargains hide in quirky places (junk debris, post-crisis Indonesia, a feud among Colorado farmers), and surveying the traps takes people. More time in the valley might have exposed Boyce; but a fund doubling every four or five years cannot grow in-house expertise as fast as capital arrives.
- Illiquid bargains, costly mistakes. Busted bonds are cheap because most investors fear them, which also makes them hard to exit; the same holds for an Indonesian bank. Liquid markets are efficient and hard to beat; illiquid ones offer bargains but punish errors.
- Illiquid assets flatter results. Without constant trading there is no objective price, so funds estimate rather than report profits. A few may exaggerate, though not sustainably; even honest funds smooth: an asset valued every few weeks that rises 5% and falls back is booked as flat, so volatility goes unrecorded and risk-adjusted returns look better than they are.
The deepest danger is that illiquid assets collapse hardest in a crisis, when investors crave anything sellable and shun the rest. LTCM's seemingly diverse book concealed a single bet on a stable world: unrelated positions fell together because many were attempts to harvest a premium for holding illiquid assets. Diversified event-driven funds may hide the same concentrated bet. LTCM paid in 1998; buyers of illiquid assets would pay heavily again in 2008.
Smoothing means reported Sharpe ratios from hard-to-mark assets overstate quality even without dishonest marks. Swensen's $7.8 billion figure was Yale's own celebration, measured against peer endowments rather than adjusted for the extra illiquidity risk. Mallaby's admiring portrait also predates the endowment model's liquidity squeeze of 2008-09, when Yale's returns fell by about a quarter, and its weaker relative record in the 2010s. The Indonesian story is told largely through Farallon's own participants (Spokes), and the inflow numbers show correlation with, not proof of, Farallon's catalytic effect.
Banks and pension funds reviewing alternative managers should de-smooth reported returns (unsmoothing autocorrelated series raises measured volatility), stress the liquidity mismatch between redemption terms and asset exits, and treat high Sharpe ratios from hard-to-mark assets as a warning, not a selling point.
Swensen died on 5 May 2021, aged 67, after growing Yale's endowment from about $1 billion to over $31 billion. In 2025 Yale moved to sell several billion dollars of private-equity fund stakes on the secondary market amid federal funding pressure, a test of the liquidity built into his model. Steyer left Farallon in 2012 for climate activism and philanthropy ran for the Democratic presidential nomination in 2020, and spent about $215 million on a bid for California governor that failed to advance from the June 2026 primary.
The Code Breakers
Renaissance Technologies, a quant shop in East Setauket, Long Island, ran perhaps the most successful hedge fund ever: Medallion returned 39% a year on average from the end of 1989 to 2006, including 71% in 1994 and 80% after fees in 2008. Mallaby sets James Simons's code-breaking method against the more theory-minded quants at D. E. Shaw and the trader-mimicking machine at Tudor. Renaissance's edge came from letting computers find nonintuitive patterns the way cryptographers and statistical translators do, inside an open, collaborative research machine that rivals could copy in pieces but never catch.
Simons: four lives in one
East Setauket is an hour's drive from Manhattan on the Long Island Expressway, cut off from the Greenwich hedge-fund cluster by a wedge of the Atlantic. From about the time Swensen backed Farallon, the township's Renaissance Technologies coined money, and by the mid-2000s its founder James Simons was the industry's top earner: not the most famous billionaire, Mallaby says, but probably the cleverest. His success came from combining four passions.
- Speculator. He dabbled in commodities from student days, and used a wedding gift to speculate successfully on soybeans.
- Entrepreneur. He launched a string of businesses; the name Renaissance Technologies reflects roots in high-tech venture capital. After riding a Lambretta scooter from Boston to Bogotá in his twenties, he co-founded a tile factory in Colombia.
- Code breaker. At the Pentagon's secretive Institute for Defense Analyses (IDA) he learned to build a research organization closed to outsiders but collaborative inside; IDA fired him for opposing the Vietnam War.
- Mathematician. He co-created Chern-Simons theory and won the American Mathematical Society's Oswald Veblen Prize, geometry's highest honour. He chaired Stony Brook's math department until quitting in 1977 for business.
Recruiting the ghost hunters
His early trading had little to do with maths: he rode the wild 1970s commodity booms and busts on hunches about supply and demand. But the mathematician in him wanted models instead of seat-of-the-pants judgment, ideally a machine that traded for him. From the late 1970s he drew in outstanding minds, through ventures whose names and ownership kept shifting (a fund in Bermuda, a company on the West Coast, the Long Island operation):
- Leonard Baum. An IDA cryptographer and preeminent modeller.
- James Ax. Winner of the AMS's top prize in number theory.
- Elwyn Berlekamp. A Berkeley mathematician and another IDA veteran.
Their military experience was directly relevant. Berlekamp had worked on battlefield systems that send signals like "ghosts": faint traces of code in seas of noise, so wispy that the enemy not only cannot decode them but does not suspect any transmission. To him, economists who declared markets random resembled those fooled adversaries: they had stared at the ghosts and seen nothing. The Simons team turned code-breaking algorithms on market data, a league economists could not play in because they lacked the specialized maths.
Medallion's kernel
The first efforts were only moderately successful. Baum tired of the hunt for golden algorithms, read the business papers and made a huge, handsomely paid discretionary bet on the British pound. Ax stuck with computer trading, but he was volatile and so were his returns. Still, by 1988 the platform was built: Simons and Ax launched the Medallion Fund, named for the medals they had won in geometry and number theory. It traded commodity and financial futures on computer signals, and its heart was unremarkable, a trend-following model much like Commodities Corporation's of a decade before. But a small slice of capital followed different rules, and that was the kernel of the Simons fortune.
The kernel came from Henry Laufer of Stony Brook's maths faculty, a figure so self-contained that when Ax punished him after an argument by refusing to speak to him for months, Laufer did not notice. In the mid-1980s he spotted patterns in how markets move right after an event perturbs them. After a data release a commodity or currency would jiggle up and down as investors reacted; to the naked eye this looked random, but with high-resolution statistical goggles patterns appeared. Not the same reaction every time (that would be too obvious), but across thousands of reactions to thousands of events, certain sequences showed up in slightly more than half of the observations. Bet on them repeatedly, often enough and big enough, and profits were assured.
- More data. Short horizons yield thousands of observations, so faint patterns can reach statistical significance.
- Faster compounding. A Tiger investor hopes a stock doubles in two years; a statistical trader who makes a quarter of a percent in twenty-four hours ends up much richer.
- More robust forecasts. Less time for unforeseen factors to knock a prediction off target, so confidence is higher.
- Safe leverage. Because the predictions were relatively robust, the team could leverage its bets and magnify profits.
The algorithms remain secret, Mallaby stresses, but these reasons for their discovery and their profitability can be roughly understood.
Crisis and the study period
At launch about 15% of Medallion's capital ran on the short-term signals and the rest on trend following. It started well, then nosedived: by May 1989 it was down almost a quarter from its peak, and Simons suspended trading. Ax insisted the model would soon recover; Simons was so sure he was wrong that he ended the partnership and, with Berlekamp and Laufer, began a "study period". The verdict: trend following had run out of juice. Too many Commodities Corporation imitators had crowded in, and brokers such as Dean Witter were selling dozens of commodity funds; trend following had become trendy. After months of deliberation, Laufer's short-term signals became the new heart of the system. In 1990, the first full year after the relaunch, Medallion made 56% after fees.
Berlekamp responded by cashing out, selling his share of the management company and returning to Berkeley. Simons, who for over a decade had charmed a shifting cast of mathematicians on the belief that cryptographers' methods would crack the market's code, felt vindicated and pressed his advantage. He bought Berlekamp's share, rolled the company into Renaissance, and used Medallion's profits to hire more mathematicians, installed in the Long Island High Technology Incubator near the Stony Brook hospital. The research paid off: patterns that worked in American commodities often worked in foreign markets, and after some setbacks the methods found patterns in equities.
Scientists, not economists
Simons added computer scientists, physicists and astronomers but never hired economists: he wanted people who saw markets as a mathematical puzzle, detached from the flesh, blood and bricks of the real economy. The abstraction had comic side effects:
- Alphabetical days. One researcher presented a week's results in the order Friday, Monday, Thursday, Tuesday, Wednesday, assuming colleagues would find computer sorting natural.
- The seating algorithm. For a dinner for 500 investors, a scientist helped Simons assign probabilities to who would get along with whom (the blackboard estimated the odds of a single female algebraic geometer hitting it off with a married male judo instructor). The program seated a long-time investor next to a woman who had sued him for sexual harassment.
Mostly the approach was gloriously successful. Simons invested heavily in computers fed with every kind of data: market prices, economic releases, newswires, even weather. One simple find: fine morning weather in a city tended to predict a rise in its stock exchange, so buying on bright days at breakfast and selling later would win, except that the effect was too small to beat transaction costs, which is why Renaissance let it be known. Most patterns were individually modest, since markets are roughly efficient, but blending many small inefficiencies into one program produced profits year after year, especially in turbulence.
- 39% a year on average for Medallion, end-1989 to 2006.
- 56% after fees in 1990, the first full year after the relaunch.
- 71% after fees in 1994, the year Steinhardt lost billions in the bond meltdown.
- 80% after fees in 2008, almost 160% before them.
- $1.5 billion. Simons's reported personal earnings in 2006, equal to the combined profits of Starbucks (115,000 staff) and Costco (118,000).
- 1993. Medallion closed to new outsiders; by the 2000s its roughly $6 billion was almost all employees' money.
When Simons retired in 2009 he was a billionaire many times over, a comb-over of white hair and grizzled beard on magazine covers. To rivals' astonishment, Medallion's magic resisted competition through the 1990s and 2000s and, as Mallaby wrote in early 2010, showed no sign of fading.
D. E. Shaw: the computer scientist
The first challenger was David Shaw, a Columbia computer scientist, who founded D. E. Shaw in 1988, the year Medallion started trading. Like Simons he worked on fairly short time scales, hired mathematicians and scientists rather than traders and economists, and pursued precision zealously: staff learned not to say a programming task would take three to eight weeks, but 5.25 weeks, with an error of two weeks. The differences, however, proved significant.
Morgan Stanley's pairs trading
Shaw entered finance in 1986, hired by Morgan Stanley's proprietary desk to build systems for its secretive Analytical Proprietary Trading unit. That unit aimed to beat Steinhardt at his 1970s game: a pension fund's big sell order can push a stock out of line, and if the sale carries no information (the seller simply needs cash), the buyer profits when the price reverts. To spot uninformed moves, Morgan's quants paired stocks: Ford with GM, American Airlines with United, International Paper with Georgia-Pacific. If one fell while its twin held steady, a block seller was probably pushing it, and the price would soon revert. The method was fallible, but it only had to be right more than half the time.
Having mastered pairs, Morgan spent on all manner of research: physicists applying chaos theory, mathematicians with complex differential equations, and, one veteran recalled, systems using 3-D glasses to hunt price patterns. Shaw thought Morgan was ignoring avenues a university computer-science team would take, both in finding anomalies and in combining them into models. Not knowing exactly where the hunch led, he quit, rented an office above a communist bookshop in Greenwich Village, and launched his firm.
Fast success and a casual culture
Within six months he had promising anomalies where Morgan, searching for complex nonlinear patterns, had found little. Like Simons's ghosts, Shaw's patterns defied familiar explanation: the squiggles on the printouts had no names. They were so unintuitive that he needed no high-speed trading, confident nobody else would find them. Morgan's frustrated bosses closed their unit; Shaw moved to a Flatiron District loft in 1989 and a futuristic tower on West 45th Street in 1991. A magazine writer visiting in 1994 found 135 employees, up to 5% of daily NYSE turnover, a casual, faintly Bohemian air, and staff rolling out sleeping bags; it was easier to focus if you did not go home, said a young employee named Jeffrey Bezos, later founder of Amazon.
What scientists do differently
Economists start from perfect arbitrage: if two securities are theoretically the same, they should converge. Shaw's scientists did not look for relationships that ought to exist but asked what relationships did exist in the data. And their data were painstakingly swept for errors, cleaner than most finance professors had. Again and again an academic would bring Shaw a profitable anomaly that turned out to be misreported numbers: a strategy of buying stocks that suddenly cratered is worthless if a series reading $60, $61, $16 is really a typo.
Options, ventures and a bond-market bruise
Shaw then reached beyond stock prices:
- Options. Early pricing models, including those of the future LTCM laureates Robert Merton and Myron Scholes, assumed normally distributed price changes; 1987 showed that was dangerously wrong, since extreme moves were far more frequent. Shaw's mathematicians, better modellers than traders and better supplied with data than academics, built a superior options-pricing model, milked misalignments in equity derivatives, notably in Japan, moved into convertibles (bonds with options attached), and opened a market-making arm that handled half the trades in some segments.
- Scale. By 1995 the firm had more than 200 staff.
- Ventures. In 1995 it launched internet provider Juno Online and FarSight, an early online bank and brokerage.
- The third market. It entered off-exchange trading of listed stocks, dominated by a genial networker named Bernie Madoff; Madoff had ways of compensating for weak analysis, and Shaw's quants failed to turn a profit.
- Bank of America, 1997. An alliance to mine anomalies, including in bonds, overlapped with LTCM-style arbitrage. In 1998 Shaw was hit hard (one trader said it could have been the end), sold part of its book and took a loss that wiped out that year's gains in all other strategies combined. It left bond arbitrage, tiptoeing back by 2002.
Shaw's firm says some strategies earned Medallion-like returns of 40% plus, but Mallaby reasons that it cannot have put as much capital behind them, or its total returns would be higher. Its willingness to experiment diversified risk and some ventures paid handsomely, but it sometimes moved into already popular fields, risking crowded trades when markets turned.
Tudor: teaching a machine to trade like Jones
A computer-trading effort also grew in a surprising corner: Paul Tudor Jones's firm. Its first systems, named Madonna and Material Girl, were statistically crude and less exciting than their namesakes. In the early 1990s Jones grew up: he lowered his profile, married, settled in Greenwich, put away the Bruce Willis sneakers, and turned Tudor from a one-trader outfit into an institutional multi-manager platform, adding Boston stock picker James Pallotta and London-based Mark Heffernan, once called the greatest discretionary trader in the Goldman Sachs empire. The computing ambitions changed after Sushil Wadhwani arrived in 1995.
Wadhwani had taught economics and statistics at the London School of Economics and would later sit on the Bank of England's monetary policy committee, but he came via Goldman, where as a strategist he advised its prop traders and clients, Jones among them. That taught him the limits of pure economic thinking: spotting an illogical anomaly is only the start, because without a trigger (an upcoming election, a chart tipping point, a shift in institutional behaviour) it might persist indefinitely. Great discretionary traders, consciously or not, blended such inputs; his mission was a machine that mimicked their eclectic thinking.
- Start naive. A simple rule set, for example buy the index if economic indicators are positive, institutions sit on large cash piles, and options-market signals suggest sentiment is about to turn up.
- Interrogate the masters. Watch Jones and Heffernan trade and probe them (why put this on at ten o'clock, why add three hours later?). They used largely the same factors but combined them differently, so each answer refined the model.
- Size by z score. A longer-horizon system in liquid markets can build any position without moving prices, so the limit is risk appetite: too small leaves money on the table, too big risks insolvency. Each trade gets a z score, larger when the confidence of winning and the likely payoff are higher; betting big on the best chances can add more than raising the hit rate.
- Contrast with short-term systems. Medallion-type systems instead cap size by how much they can trade without moving the price against themselves.
- Shift weights by regime. In turbulence, sentiment (options data) matters more; in recession each bad number hurts more than the last.
- Switch strategies. Run LTCM-style arbitrage in calm times and trend following in storms, ideally switching automatically.
Wadhwani left for the Bank of England in 1999 before finishing. His first system, Techno-Fundamentals, a trend follower, ran money successfully from the end of 1997; the regime-switching design was left undone until he founded Wadhwani Asset Management in London in 2002. Tudor's program trading kept growing: by 2008 more than 50 people worked on it and algorithms ran more than $3 billion of Tudor's $17 billion.
Mallaby sees limits here too. A system built by a Goldman economist from the instincts of a Goldman trader came from the heart of the establishment, where others would hatch similar ideas: a handful of prized experts circulating among a handful of firms, each move cutting the odds that any firm's system was unique.
Simons's separate road
Simons hired established scientists, not Shaw's young quants and never Wall Street veterans; he limited cross-fertilization by staying on Long Island, far from New York, Greenwich and London; and he ignored academic finance (the faculty once met weekly to discuss journal articles, then dropped it as fruitless). Wadhwani, echoing the industry, called the Renaissance researchers "the gods of the business" and himself a mere mortal.
Brown, Mercer and statistical translation
In 1993 Simons added two IBM researchers who would drive Medallion for years and eventually run the firm. Peter Brown was a magnesium flare: five hours' sleep a night, riffs on every topic, and for a while a unicycle around the office. Robert Mercer was the calm half: an icy poker player who never recalled a nightmare, jokingly called an automaton by his IBM boss. They had dabbled in cryptography, but their real achievement was to upend machine translation.
Translation programmers had worked from the inside: know the grammar and syntax, teach the computer that la fille means "the girl" and les filles is the plural, as you would teach a middle schooler. Brown and Mercer spoke no French and ignored its rules. They took Canada's parliamentary records (Hansard), thousands of pages of paired French and English passages, fed them to an IBM workstation and told it to find the correlations. Unlike their later Renaissance work, this was published.
- Scrub the data. Just as price histories must be cleaned of "bad tics" ($16 reported for $61), the Hansard's misprints had to go.
- Co-occurrence. At the outset any English word might match any of the 58,000 French words in the sample; checking which words appeared in which twinned passages eliminated nearly 99% of the uncertainty at once.
- Subtler tests. For instance, assume an English word most likely matches a French word in the same sentence position.
- Loud and quiet pairs. Couplings like lait/milk and pourquoi/why shouted from the data; softer ones (marqué d'un astérisque/starred, qui s'est fait bousculer/embattled) emerged only after combing the data many times with slightly different algorithms.
- Generalize. Once trained, the program could translate a French newspaper article by matching its phrases against the decoded speeches, outclassing rival systems by a wide margin.
The IDA code crackers would not have been surprised: Brown and Mercer used the expectation-maximization algorithm and cited Leonard Baum, the same IDA man who had worked for Simons. Traditional translators were outraged. A reviewer protested that the crude force of computers is not science; at a conference, listeners shook their heads and grunted in disbelief, asking where the linguistic intuition was (the answer being that there was none, which was the point). IBM manager Fred Jelinek rubbed it in: every time he fired a linguist, his system improved. By 1993 the sceptics were capitulating, and statistical machine translation was hailed as an intellectual revolution.
Brown and Mercer's lesson about artificial intelligence: computers work best when they do not imitate humans. People learn grammar because it prompts their memory; computers can memorize millions of sentences (la fille est intelligente, les filles sont intelligentes and dozens of variants) and learn in chunks without knowing that est and sont are forms of être. Brute memory can substitute for human notions of intelligence.
What it suggests about Medallion
Possibly nothing, Mallaby concedes, since the reasons for Medallion's success are secret. But Brown and Mercer clearly programmed differently from other quants:
| Firm | Origins | Method | Hires | Mallaby's verdict |
|---|---|---|---|---|
| Renaissance | Technical commodity trading; price data paramount | Data first; let the computer find answers; nonintuitive signals welcome | Established cryptographers and translation programmers, expert at telling fake ghosts from real | In a class of its own |
| D. E. Shaw | Equity statistical arbitrage, rooted in intuitions about stocks | Often theory first, tested on data; signals should make intuitive sense | Young quants of all kinds, recruited in their twenties | Strong and broader, but less capital on its best ideas |
| Tudor | Discretionary macro | Machine trained to think like human traders | Goldman economist and Goldman trader | Less original; establishment ideas leak between firms |
Shaw's approach recalls the programmers who taught computers French grammar; Brown and Mercer resemble code crackers, who have no grammar book and must sift apparently random data repeatedly for ghosts invisible to the human eye.
D. E. Shaw (Eric Wepsic, head of statistical arbitrage). Beware fake ghosts: the old claim that markets rise when an original-NFL team wins the Super Bowl may hold statistically but is coincidence. Trade patterns you can explain: in the 1990s odd correlations among cable, media and consumer-electronics stocks looked like a fluke until tied to tech euphoria, which made them tradable. If you know why a signal works, you know why it might stop. Pure pattern recognition is a small part of Shaw's business.
Renaissance (Robert Mercer). A theory linking the phases of Venus to markets would need a lot of evidence, but some signals that make no intuitive sense do work, and they are often the most lucrative. The signals traded without interruption for fifteen years make no sense; otherwise someone else would have found them.
The research machine
By the late 2000s Renaissance had left the incubator for a campus with a gym, lighted tennis courts, a pond of bulbous goldfish and a skylit slate staircase: upmarket, low-key, eerily clean. Someone had taped to an office door an article titled "Why Most Published Research Findings Are False." Server rooms had elaborate key systems, but the striking thing was internal openness. Other quant funds enforced Chinese walls and need-to-know rules; at Renaissance scientists roamed freely, limited only by the risk of colliding with Brown's unicycle (mirrors were placed at key corners to see him coming). Like IDA, it was closed outside, open inside.
- The Big Meeting. Every Tuesday morning about ninety PhDs gathered. Each refinement began with a presentation, with simulations of how the signal would blend with existing ones, followed by questions: how would it have fared in the LTCM crisis, or in a period of low volatility?
- Peer review. For days afterwards colleagues could drop into the proponent's room with follow-ups.
- The Small Meeting. Only those with remaining questions attended; Brown and Mercer decided whether to go ahead.
- Final check. Henry Laufer, still chief scientist, held a veto.
- Pay. Researchers were paid on firm-wide profits, not a subunit's results.
- Shared code. As at IBM, Brown and Mercer built a system many programmers could work on at once; a researcher could even extend the in-house programming language to express a new idea.
- Data windfall. Globalization brought foreign statistics; digitization brought e-commerce sales and web-surfing data; electronic markets exposed every bid and offer, including unconsummated ones. The complexity exceeded a few minds, but a collaborative faculty could thrive on it.
Engineering loyalty
Openness assumed nobody would walk off with the secrets. Like the firm in John Grisham's novel, Renaissance thought hard about loyalty: applicants were told they could never work elsewhere in finance; it rarely hired from Wall Street, since someone who had left one trading team might leave another; researchers had to invest a fifth of their pay in Medallion, locked up as a kind of bail for four years after leaving, to enforce noncompete and nondisclosure agreements; and the remoteness of East Setauket helped, since once children were in local schools, nobody wanted to move.
Seeking the best brains, Renaissance discovered the Soviet system's prodigies, gathered from fifteen republics into Moscow's Institute of Physics and Technology; one recruit recommended another until there was a sizable Russian caucus. In 2003, having learned the secrets, two Russian researchers refused to sign the noncompete, demanded higher pay and threatened to join a rival. Simons refused to be blackmailed; they left for another fund. Because of the open structure they knew a lot, and financial innovations lack patent protection, like pirates copying a drug company's blockbuster. Renaissance sued the pair and their new employer, and a 2006 settlement required them to stop trading. Yet during the two or three years they ran a rival system, Medallion still did extremely well, perhaps less than it would have otherwise.
Lawyers are only part of the answer. Mallaby's lesson is that infrastructure matters as much as research, and research never stops: it takes enormous time and money to build systems that absorb a trillion bytes a day, make them usable to researchers, and turn findings into hundreds of thousands of glitch-free automated trades from Spain to Singapore. While the defectors struggled to build a halfway comparable platform, each Tuesday's Big Meeting moved the target further. Simons, like a magician who drinks poison and survives, looked more mysterious than ever.
The Formula One effect
Simons had a paradoxical effect on the industry. Medallion was like a carmaker's Formula One racer: few customers could ride it, but it sold ordinary cars. Because short-term trading cannot grow too large (liquidity and time limit how much goes on each trade), it closed to outsiders in 1993. Yet its fame produced a halo effect, offsetting the damage LTCM had done to the reputation of black-box trading; each magazine cover brought more institutional money into quant systems.
Simons cashed in. In 2005 he launched the Renaissance Institutional Equities Fund (RIEF), designed to absorb an eye-popping $100 billion. That scale required more liquid, longer-term strategies, and since pure pattern recognition works best at short horizons, RIEF had to rely on different signals, ones rivals such as D. E. Shaw might already have mined. By summer 2007 it had raised over $25 billion, one of the world's largest hedge funds. Then the crisis hit, and Simons, like almost everyone, felt it.
Mallaby explains the defection by a "Soviet upbringing" that bred cynical individualists, a sweeping national stereotype resting on two people; he also leaves them unnamed (court filings identified Alexander Belopolsky and Pavel Volfbeyn, who joined Millennium). He calls Baum the EM algorithm's inventor; more precisely the Baum-Welch algorithm is a special case of the expectation-maximization method generalized by Dempster, Laird and Rubin (1977). All Medallion figures are self-reported by a secretive firm, and Mallaby concedes its methods are unknown, so the translation analogy is informed speculation.
The Brown-Mercer bet, that data and statistics beat hand-coded rules, is the lineage of modern AI: statistical translation gave way to neural translation in the mid-2010s and then to transformer language models trained on vastly larger corpora. Jelinek's quip about firing linguists anticipated what researchers now call the "bitter lesson." For banks, the Renaissance research process (peer-reviewed signals, backtests against crisis regimes, a single shared code base, a chief-scientist veto) is a strong template for model risk governance.
Medallion's record held up: per Gregory Zuckerman (2019), about 66% a year gross and 39% net from 1988 to 2018; in 2020 it gained 76% while Renaissance's outside funds lost heavily. Mercer stepped down as co-CEO in 2017 amid controversy over his political funding; in 2021 Renaissance insiders agreed to pay about $7 billion to settle an IRS dispute over basket options. Simons died on 10 May 2024, aged 86.
Premonitions of a Crisis
By the mid-2000s hedge funds had become an industry of alpha factories: more than 8,000 funds, and multistrategy platforms promising to manufacture uncorrelated returns like widgets. Amaranth, the model factory, bet itself on a young gas trader, Brian Hunter, and lost $6 billion of investors' equity in 2006, more than LTCM. Mallaby reads the episode two ways. Amaranth showed a hedge-fund bubble that let inexperienced traders blow up, yet its failure harmed nobody beyond its investors, and Citadel's overnight takeover of its book showed that hedge funds could be firefighters as well as arsonists.
The age of the manufacturer
The first generation of titans had been seen as freakish geniuses whose returns were possibly lucky and certainly not reproducible. By 2005 that view was untenable: more than eight thousand funds existed, and long track records made success hard to dismiss as luck. At conferences from Phoenix to Monaco, consultants and gurus now spoke of a scientific product called alpha. Its appeal was that it could be explained: Steyer's merger arbitrage or D. E. Shaw's statistical arbitrage delivered uncorrelated, market-beating profits that professionals could understand, replicate and manufacture. Innovation gave way to the alpha factory.
Franchising Tiger
By the early 2000s nobody doubted that long/short stock picking in the Tiger style could beat the market; the challenge was implementation, not invention. Dozens of Tiger look-alikes sprang up, many run by Tiger alumni, and consultants and funds of funds arose to steer capital to the best. The biggest sponsor of clones was Julian Robertson himself. After closing Tiger in 2000 he turned his offices into an incubator for "Tiger seeds", which managed his money, took his coaching and used his prestige to raise outside capital. Where he had once made every big call, he now let protégés run their own shows: he had switched from inventing a technique to franchising it. By 2006 the reinvented Tiger complex ran $16 billion, and the Park Avenue premises were bigger than ever.
The multistrategy idea
The purest form of factory chic was the multistrategy fund. Rather than claim an edge in one style, it held that you could have an edge in any style you liked, if you hired the people. Like a drug giant hoovering up ideas from university labs and biotech start-ups, the factories gathered many strategies under one roof, blended them to diversify risk and shifted capital among them as conditions changed. They talked little of invention and much of process, treating hedge funds as financial products.
Chicago's Citadel was the prime exponent; an executive described the goal as turning investing into widget making. Ken Griffin, its thirtysomething boss, devoured management books (staff peeked at the tomes on his desk to brace for the next six-step plan), pushed people out with mechanical determination, and compared a hedge fund to a bus: people get on, people get off, the bus keeps rolling.
- $2bn to $13bn. Citadel's assets from the start of 2000 (Griffin, who began trading convertibles from his Harvard dorm, was 31) to 2007.
- More than 5%. Expenses Citadel billed clients, as a share of capital, before its 20% performance fee. Griffin's earnings were said to be the industry's second highest, behind Simons, and he spoke of rivalling Goldman Sachs and Morgan Stanley.
- $3bn to $11bn. Eton Park, launched in 2004 by ex-Goldman merger arbitrageur Eric Mindich: raised before opening, managed four years later.
- 60%. Share of industry assets held by funds of $5 billion plus by 2007, after a 1990s in which top funds had returned capital to fight bigness; a magazine's "Billion-Dollar Club" list defined who mattered.
- Almost $400bn. Fund-of-funds assets by 2005.
- $5bn to $100bn+. Hedge-fund capital in energy trading, 2001 to 2006.
The rush to size had real logic. Small firms generate ideas; big firms implement them. After the garage-workshop phase, it took professional outfits to bring inventions to market: lightning-fast computers, legal departments fluent in many countries' rules, treasury desks squeezing brokers for terms, marketing teams churning out glossy monthly reports for demanding institutions. If the edge lay in the platform rather than in original ideas, it made sense to run many strategies on it, and many strategies meant huge capital. Customers pushed the same way: funds of funds had grown by promising to shift money nimbly among strategies, and MBA-minded managers reasoned they could cut out the middleman by building that flexibility in.
Yet wise observers were uneasy: too many people were making too much money too fast. Consultants ran workshops on launching a fund, Hedge Funds for Dummies appeared in bookshops, and grandees with no investing background, such as former secretary of state Madeleine Albright, joined in. The frenzy recalled the 1980s buyout boom and the dot-com mania. Surely it would end painfully.
Amaranth: a factory without an edge
Amaranth was the very model of the modern alpha factory. Founder Nick Maounis, a convertible-arbitrage specialist, hired experts in merger arbitrage, long/short equity, credit arbitrage and statistical arbitrage, and in 2002, after Enron's collapse, snapped up stranded Enron staff to open an energy-trading desk. He made the standard case for mission creep: blending strategies diversifies risk, and capital would move aggressively among them. The shape-shifting was a point of pride:
- Launch, September 2000. Nearly half the capital in merger arbitrage.
- A year later. Merger arb cut to practically zero; more than half in convertible arbitrage.
- Another year on. Shifts into bonds, then statistical arbitrage and energy.
Why should a pension plan pay a fund of funds when it could go straight to Amaranth? Results were excellent: 22%, 11% and 17% in its first three full years, while the S&P 500 mostly fell.
But Mallaby finds the factory hollow. Despite the marketing, Maounis had no clear edge in choosing which strategy to shift into: he raised allocations to styles that had recently worked and cut those that had not, with no sure way to know what would work next. Because alpha had become a commodity, dozens of rival factories were driving down returns by making the same thing, so the shape-shifting was less clever cycle-timing than a desperate search for the next trick. And since he was allocating to specialists whose books were hard to understand, his decisions rested on instinct, and gut feelings about individual traders could sway him dangerously.
Brian Hunter's winter-gas trade
Those dangers converged in Brian Hunter, a six-foot-five Canadian with a graduate degree in maths, who sometimes wore a Calgary Flames jersey and was earnest, soft-spoken and unfailingly calm. From his arrival in 2004 his natural-gas returns stood out. He had spotted an anomaly:
- Rigid supply. Unlike oil shipped by tanker, gas moves mainly by pipeline, so supply cannot be rerouted to fill sudden local shortages; a cold snap makes heating demand spike and prices leap.
- Cheap insurance. Options that would soar in a shortage were strangely cheap: bargain weather insurance with an asymmetric payoff (lose a small premium, or win many times over).
- The spread version. Go short a summer futures contract and long a winter one, betting the narrow spread widens if winter prices jump.
- Payoff. It had worked in recent winters, and in November 2004 gas rose about 80% above its summer low; Amaranth cashed in.
Most of Amaranth's strategies were faring poorly: convertible arbitrage had hit a wall, and a heavy push into statistical arbitrage (Maounis wanted a piece of Simons's action) had little to show. In April Maounis learned that SAC Capital had offered Hunter a $1 million bonus to defect. Back to the wall, he promoted him to co-head of the energy desk with authority to place his own trades, and raised energy's share of capital from 2% the previous spring to around 30%. He had effectively bet the company on a 32-year-old who had been with him barely a year: the sort of thing, Mallaby notes, that happened in the go-go mid-2000s.
The promotion was the more remarkable given Hunter's past. At Deutsche Bank, where he supervised gas trading, his group lost $51 million in a single week in December 2003. In a later New York lawsuit he blamed an "unprecedented and unforeseeable" price run-up, as if failing to foresee the market absolved him, and cited the bank's well-known software problems in exiting losing trades, as if his risk-taking were management's fault. He argued that he personally had earned $40 million for the bank in 2003 and so deserved a bonus. By February 2004 Deutsche had decided there was no place for him.
Katrina makes a hero
Four months after the promotion came the mother of weather shocks. Hurricane Katrina struck Louisiana in August 2005, flooding New Orleans and wrecking Gulf of Mexico gas rigs; Hurricane Rita followed in September, and gas prices hit a record. Amaranth, down 1% in the first half, finished 2005 up 21% against 9% for the average fund. Hunter's trades earned $1.26 billion, nearly all of Amaranth's profit, and he reportedly kept a tenth. Assets swelled to about $8 billion, the world's 39th-largest hedge fund. The Christmas mailing sent clients toy gasoline pumps and a card quoting Benjamin Franklin on energy and persistence altering all things; seldom, Mallaby jokes, had a Founder been so taken out of context.
The hurricanes were freak events, but Hunter, who had blamed unforeseeable extremes for his Deutsche losses, happily took credit when they made him a hero. Maounis grew enamoured, seeing him as a convertible-arbitrage trader in a different space: profits with little risk, because winter gas could never fall below summer gas, so losses seemed capped. The risk department reinforced the view; one gas risk officer called Hunter the greatest commodity trader he had ever seen. Hunter moved his family and team to his native Calgary, commuting in a Ferrari (a Bentley when it snowed).
Trading on his own momentum
Early 2006 went even better: about $2 billion of profit from January to the end of April, again nearly all of Amaranth's performance, leaving it up roughly six times the 5% of the average multistrategy fund. Maounis began saying that although he had not struck gold in stat arb, he had found an equally potent secret weapon. Savvy observers were alarmed. Such profits implied outlandish risks, and the energy trade was crowding: with hedge-fund energy capital up from about $5 billion to over $100 billion, any strategy energy specialists liked would come good as others piled in, and blow up if the stampede reversed.
With Hunter an extreme version was under way: the weight of his own money was driving his profits.
- Market share. By the end of February he held about 70% of NYMEX gas futures for November 2006 delivery and about 60% for January 2007, betting November would fall and January rise; so long as he kept adding, the bet was self-fulfilling.
- Fundamentals against him. Output had recovered from the hurricanes and a mild winter cut demand; by April gas in storage was nearly 40% above the five-year average.
- Total book. By about the end of April Amaranth owned over 100,000 NYMEX contracts, more than 40% of open interest across all months. Hunter was, in Mallaby's phrase, a momentum trader trading on his own momentum.
Blackstone walks
In May 2006 a team from Blackstone, whose long-standing fund of funds had $125 million in Amaranth, visited Calgary. Amaranth was up a whacking 13% in April alone, but that showed how big the bets were. The risk department's maximum monthly loss of $300 million (about 3% of equity) rested on the claim that winter could not fall below summer. Hunter's own buying had made that obsolete: the summer/winter spread had widened from $1.40 in mid-February to $2.20 in late April, leaving plenty of room to shrink. And in a small market whose main buyer was Hunter himself, there would be nobody to sell to. Blackstone redeemed at the next opportunity, gladly paying the short-notice penalty.
Maounis finally told Hunter to cut risk, but nobody would buy. As soon as Hunter sold, the market turned: by the end of May Amaranth was down more than $1 billion, nearly four times what the risk department had deemed possible. Other desks were ordered to shrink to free capital, and Amaranth paid Morgan Stanley a large fee to take some of the exposure, but it was too little, too late. Rivals had noticed, and the positions were no mystery: anyone could see which contract pairs had widened in March and April. Like LTCM in its bond trades or Robertson in US Airways, Amaranth was trapped; one rival said it was naive to think anyone could exit 100,000 positions.
Liquidity shrinks as your size grows. Hunter's "maximum loss" assumed an exit that his 40% market share had destroyed, the same trap as LTCM's bond trades and Tiger's US Airways stake. Position limits measured against open interest, and risk models that ask who the buyer will be, are the banker's answer.
The summer waiting game
Unable to cut without crippling losses, Hunter waited for something to let him out. At the end of July a rumoured late-summer hurricane revived his bravado: he jacked up his bets, widening spreads and sinking MotherRock, a rival fund on the opposite side. In August Maounis told the Wall Street Journal that Hunter was really good at taking "controlled and measured risk"; a veteran later likened his faith to a jilted lover's. Losses at the end of August brought margin calls, the hurricane season passed quietly, and predators began targeting the positions.
Citadel's weekend
On a rainy mid-September day Maounis took a limousine (he should have taken the train) from Greenwich to the Pierre Hotel for a Goldman Sachs hedge-fund conference that captured the era: dozens of tables of fund chieftains and investor groups hurrying from stall to stall like speed-dating. He was reciting his patter (world-class equity, credit, quant and commodities teams in world-class infrastructure) when an e-mail warned of trouble at the office.
Thu 14 Sep Amaranth loses $560 million in a day as a key summer/winter spread shrinks to a third of its early-September width; at Maounis's home that night the top brass agree to raise capital for margin calls; he calls Goldman about buying the energy book, others call other banks.
Sat 16 Sep Bankers jam the Greenwich car park and devour the Pop-Tarts; analysts confer with bosses in the Hamptons while positions bleed after hours. Goldman bids, then Merrill Lynch.
Mon 18 Sep, early Amaranth thinks it has a deal to sell a chunk to Goldman, an apparent LTCM replay; a letter tells investors it has lost half its peak capital and that new capital is "near completion."
Mon morning Clearing broker J.P. Morgan, which had financed the gas trades against futures collateral, will not waive its right to claw back assets through bankruptcy; Goldman cannot value the book under that threat and the deal falters.
Mon morning, Chicago Griffin, on an elliptical trainer rigged with monitors, reads an e-mail from investor-relations head Scott Rafferty: Amaranth down 50% in a month. He recalls Vinny Mattone's warning to Meriwether: down by half, you do not recover. He phones Amaranth COO Charles Winkler, a former Citadel employee and friend, asking how Citadel can help; Winkler wants a bridge loan and a couple hundred million.
Mon lunchtime Losses force a sale of all energy positions, not just gas; about 40 Citadel staff work in teams and two young lieutenants fly to Greenwich.
Mon evening Citadel's talks eclipse the banks', but bankruptcy review risk remains; Griffin's team works through the night.
Tue 19 Sep J.P. Morgan offers to waive clawback if it can buy 50% of the positions; Griffin accepts. A last worry, hidden claims from other brokers, is settled by Bob Polachek, who had checked every Amaranth brokerage statement overnight and found no holes.
Wed 20 Sep, 5:30 a.m. The sale goes through.
The deal tested the multistrategy premise, that the edge lies in the platform. Griffin had to grasp the trades' logic, financing, counterparties and whether Citadel's systems could carry a book that was a large chunk of the entire gas market, which by Monday had already bled $4 billion to $5 billion. Citadel, at $13 billion, should have struggled to digest it. But each time the goalposts moved, the banks had to consult up and down a chain, while Griffin, chief executive and chief deal maker in one, decided instantly. The speed ran down the firm: Citadel's technology chief could move Amaranth's trade data into Citadel's accounting and risk systems without the committees a bigger bureaucracy would need.
The outcome was a triumph. Thanks partly to J.P. Morgan's offer but also to a platform at least as good as the top banks', a hedge fund had stolen a deal Wall Street considered its own; in 1998 Goldman had shown interest in LTCM's portfolio but never closed. Once Citadel and J.P. Morgan owned the book, predators saw that no dump was coming, cut their bets, and prices recovered. Citadel eventually made about $1 billion, a long step toward Griffin's next Goldman Sachs.
| LTCM, 1998 | Amaranth, 2006 | |
|---|---|---|
| Fatal trade | Leveraged bets on converging spreads in illiquid bonds | One trader's outsized NYMEX gas spreads |
| Investor loss | Smaller than Amaranth's | $6 billion of equity |
| Who stepped in | Wall Street banks under New York Fed aegis; Goldman's bid failed | Citadel and J.P. Morgan, in about two days |
| Official role | Round-the-clock Fed meetings | None; no taxpayer cost |
| Systemic effect | Feared contagion across markets | About a bad day for the S&P 500 |
A bubble, but not a systemic one
The bubble. Hedge funds had grown faster than the available talent, and under pressure to perform they gave inexperienced traders room to blow up. The factory format made it worse. Veterans such as Druckenmiller or Louis Bacon grasp risk because they trade daily with their own savings and reputations at stake; factories delegate, so the boss cannot master every risk, especially when rapid growth forces rapid moves into new styles. Delegation imports a mild form of big-bank incentive problems: traders want bonuses, bonuses come from betting big, and central risk departments seldom restrain wizards with an aura of invincibility.
The non-crisis. The calls for regulation missed the mark. Senate Finance chairman Charles Grassley wrote to Treasury secretary Henry Paulson that pension savers were exposed and attacked hedge-fund secrecy, yet Amaranth reported its strategies to investors monthly, and that openness made it a target. Most economists in a Wall Street Journal survey favoured tougher oversight, often SEC registration, but its benefit was unclear. Taxpayers lost nothing; there was no crisis meeting at the New York Fed and no visible damage to the system; the hit to pensions, for which Amaranth was a tiny holding, was like a bad day for the S&P 500.
Mallaby's conclusion: the market disciplined Amaranth while inflicting minimal wider damage, and no regulatory clamp could have done better. Since LTCM, Wall Street had dreaded the next blowup; it came, and the scars were barely visible. Critics still feared leveraged funds could start systemic fires, as LTCM had, but Citadel's lightning purchase showed the other side: hedge funds might occasionally ignite fires, but they could also be the firefighters.
Mallaby's "no harm done" verdict treats Amaranth only as an investor loss. Regulators saw market manipulation: in 2007 FERC and the CFTC charged Hunter and Amaranth over their gas trading, and a Senate subcommittee argued that Amaranth's positions had distorted prices paid by utilities and consumers. The "premonition" also cut against his own reading: the factories' delegated risk-taking and crowding in illiquid trades were exactly the weaknesses exposed a year later.
The legal sequel came after the book: in March 2013 a US appeals court ruled that FERC lacked jurisdiction over Hunter's futures trades, voiding its $30 million fine, and in 2014 he settled with the CFTC for $750,000. The factory model Mallaby describes went on to dominate, as multistrategy "pod" platforms led by Citadel and Millennium drew the largest inflows; Citadel moved its headquarters to Miami in 2022, and its flagship Wellington fund gained 15.1% in 2024.
Riding the Storm
The subprime bubble was the policy error of the 2000s, and John Paulson played it as Soros had played sterling, turning a few hundred million dollars of insurance premiums into an estimated $15 billion of profit for his firm in 2007. Mallaby uses that year to stage his central comparison. Freestanding hedge funds mostly dodged toxic mortgages and even rescued one another, while banks such as Citi, UBS, Merrill and Bear Stearns bought the garbage by the bucketful and ended up on the taxpayer's books. The August quant quake showed that hedge funds could still hurt themselves badly, but in Mallaby's reading it did not hurt the system.
Daniel Sadek and the mortgage mania
Mallaby opens with a lender who personifies the bubble. Daniel Sadek, born in Lebanon in 1968, had his schooling broken by the civil war and carried scars from a gunshot and flying shrapnel. He moved to France, reached California at eighteen, pumped gas and then sold cars. Around 2000 he noticed that many of the buyers of his Mercedes worked in mortgages, and found he could get a licence to sell home loans without taking a single class; a would-be barber, by contrast, needed 1,500 hours of training for a state licence.
By 2005 his firm Quick Loan had 700 employees and ranked among the nation's top fifty subprime lenders, meaning it lent to people too risky for normal mortgages. Its ads promised no income verification and instant qualification, and another welcomed prior bankruptcies, tax liens and foreclosures, as if hunting for deadbeats. The California Department of Corporations logged complaints of fraud and underwriting errors. Sadek bought a coastal mansion, a Las Vegas flat, fast cars, restaurant stakes and a film company, and bankrolled Redline, a car movie built around his actress girlfriend that the Boston Globe panned; filming and promotion destroyed more than $2 million of his sports cars. Its tagline urged viewers to fear nothing and risk everything.
The skeptics: Kyle Bass and Michael Litt
Kyle Bass, a Dallas hedge-fund manager at Hayman Capital, heard the Sadek story in 2006 and took it as confirmation that the mortgage market was a bigger bubble than anything in hedge funds. Subprime volume had quadrupled between 2000 and 2005, and more of it went to borrowers who could not repay. A lender who shrugged off Porsches being catapulted for a film was, Bass said, exactly the man to bet against. Around the time Amaranth was failing, Hayman traced how loans like Sadek's were bundled into mortgage bonds, shorted a large quantity of them and waited.
Michael Litt, co-head of the alpha factory FrontPoint Partners, had his epiphany in two steps:
- The suits. Touring Lehman Brothers' new mortgage trading floor, he heard traders mock a colleague for ordering only one $6,000 suit from a visiting tailor.
- The report. Flying home from London he read a Bank for International Settlements paper crediting sophisticated finance with suppressing volatility. Gazing at Greenland, once habitable and later frozen, he concluded that volatility was low because the world was awash with wild money, not because of any structural improvement; investors from Asia to Arabia were wiring billions to New York to buy every mortgage paper Lehman could sell, and the BIS was missing the coming freeze.
Litt rebalanced for a shock: in the autumn of 2006 FrontPoint went short subprime and turned bearish in several other strategies. Both Hayman and FrontPoint profited handsomely over the following year.
John Paulson, Mr. Average
The biggest killing went to John Paulson, whom Mallaby paints as unremarkable in every visible way: middling height, looks and manner. After Harvard Business School (1980) he moved from a management consultancy to Odyssey Partners, an early hedge fund, then to Bear Stearns in mergers and acquisitions. In 1994 he launched a tiny fund in a Park Avenue building that incubated several start-ups. Paulson and Company grew from $2 million at inception to $600 million in 2003, then rode the asset boom to $4 billion by 2005, still with only seven analysts and a deliberately low profile.
He was a loner and contrarian who needed no affirmation and had honed the unconventional long shot:
- Long-odds merger arbitrage. Besides collecting the usual small premium on deals expected to close, he bet against mergers that might collapse or on deals where a surprise rival bid could shock the market.
- Calling the cycle. In booms he looked for the moment to go short, and the reverse in busts.
- Capital-structure arbitrage. Because senior bonds are repaid first in bankruptcy and junior bonds get the leftovers, the two trade at similar prices while a firm is healthy. When the economy weakened, Paulson picked companies likely to fail (cyclical, over-indebted) and shorted the junior bonds, sometimes hedging by going long the senior ones.
His team shared the maverick streak. Paolo Pellegrini, a tall, elegant Italian with heavy-framed glasses who called himself a romantic, had spent years chasing offbeat ideas without success; joining as the analyst for financial firms in 2004, he found his unconventional style welcomed for the first time.
Hunting for capitalism's weakest spot
By early 2005 Paulson sensed the cycle turning: bonds he had bought cheaply in the last recession traded at silly heights, and years of low interest rates had made markets frothy. He wanted the target that would blow up loudest and fastest in even a mild slowdown, combining three weaknesses: a cyclical industry, too much debt, and debt split into senior and junior layers so he could short the junior part where risk concentrated.
He tried shorts on car-company bonds (consumers were taking on unaffordable car loans), an insurer and a couple of home lenders. All shared a flaw: they had franchise value (buildings, brands, customer relationships), so they would be worth something even after collapsing under debt. A true contrarian wanted something that could go to zero. In the spring of 2005 he found it in mortgage securities.
They had every charm a short seller could want. Home prices were cyclical, whatever the public believed; they rested on mountains of household debt that families could not service in hard times; and the seniority structure was a feast. Lenders like Sadek sold loans to Wall Street banks, which made mortgage bonds; other banks bought those bonds, rebundled them into collateralized debt obligations (CDOs) and sliced them into tranches from AAA down through AA to BBB and below, sometimes eighteen layers deep.
If the pooled mortgages repay at least 95%, the first 5% of losses fall on even more junior layers and BBB bonds are untouched. But the BBB tranche was only 1% thick, so a 6% non-payment rate wiped it out completely. Unlike a bankrupt carmaker, a pile of loans that pays nothing has no franchise value: it is worth exactly zero.
Pellegrini's numbers
In April 2005 Paulson made his first bet: a credit default swap, in effect insurance against default, on $100 million of BBB-rated subprime debt for $1.4 million a year. The payoff was hugely asymmetric, but a big payout means little if the odds are terrible, like a single number at roulette. To size the odds Pellegrini got a $2 million research budget, bought the largest mortgage database in the country, hired an outside firm to store it and added analysts to study past default rates.
- Discovery one (discouraging). Families with unpayable loans did not necessarily default: while prices rose, they could refinance their way out, so defaults stayed low during the boom.
- Discovery two (decisive). The industry held that house prices, then rising about 15% a year (summer 2005), had never fallen nationally at once, so CDOs mixing mortgages from many states were treated as near riskless. Pellegrini, a newcomer free of that faith, found that adjusted for inflation there had been national slumps in the 1980s and 1990s.
- Flat is enough. Prices did not need to fall. If they merely stopped rising, owners would lack the collateral for new, larger loans, refinancing would stop, and his model implied defaults of at least 7%, enough to erase every BBB bond. The team's shorthand: zero would mean zero.
The payoff table
By early 2006 the first mortgage bets had lost money and some investors grumbled that Paulson had strayed beyond his competence. He grew more convinced. With the Fed's rate rises pushing up mortgage costs and slowing appreciation, the chance of flat prices was at least even, like betting on red; yet the payout was seventy or eighty times the stake, double a single-number roulette win. He drew a simple table for a $600 million fund spending 7% of its capital a year on BBB insurance.
| Scenario | Default rate on BBB bonds | Fund return | Dollars |
|---|---|---|---|
| Worst case | Housing holds up; premiums lost | about -7% | -$42m |
| Mild | 30% | +341% | +$2.0bn |
| Middle | 50% | +568% | +$3.4bn |
| Paulson's base case | 80% (he thought it highly likely) | +909% | +$5.5bn |
Some investors thought a 909% projection proved he had gone mad. Undeterred, in the summer of 2006 he launched a dedicated credit fund to do exactly what the table said, seeded partly with his own money and co-managed by Pellegrini.
Buying in size: the ABX
Scale was the obstacle. Shorting mortgage bonds by borrowing them was cumbersome, and buying swaps depended on finding a bank willing to sell. In July 2006 the big investment banks, hungry for commissions, launched the ABX, a subprime mortgage index, and insuring $10 million at a time became easy. A week or two later an ABX trader at a big bank phoned, asking for Paulson's full appetite. Wary of revealing it and seeing the price move against him, Paulson offered $500 million; the trader simply said "done", and did the same for a second $500 million. The next day Paulson bought protection on another $1 billion. In the first half of the year he had scrambled to assemble $500 million; now he had four times that in two days.
Recalling Soros's advice to go for the jugular, he offered to do another $3 billion. The line went silent; the trader agreed to $1 billion more and stopped. Calling round other banks, Paulson built positions of $7.2 billion in the credit fund, and at the end of 2006 opened a second fund with the same strategy.
New Century Financial, the country's second-largest subprime lender, announced that its fourth-quarter results, due the next day and expected to show a healthy profit, would be postponed while it calculated losses; some borrowers had missed their very first payments. The same day HSBC, the third-largest US subprime lender, set aside $10.6 billion for bad mortgages. Next morning the ABX fell five points. At $250 million per point, Paulson netted $1.25 billion in a morning, roughly Soros's haul against sterling. When he reported February results a client rang to ask whether 66% was a misprint for 6.6%. It was not.
Once prices stopped rising, over-indebted families handed in their keys and BBB mortgage securities became nearly worthless. Within months New Century was bankrupt and HSBC had fired its US executives; every blow to the industry was a bonus for Paulson. On a hot summer day he left two prospective clients in a stuffy conference room after a colleague whispered to him, came back unable to stop grinning, and finally confessed that the day's marks showed a billion dollars made.
The new Soros
Mallaby finds the Soros echo more apt than Paulson knew. Each era of hedge-fund triumph exploited the official blunder of its time, and the subprime bubble was the twenty-first-century version.
| Era | The policy error | Who cashed in |
|---|---|---|
| 1970s | Incompetent central banks stoke inflation | Commodity traders riding trends |
| 1990s | Central banks defend untenable currency pegs | Macro traders such as Soros |
| 2000s | Victory over inflation licenses cheap money; stable currencies embolden Wall Street to lever up | Credit funds such as Paulson's |
The heyday of macro might be over, but a new heyday of credit hedge funds had arrived: Paulson was the new Soros. The parallel ran deeper. Soros's 1992 coup needed a sucker on the other side, the British government. Paulson could build vast shorts only if someone was buying the mortgages, so the puzzle was who, and why so eagerly.
Who bought the garbage
When the bubble burst in 2007-08, losses surfaced everywhere: loans like Sadek's had been packaged and sold to buyers from Japanese insurers to Norwegian pension funds. Some hedge funds were caught. Peloton Partners and Sailfish Capital, both medium-sized, sank under mortgages; Peloton's London managers had become notorious when their secretary stole £4.3 million without their noticing, though they told her trial the account had felt a million or two light. Yet the sector as a whole came through remarkably well, mostly avoiding toxic paper and often shorting it. Per Hedge Fund Research in Chicago, funds specializing in asset-backed securities (mortgages included) were up 1% on average in 2007, and hedge funds overall gained 10%, respectable for a crisis year.
The heavy buyers were banks and investment banks, firms such as Citibank, UBS and Merrill Lynch, which were proud of their proprietary trading desks and yet loaded up on subprime. The consequences were public:
- Citi. Losses so vast that the US government had to rescue it, buying more than a third of its shares.
- UBS. Needed a lifeline from the Swiss government.
- Merrill. Sold itself to Bank of America to avoid collapse.
Whereas the failures of Peloton, Sailfish and Amaranth cost taxpayers nothing, the failures of Citi and its peers imposed enormous burdens on government budgets and the world economy.
- 1. Regulation became a crutch. Deposit-taking banks must hold minimum capital, which should have made them sturdier. Instead they ran their books to what the rules deemed safe rather than what analysis showed. Toxic mortgage bonds won AAA ratings, partly because issuers paid the rating agencies; AAA paper needed little capital, so banks held it happily. Ratings and rules replaced analysis; hedge funds made their own risk calls and often did better.
- 2. Incentives. Hedge-fund pay is imperfect: managers keep a fifth of profits without returning a fifth of losses, which tempts gambling. But they usually keep their own wealth in the fund. Bank prop traders risk no personal savings; they may own bank stock, but a prop desk's losses are diluted by other business lines and barely move the share price, too weak a signal to change behaviour.
- 3. Distraction by fee businesses. Prop desks sat beside advisory, underwriting and asset-management arms, and the chase for fees warped investment choices. Merrill reportedly sold about $70 billion of subprime CDOs at a 1.25% fee, or $875 million; CEO Stan O'Neal, obsessed with the mortgage league tables, would finance home lenders at no profit to be first in line for their loans. To feed the CDO machines banks kept large inventories of mortgage bonds, so when CDO demand died in early 2007 billions of unsold inventory landed on their balance sheets: they became huge mortgage investors by accident.
- 4. Culture: paranoia versus complacency. Hedge funds fear margin calls and redemptions, live by returns, and are run by a founder who can flip positions fast. Banks have many revenue streams and seemingly secure funding: government-guaranteed deposits, or, for investment banks, market access they wrongly thought made them impregnable. After Russia and LTCM in 1998, hedge funds locked up investors longer and won broker guarantees, while bank equity buffers fell by about a third from the mid-1990s to the mid-2000s. Even in 2006-07 banks sold Paulson billions of protection through the ABX without asking what his buying signalled.
Bear Stearns: bank culture in hedge-fund clothing
Bear Stearns summed up the contrast, with a twist. Known as a vigilant risk manager, it was the last firm one would expect to buy poison. But by the mid-2000s it had become Wall Street's number one packager of mortgage-backed securities, up from third in 2000, and had bought subprime-lending subsidiaries in the US and Britain to keep the sausage factory supplied. Managers came to care more about making mortgages than about what they were worth.
In 2003 Bear also set a "10 in 10" goal: asset management, then insignificant, would supply 10% of revenues and profits by 2010. Chasing that fee income, it hurriedly put unqualified executives in charge of launching internal hedge funds, some of which loaded up on subprime. That misstep led to Bear's collapse the next year and to the Federal Reserve absorbing $29 billion of its toxic securities.
Mallaby rejects reading this as proof of hedge-fund danger: the Bear funds were products of bank culture. Like other bank-sponsored funds they were run by people seeded inside a big firm, not founders of independent ventures; they raised money through the parent's network and brand, which lowered the entry barriers freestanding funds face; and they knew the parent might bail them out, which dulled vigilance.
Two of the funds were run by Ralph Cioffi, who had spent years on Bear's sales desk selling mortgage-backed securities to institutions. His plan was to buy those same securities and lever them 35 to 1, a bet that suited a fee-hungry parent but would rarely fly at a real hedge fund. He had almost no experience of controlling portfolio risk; Paul Friedman, COO of Bear's fixed-income desk, recalled many internal skeptics, since Cioffi was bright but had never managed money or written a ticket that wasn't someone else's money. He had also disappointed in a brief supervisory stint.
Even with Bear smoothing the way, Cioffi struggled with administration: he failed to get his independent directors' approval before buying securities from other Bear divisions, and the paperwork was so chaotic that a law firm was hired to investigate. One outside investor complained that it had invested on the strength of Bear's reputation for managing its own risks, and that Bear treated outside clients differently.
Bear and other banks that jumped on the hedge-fund bandwagon cared more about leveraging their brands than managing risk. The non-hedge-fund character of Bear's internal funds goes far, Mallaby argues, toward explaining why Paulson made billions while his former employer went under. In June 2007 Cioffi's funds blew up, and because they had been sold on Bear's name, Bear felt obliged to rescue them with an emergency loan, exactly the parental cushion that erodes vigilance.
- 700%. Cumulative net-of-fees gain of his flagship mortgage fund by the end of 2007, the biggest killing in hedge-fund history.
- About $15 billion. Estimated profits generated by Paulson and Company.
- $3-4 billion. Paulson's personal take; one magazine profile dubbed him the man who made too much.
- 2008. When he proposed changes to Treasury secretary Hank Paulson's bank-bailout plan, Representative John Tierney of Massachusetts quipped that Washington might have the wrong Paulson in charge, a deference recalling Soros's.
Bear's failure and Paulson's triumph looked like a win for hedge funds, but it was too early to tell whether they would survive the shocks to come.
Citadel puts out Sowood
The month after the Bear funds failed, Ken Griffin of Citadel was on holiday in France. He was at his peak: a landmark tower in downtown Chicago, a Jasper Johns bought for $80 million, a recent wedding to his French bride at Marie Antoinette's mock village, the Hameau de la Reine at Versailles. But each day began and ended with calls to Chicago, and by Friday morning he told his wife she could come or stay; he was going home.
That Friday, 27 July 2007, the subprime trouble became a broader credit crisis. Until then the Porsche-catapulting lenders, eighteen-layer CDOs and ABX side bets could be dismissed as a mania confined to one corner. Now Sowood Capital Management of Boston caught fire: its $3 billion portfolio was falling and brokers were making margin calls.
Sowood had avoided subprime. Its boss Jeffrey Larson had made his name at the Harvard endowment, which matched Yale's enthusiasm for absolute return with its own stable of in-house managers; by the end of twelve years there he ran $3 billion of its $20 billion. In 2004 he persuaded Harvard to seed an independent multistrategy fund, which grew to about 70 staff and made about 10% a year in its first three years, mainly in credit. In early 2007 he rightly expected rising defaults and, in a version of Paulson's capital-structure arbitrage, went long senior bonds and short riskier paper. Bear's subprime woes seemed harmless or even helpful; in early July he put another $5.7 million of his own savings into the fund.
His error was soon exposed. Leveraged players who had lost on mortgages raised cash by dumping non-mortgage holdings, and in the third week of July Sowood's corporate bonds fell hard. High-quality bonds, supposedly safe in a downturn, were the easiest to sell, so they went first; when everyone deleveraged indiscriminately, the logic of capital-structure arbitrage broke down. Harvard declined an emergency injection as too risky. By Friday morning Wall Street knew, and traders positioned for a contagious spiral: margin calls would force a fire sale, hit other bond funds and trigger more selling. The fear was self-fulfilling; bonds fell hard that afternoon, and Sowood kept bleeding in after-hours trading.
Around Sunday lunchtime Larson rang Griffin, recalling the Amaranth purchase, and asked whether Citadel would buy Sowood's book before markets opened, so predators would stop targeting it. Griffin phoned his lieutenants. Gerald Beeson, one of the two who had parachuted into Amaranth, had just begun a beach holiday near Chicago; he raced home, pulled on long trousers and drove to the airport, phoning colleagues on the way. By about 7 p.m. half a dozen Citadel staff had taken a conference room in Boston, where they found Morgan Stanley also summoned. Much of the book was hard to value: over-the-counter derivatives that only a firm trading them all could price, a tangle of legal arrangements with brokers and counterparties, and trade data to load into the buyer's systems. Around 9 p.m. Morgan's team leader told Griffin they would resume in the morning; Griffin replied that Citadel would be done by then, and was unsure whether the banker was laughing. By 7 a.m. Monday Citadel owned Sowood's entire trading book; Larson told investors Citadel had offered the only immediate, comprehensive solution.
The costs were real: Sowood's two funds were down 57% and 53% for the month and Harvard's endowment lost $350 million. But the announcement killed fears of a dump, the bond market recovered more than 4% that day, and the panic ended. One hedge fund had threatened to start a systemic fire, and another had arrived as the fireman. Almost at once, a new fire broke out.
The quant quake
Bear's conference call
On Friday 3 August a ratings agency warned that Bear Stearns' debt might be downgraded, the first time in the crisis that a Wall Street firm's own health was questioned. Bear's stock plunged and its bosses held a conference call to reassure investors. More than 2,000 people dialled in, to no effect: the CFO said the credit markets were the most extreme of his long career, and Bear's treasurer later complained that he had blown the market up. Hours later on CNBC, Jim Cramer berated the Fed for doing nothing and declared Armageddon.
Signals that stopped working
The next fire came from an unexpected quarter. In the second half of July, computerized equity quant systems had stopped performing, some losing money. The quants found that no new risk was swamping their signals; the signals themselves had failed, all at once:
- Momentum programs, betting that price trends continue.
- Reversal programs, betting that trends turn back.
- Value programs, betting that cheap stocks with low price-to-earnings ratios beat expensive ones.
The likely cause: one or more quants liquidating holdings, perhaps to raise cash after mortgage losses, driving prices against anyone with a similar portfolio.
At the end of July Mike Mendelson, a hard-charging ex-Goldman quant, trimmed leverage at AQR, founded in 1998, which ran $10 billion in hedge funds and $28 billion in traditional funds. Its chief founder Cliff Asness had written academic papers on stock-pricing anomalies and programmed computers to harvest them, producing steady, uncorrelated returns. AQR's July losses were modest, but Mendelson had heard of a bad loss at another big quant fund. In the first days of August the models worked again; whoever was selling seemed to have stopped.
Monday, 6 August
After routine meetings at AQR's plain suite beside Greenwich station, Mendelson walked to a Subway sandwich shop around mid-morning and checked his BlackBerry in the queue. Every number was red: tens of millions lost in three hours. Some quant was deleveraging on a monstrous scale, or several at once. His one certainty was that AQR had to cut leverage fast, because of arithmetic he and colleagues sketched on a notepad as a hyperbolic curve.
- 8 to 1. Start with $100 of capital supporting $800 of positions.
- Over 12 to 1. A 5% loss ($40) leaves $60 of capital against $760 of positions.
- About 33 to 1. A second 5% loss leaves roughly $22 against $722: leverage nearly triples rather than rising another half.
- Infinity. A third 5% loss makes capital negative.
- The survival rule. Stay on the flattish left side of the curve: a fund down 5% might have to sell almost two fifths of its positions to hold leverage steady, or enter a death spiral.
Crowding in a panic
The same drama played out across the industry. Most quant funds were not like Simons's Medallion: they traded well-known anomalies, not esoteric secrets, and Asness admitted there was no E=MC² under the hood. Even Simons conceded that his big institutional fund used signals others understood; everyone had read the same academic papers, studied the same data and made similar bets on momentum and value. In calm markets this mattered little, since there were dozens of ways to measure value. In a panic crowding bit: one big fund's selling inflicted losses on others, especially now that quant strategies were large enough to move prices, and the hyperbolic curve then forced those others to sell too. By Monday afternoon Mendelson saw that AQR would have to sprint to stay still, racing rivals to shrink.
Tuesday was worse: AQR's models lost money twice as fast, and it ordered its computers to dump billions of positions. Technology let it liquidate far faster than in earlier years, but rivals could do the same; every quant was torpedoing every other, with rumours of funds down 10% or more. At Renaissance's Long Island campus Simons and his lieutenants tweaked model parameters like pilots in a hurricane. Asness erupted and smashed computer screens, then took a call from Griffin, by now famous for buying corpses out of car wrecks.
Washington in a parallel universe
Policy makers seemed oblivious. On Tuesday 7 August, day two of the quake, the Fed's rate committee warned about inflation. On Wednesday President Bush visited the Treasury and voiced hope for a soft landing if markets functioned normally. The tone shifted on Thursday 9 August, less because of the quants than because BNP Paribas froze redemptions in three money-market funds, citing vanished liquidity. The European Central Bank injected $131 billion; by afternoon Ben Bernanke had turned his office into a war room with deputies dialling in from holiday, and early Friday the Fed reversed course, promising cash to keep markets orderly.
Mendelson's detective work
Mendelson's relief owed nothing to the Fed. From Tuesday he phoned brokers and friends to learn how far the big leveraged quants had cut; once all were on the flat part of the curve, the selling would stop. The business was dominated by a handful of firms:
- Renaissance' new institutional fund, with more than $25 billion.
- Highbridge Capital, a J.P. Morgan subsidiary.
- D. E. Shaw, Barclays Global Investors and Goldman Sachs Asset Management.
Just after midnight on Tuesday he reached a friend who had rushed back from holiday and was exhausted and babbling; Mendelson could tell he was about to liquidate everything, and ticked that firm off. By Thursday evening he concluded that only one big player had not cut: probably Goldman's $5 billion Global Equity Opportunities Fund, whose positions Goldman judged too big to sell, so its leverage had soared as its bets were hammered. On Friday it would either liquidate, hitting quants for a fifth day, or be recapitalized by its parent.
Late July Quant equity models falter; AQR trims leverage, then models recover in early August.
Fri 3 Aug Rating agency flags Bear Stearns; Cramer cries Armageddon on CNBC.
Mon 6 Aug AQR loses tens of millions in three hours; the deleveraging race begins.
Tue 7 Aug Losses double; AQR dumps billions; the Fed warns about inflation.
Wed 8 Aug Bush hopes for a soft landing.
Thu 9 Aug BNP Paribas freezes three funds; ECB adds $131 billion; Mendelson pins the last holdout on Goldman's GEO fund.
Fri 10 Aug Fed pledges liquidity; within minutes AQR's model turns profitable and it re-levers.
Mon 13 Aug Goldman announces a $3 billion recapitalization of the fund.
On Friday morning Mendelson ignored the Fed and watched his model, which within minutes was making money, all but announcing that the Goldman fund had been rescued and the liquidations were over. AQR re-levered as fast as possible to catch the rebound and recoup four days of carnage; the next Monday Goldman confirmed the $3 billion injection. The quake ended as abruptly as it began: Friday and Monday were great days, and most quants won back at least part of their losses.
Was the system at risk?
The debate outlasted the turmoil. Amaranth and Sowood had been saved by a fellow hedge fund, and freestanding funds had dodged the mortgage bullet; but the equity storm looked like a warning that the most sophisticated funds had lost control of their models, rocket scientists blowing up their own rockets.
Andrew Lo and Richard Bookstaber. Lo, an MIT professor with his own hedge fund, wrote a much-cited post-mortem; Bookstaber, an MIT alumnus who had worked at several big funds, pressed the warnings of his recent gloomy book. Leveraged multistrategy funds let losses in mortgages or credit force sales of equities, so distress jumps between sectors and the whole system is riskier. Too many quants chase small anomalies; crowding erodes returns, which drives funds to ever more dangerous leverage.
Mallaby. There is truth in it: in 1998 credit turmoil left quant equity funds untouched, and leveraged traders explain why the fire jumped the firewall this time. But crowding in a panic differs from crowding that depresses returns in good times. Lo's evidence, a simple contrarian strategy whose profits had decayed since 1995, was not central to the quake; the big money was in value bets, and the value-growth price gap had shown little sign of narrowing before 2007 (if strategies were overcrowded, Asness quipped, nobody had told the prices). Marek Fludzinski of Thales tested near-identical strategies and found one did not erode the other's returns. Any strategy feels crowded when investors rush for the exits, quant or not.
On systemic risk Mallaby also qualifies the critics. In the quake's first three days the model signals performed horribly but the broad market stayed calm; a household with savings in an index fund would have noticed nothing. The hard fall on the fourth day reflected the credit crisis more than quant selling. Unwinding market-neutral strategies is, by design, neutral for the market: each stock sold matched a short covered. Had Goldman not rescued its fund, deleveraging could have run longer and spread to other markets through multistrategy funds. As it was, the market punished overleveraged traders while the wider system took little harm; this, he says, is how capitalism should discipline its children, and no regulator could have done better. He closes with Mendelson's verdict: a lot of quants lost a lot of money and had an exhausting week, but it was not a public-policy issue.
The "hedge funds dodged subprime" story leaves out how some shorts helped manufacture their sucker. Paulson helped select the mortgage bonds for Goldman's Abacus 2007-AC1 CDO, which he then bet against; the SEC charged Goldman (not Paulson) in April 2010, and Goldman settled for $550 million that July. Funds such as Magnetar bought CDO equity while shorting senior layers, arguably prolonging CDO issuance. And Paulson's winnings needed counterparties who could pay, a chain that ran through banks and AIG, which taxpayers rescued. Mallaby's 1% average for asset-backed funds is accurate, but the line between hedge-fund and bank culture was blurrier than his four-reason contrast implies. A second caution: Mallaby's "no regulator could have done better" verdict rests on a quake that ended in five days because Goldman recapitalized its fund. The Lo-Bookstaber mechanism, forced selling that leaps across asset classes, operated on a far larger scale in autumn 2008. On Bear, note that the Fed's $29 billion was a loan to the Maiden Lane vehicle that took over Bear assets when J.P. Morgan bought the firm in March 2008, not an outright absorption; and Cioffi and his co-manager Matthew Tannin were acquitted of securities fraud by a jury in November 2009, though they later settled SEC civil charges.
Paulson never repeated 2007. Heavy losses on gold and bank stocks drained client money through the 2010s, and in July 2020 Paulson & Co. returned outside capital and became a family office. AQR, after a painful 2018-20 drawdown in value strategies, recovered strongly as value rebounded from 2022.
"How Could They Do This?"
In 2008 the balance of power between Wall Street and its hedge-fund clients flipped: Bear Stearns and Lehman Brothers begged short sellers for mercy, then blamed them for their fall. Mallaby's crisis story turns on Lehman's weekend in September 2008, when Hank Paulson weighed the odds of chaos like a policy maker instead of the payouts like a trader. Hedge funds were whipsawed by the panic, the short-selling ban and the loss of leverage, yet the industry fell 19% against roughly twice that for the S&P 500, and not one fund cost taxpayers a cent. Citadel's near-death, survived by funding discipline rather than a Fed lifeline, is his model of how finance ought to fail.
The rabbit pleads with the python
On a Thursday evening in March 2008, James Chanos left his midtown office to dine with Carl Bernstein of Watergate fame. Chanos felt a kinship with investigative reporters: his fund, Kynikos Associates, dug up dirt on companies, shorted them and profited when the truth came out. Short selling had been a thankless niche in the bull markets of the 1980s and 1990s, but the flat 2000s were glorious for it. Chanos had been among the first to see through Enron, had startled the world in 2005 by calling the insurer AIG "Enron-like," and in 2007 returned over 30%, mostly by shorting financial firms slow to own up to mortgage losses.
Walking up Madison Avenue he took a call from Bear Stearns, a bank whose relations with hedge funds had soured. Worried by Bear's mortgage losses, a string of famous funds had shut the margin accounts they kept there to finance their trades; because Bear borrowed against the assets in those accounts, its own funding was collapsing. Bear's executives suspected the funds of ganging up: shorting the stock and reinforcing the raid by pulling their accounts.
The caller was Bear's chief executive, Alan Schwartz. He thanked Chanos for staying a client and asked him to go on CNBC's Squawk Box next morning to say he still had money at Bear and that all was well. To Chanos the scene was absurd: the bank that had railed against short sellers now wanted one to vouch for it. When he asked whether things really were fine, Schwartz said Bear would report record earnings on Monday. Chanos snapped that Schwartz had just made him an insider without his asking, and that the real problem was clients shrinking their margin balances and causing a funding squeeze. Schwartz conceded the point "to some extent."
Chanos declined. As Wall Street's most visible short seller he could have propped up confidence, but he would not stake his credibility on a bank that might be imploding, and he was flying to the Bahamas at seven the next morning anyway. He went on to steaks at the Post House with Bernstein and his five partners. His irritation turned to fury at 6:30 a.m. on Friday, while Squawk Box was on air, when word spread that the Fed was brokering Bear's rescue: the margin-account exodus had become a classic bank run. Schwartz must have known the night before that Bear was finished. Chanos's later verdict was that Schwartz had been ready to throw him under the bus.
The tables turn
The exchange showed how completely the relationship had reversed. In 1994 Bear had sunk Askin Capital, forcing default and seizing much of its collateral; in 1998 it told Long-Term Capital Management it was finished, refused to clear its trades and slammed the door on the Fed-brokered rescue. In 2008 Bear was the one that was toast, and hedge funds held the power. Wall Street hummed with conspiracy talk: a vivid Vanity Fair account repeated a "vague tale" that the ringleaders had celebrated at a Sunday breakfast and plotted their next assault, on Lehman; one Lehman executive heard it was at the Four Seasons, with mimosas made from $350 bottles of Cristal.
Mallaby is unconvinced. The SEC investigated and prosecuted no one, and the scene itself defies belief: a real conspiracy would have been criminal, and its members would hardly incriminate themselves by chest-bumping in public the next weekend.
Lehman versus the shorts
Overwrought or not, there was no doubt that funds were shorting the banks and that Lehman was next. Short interest rose above 9% of its shares (nearly one share in ten borrowed and sold), and the stock was down more than 40% since January. Lehman accused the funds of "short and distort," and its combative chief Richard Fuld told the annual meeting that hurting the shorts was his goal.
Fuld asked Washington for help, for instance by restoring the defunct uptick rule, which barred shorting a stock while it was falling. In April the SEC's Erik Sirri asked him for evidence of collusion: a name, anything. Fuld gave none. His lieutenants had passed on rumors that traders were gunning for Lehman, but proving a conspiracy needs subpoena power, and Fuld thought it upside-down for the SEC, which had that power, to ask Lehman for the evidence. He then invited the TV pundit Jim Cramer to breakfast and pressed him for the names of people badmouthing Lehman; Cramer insisted there weren't any.
David Einhorn of Greenlight Capital, the boyish-looking manager who had shorted Chemdex and other dot-com froth, told a big New York investment conference that Lehman was underplaying its balance-sheet problems. It held $6.5 billion of dicey collateralized debt obligations but had marked them down only $200 million in the first quarter, a suspiciously small move given the slide in credit markets. It had told investors on a call that it would book a loss on hard-to-value "Level Three" assets, then reported a profit. Like a prosecutor, he urged regulators to make Lehman own up before taxpayers paid, and closed by observing that managements who complain about short sellers are usually distracting from serious problems. Lehman's shares fell almost 3% the next day and kept sliding; in June it reported a $2.8 billion second-quarter loss, and commentators credited Einhorn with forcing it to come clean.
Not everyone cheered. One Wall Streeter granted Einhorn was right but said the timing was wrong: two years earlier the crusade would have been heroic, but if he brought Lehman down, taxpayers would be the guarantors.
Hedge funds feel their oats
Short sellers were not the only ones riding high. In the week after the August 2007 quant quake, George Soros hosted Julian Robertson, Chanos and a few other heavyweights for lunch (striped bass, fruit salad, cookies) on his Long Island estate. The guests judged a full recession unlikely; Soros disagreed so strongly that he returned to active investing. Since Stan Druckenmiller's departure he had farmed out most of his fortune to outside managers; now, at 77, he took back the reins, ended 2007 up 32% and was the industry's second-highest earner.
Other macro traders thrived too, riding the implosion of the rich-world credit bubble and continued emerging-market growth. The typical book was short rich-world markets and the dollar, long emerging markets, oil and often other commodities. In the eighteen months to June 2008 the average macro fund rose about 17% (per Hedge Fund Research), a fine return in a financial crisis.
Activism
The swagger showed in activism: buying big stakes and demanding strategic change.
- Phil Falcone at the New York Times. The former ice-hockey player, whose Harbinger Capital had made a killing shorting subprime in 2007, bought 4.9% of the Times in early 2008 and demanded four board seats for allies. His thesis: dump the peripheral holdings (a slice of the Boston Red Sox, a NASCAR team, regional papers and TV stations) and build the core brand online. In March the Sulzberger family, holders of most voting shares, conceded two seats; Falcone raised his stake to 19%. In April, chairman Arthur Ochs Sulzberger Jr. told the annual meeting the company was not for sale, an implicit admission of how serious the threat was.
- Chris Hohn in Europe and America. Hohn ran The Children's Investment Fund, part of whose profits funded child-survival and AIDS programmes. High-minded but hardball, he bought into Deutsche Börse in 2005 and forced out its chief executive, telling him he was strong enough to seat Mickey Mouse and Donald Duck on the supervisory board. In 2007 he pushed ABN AMRO into selling itself to a trio of suitors. In summer 2008 he bought 8.7% of the US railroad CSX and sought board seats. CSX fought back: friends in Congress hauled his people before a committee, it sued in New York alleging breaches of SEC disclosure rules, it painted him as a British raider on strategic infrastructure, and in June 2008 it held its annual meeting in a tent in a steamy Louisiana rail yard, where one shareholder reminded the British that America had saved them from the Huns twice. Despite delays and fudged voting, Hohn's candidates won four seats by mid-September.
Pretending to be banks
The funds' reach extended to emerging markets. Kazakhstan, known on Wall Street mainly through Borat, was growing 8-9% a year on oil and running an export surplus, so its currency was a near-certain bet to rise against the dollar. But oil revenue meant the government issued no debt, so foreigners had no sovereign bonds to buy. From about 2003 hedge funds found a workaround: they lent directly to Kazakh banks, gaining currency exposure plus a higher interest rate. They repeated the trick worldwide, lending by 2008 to everyone from Brazilian coffee exporters to Ukrainian dairy farms; Kiev's top hotel charged €800 a night.
Mallaby spells out the flaw. To ride the currency, the funds took on the default risk of coffee exporters and dairies; they were acting as commercial lenders without a bank's capacity for due diligence. A second-tier Russian firm could borrow without anyone visiting its offices, reading its books or working out how to recover the loan in bankruptcy. Collateral might be dairy herds half a world away; the idea of a fund turning up in the Ukrainian countryside to collect live cows was farcical. In the mid-2000s nobody cared: hedge funds could do no wrong, loan officers were dull, and the subprime crisis, which exposed banks' risk management, only confirmed the verdict.
In April 2007, German regulators in Washington asked Chanos about hedge funds and financial stability; he told them they were looking the wrong way, because the danger was the banks. By mid-2008 he was plainly right: Bear had failed, Lehman was under siege, and Fannie Mae and Freddie Mac, which had a whole government agency dedicated to overseeing them, were about to fall into taxpayers' arms. The unregulated funds were meanwhile stalking targets from the Times to Kazakhstan and dancing between the land mines.
Paul Tudor Jones reads the charts
By summer 2008 Paul Tudor Jones had moved his firm into a mansion on a broad lawn outside Greenwich, courtly and brash like its owner: twin curving staircases, marble and antique rugs, deer antlers and a stuffed bear (a 50th-birthday joke) in his office, a huge mural of a killer shark guarding the approach, six screens set into the panelling, and a yellow legal-pad page pinned by the window reminding him to look for a trending market.
In late June he decided the trend was down. The S&P 500 had jumped in April and kept rising in May, but Jones saw a sucker's rally in the greatest credit bubble ever. He studied every precedent (Japan 1989, the US in the late 1920s, Sweden in the 1990s) for price patterns. At 3:05 a.m. on Saturday 28 June he sent colleagues a eureka email headed with an apology for alarmism: the weekly S&P 500 now looked like the daily Dow of 1987.
- Oil as the new bond rout. In 1987 falling bonds spooked stocks because higher rates meant less money for equities; in 2008 dear oil kept the Fed from cutting rates and drained liquidity from asset markets. Same plot, different characters.
- Fear the absence of fear. Inverting Roosevelt, he worried about the lack of fear in options markets: complacent sentiment had plenty of room to sour.
- A record signal. The Dow had closed at a 250-day low while sentiment was bullish, something that had never happened in the indicator's 21-year history, 1987 included.
- Chart overlay. The two indices, on different time scales, fitted almost perfectly, which convinced him the third quarter would be the ugliest in history.
Over the next two months his analogies wandered: sometimes the chart looked like the 2001 recession, sometimes like 1987. What counted was the instinct beneath them. Jones saw an asymmetric bet: a leveraged system in a credit crisis is like a high-wire artist in a storm, who may fall a long way but will certainly not float upward.
He had been building to this for years. He had watched leverage grow exponentially since the 1980s; during the dot-com mania he wrote to Alan Greenspan urging higher margin requirements; in the mid-2000s a senior Fed official phoned him regularly for his read on risks, and he kept answering that debt was piling on debt, so the higher the pyramid the greater the catastrophe when some part cracked. Tudor was part of the edifice. Each year, reviewing pay late at night with president Mark Dalton, Jones would ask what would happen if "this enormous contraption" ever had to be unwound; Dalton said he did not want to think about it.
The trader's question Washington forgot
At about 6 p.m. on Friday 12 September 2008, a column of dark cars delivered Wall Street's chief executives to the New York Fed, where Treasury secretary Hank Paulson, New York Fed president Timothy Geithner and SEC chairman Christopher Cox told them there would be no public money to save Lehman. Paulson, the politician of the trio, felt he had risked enough taxpayer money already: he had backed the rescues of Bear and of Fannie and Freddie, and Senator Jim Bunning (Republican, Kentucky) had accused him of acting like China's finance minister. Lehman offered a chance to draw a line and teach bankers to bear the cost of their errors. Markets might react badly, but Lehman had been in the emergency ward for months, so its counterparties had presumably prepared. The team would hunt for a private buyer; failing that, it would step aside and bet that failure would not cause chaos.
Had Paulson's team thought like hedge funds, Mallaby argues, they would not have made the call that produced the worst financial freeze since the 1930s. They repeated the Bank of England's 1992 trap: they asked how likely chaos was, not what each outcome would pay. A policy maker who sees a fifty-fifty chance of calm may run the risk to teach a lesson. A trader sees that in the calm world markets are flat and in the chaotic world they crater, so shorting everything costs nothing in one case and pays hugely in the other. Every rational fund bets on collapse, and because they all do, collapse becomes inevitable.
Jones needed no detailed census of who had prepared: everyone knew Lehman would be "the kickoff" for a big fall. Lehman sat in a bewildering daisy chain of interlocking trades; the system was leveraged to the eyes; and the symbolism of a firm that had survived the Depression and two world wars going under would scream that nothing was safe. People would wonder whether even their own son was good for a loan, Jones said, and would shoot first and ask questions later.
Sunday: the London surprise
News of the bankruptcy leaked around Sunday lunchtime. Funds that had positioned for bad news found they were not fully protected. American managers belatedly grasped that Lehman's London arm would go bankrupt under British law, so client accounts that US rules would have kept segregated and safe would instead be frozen. Lawyers raced in from the Hamptons to work out which regime covered their assets, outside counsel were put on speed dial, panicked investors demanded exposure numbers, and nobody had clear answers. LTCM veteran Eric Rosenfeld, hearing the news on his car radio, was so shocked he nearly hyperventilated: "How could they do this?"
Monday: Tudor trapped
On Monday 15 September Jones lived his own extremes. His personal book was perfectly placed and rode the market down as the S&P 500 fell 4.7%. Yet it was the worst day of his professional life. Tudor's request the week before to pull its remaining assets from Lehman had arrived a day late, leaving $100 million frozen in London; Tudor wrote it all off, and that was the least of its problems.
Mallaby calls Tudor's real error as egregious as Paulson's. Its emerging-market credit team had built a giant portfolio of loans (Kazakh and Russian banks, Ukrainian dairies) that formed a significant part of the flagship BVI fund. As with Brian Hunter at Amaranth, the opportunity had been real at first: currency exposure, interest rates that more than paid for default risk, and, as other funds crowded in, rising prices that boosted returns and drew more capital from Tudor's bosses. After Lehman the true risk surfaced: if storied US banks could not raise money, banks in Kazakhstan and Russia were bound to suffer. The loans lost about two thirds of their value, costing Tudor more than $1 billion.
Worse for a trader who never opened a position without a stop-loss, the loans were utterly illiquid: nobody would buy loans at any price. Jones called it the worst moment of his life. He had always wondered how anyone in 1929 could have been so boneheaded as to lose everything; now he saw that the victims had not merely been long the market, they had held things they could not get out of. Tudor had been up 6-7% for the year before Lehman, impressive in a falling market; it finished down 4%, imposed gates that suspended investors' access to their money, and Jones promised to stick to the liquid markets he knew. The age of the diversified alpha factory, Mallaby suggests, was perhaps receding.
The investment-bank model dies
Lehman's failure ended the modern investment-bank model. Lehman and its peers borrowed billions in short-term money markets to buy assets that could not be sold quickly. When the crisis hit, short-term lending vanished, and fear of a crunch was self-fulfilling. Commercial banks are protected against such runs by deposit insurance and access to the Fed's emergency lending; investment banks had neither, yet, believing themselves invincible, they had behaved as though they did.
Sun 14 to Mon 15 Sep, 1 a.m. John Thain, shuttling between the New York Fed and Ken Lewis of Bank of America, sells Merrill Lynch (and its "thundering herd" of nearly 17,000 brokers) for $29 a share, after rebuffing an offer worth $90 almost a year earlier. A paper likens it to Wal-Mart buying Tiffany's.
Mon 15 Sep Lehman files; the S&P falls 4.7%.
Tue 16 Sep AIG is rescued with an $85 billion government loan after its credit default swaps lose billions.
Wed 17 Sep Rumors of AIG exposure drive Morgan Stanley down 42% by mid-afternoon; John Mack blames short-selling conspirators; Senators Chuck Schumer and Hillary Clinton urge an SEC ban.
Thu 18 Sep Cox doubts, but Paulson insists they must be saved now; at about 1 p.m. London's FSA bans shorting 29 financial firms for 30 days; some three dozen Goldman traders stand, hands on hearts, for "The Star-Spangled Banner." That evening the SEC bans shorting of about 800 financial companies.
Sun 21 Sep Morgan Stanley and Goldman Sachs come under the Fed's protection as bank holding companies.
AIG, in Mallaby's account, showed what happens when an ambitious trading operation sits inside a well-capitalized firm: the traders felt invulnerable and sold protection against default on all manner of bonds. With Bear, Lehman and Merrill gone, Morgan and Goldman's reliance on short-term funding and extreme leverage made them obvious run candidates, and their share prices stayed pinned to the top of CNBC's screen in what traders called the "death watch." Hedge funds rushed to pull assets from Morgan to avoid another Lehman-style trap.
The short-selling ban
Morgan and Goldman then followed Lehman's summer playbook and lobbied for curbs on short sales of their stock. The demand was awkward: their proprietary traders happily shorted, and both ran thriving prime-brokerage arms that financed and executed hedge funds' short sales. In the frenzy, neither stood on principle.
The ban bought the two banks brief breathing room but was, in Mallaby's words, a frontal assault on hedge funds' viability. Stock pickers lost hundreds of millions; one Tiger cub said the game had switched from chess to rugby at halftime. The claim that it protected the system was a stretch, since the list included internet incubators and retailers. Nor did it end the banks' lobbying. In good times Morgan and Goldman had prized their freedom from Fed oversight; now they made a swift U-turn, asking to be brought under the Fed's wing for guaranteed emergency credit. On 21 September they got it, and their vulnerability ended.
Citadel's deathwatch
Around the time Morgan was plunging, Ken Griffin worked out the odds that Citadel might fail. He had set out to build a firm like Morgan or Goldman and shared some of their weak points. He had leveraged capital more than 10 to 1: far tamer than the investment banks' typical 30 to 1, but well above the single digits normal for hedge funds. And because Citadel had issued a small amount of five-year bonds, a market existed in credit default swaps on its debt, giving traders a way to broadcast doubts about it.
Griffin sketched a quick probability tree:
- 50%. His estimate of Morgan Stanley's chance of survival.
- 95%. Chance Goldman follows if Morgan fails.
- ~100%. Chance Citadel collapses if Goldman fails, since forced selling by the two banks would wreck the value of its holdings.
- ~55%. Citadel's overall survival odds as Mallaby reports Griffin's sum: twenty years of work riding on firms he could not influence.
Kicked by the government
Being classed as a hedge fund, Citadel got no access to emergency Fed lending; instead the ban hit it hard. It held a giant portfolio of convertible bonds, hedged by shorting the issuers' shares on the view that the embedded options were underpriced relative to the stock. With shorting barred, new convertible positions could not be hedged, demand for convertibles collapsed and Citadel took shocking losses. Its main funds were down 20% for September, and as equity shrank, leverage rose. Since 1990 Citadel had grown from nothing to $15 billion of assets and 1,400 employees; now its survival was in doubt.
Griffin's team faced a dilemma. Cutting leverage by dumping convertibles would show rivals he was desperate and invite them to squeeze his portfolio; doing nothing would soon leave him unable to meet margin calls. A third danger loomed: worried counterparties could mark down the value of Citadel's derivatives, forcing it to post cash until it ran dry, exactly what brokers had done to Askin, LTCM and most failing institutions since.
- Quiet sales. It shed assets outside its main strategies, raising capital without advertising distress.
- Exchanges over dealers. It closed bilateral derivatives with other firms and in some cases replaced them with exchange-traded contracts; an exchange, unlike a bank, would not squeeze it.
- Back office as weapon. Where contracts could not be closed, its state-of-the-art infrastructure, data feeds and models tracked the daily value of every derivative bought from a bank, so counterparties found it harder to push marks against it. Recalling LTCM's dream of "Salomon without the bullshit," a staffer joked that the back office had been Salomon's true edge and the LTCM partners were the bull.
- Matched funding (the key advantage). Treasury mapped how long each asset took to sell and borrowed with the same mix of maturities: overnight money only for overnight-sellable assets, harder-to-sell holdings backed by harder-to-yank loans, the unusual five-year bonds, and bank lines locked for up to a year. In theory, no crisis could force a death spiral of fire sales.
Practice was shakier. Citadel had planned for a crisis, but not one this big. Less securely financed funds, hit by margin calls, dumped convertibles and other positions, inflicting further losses on Citadel. Rumors multiplied: margin calls, the Fed phoning Citadel's counterparties about their exposures. The Fed was in fact calling around Wall Street telling banks not to pull loans; whether that saved Citadel or fed the rumor mill is debatable. On some October days CNBC parked a truck outside the Citadel Center. A new deathwatch had begun.
Treasury head Dan Dufresne took a call on his way to the train from a European bank's New York office: rumors were gathering pace in Europe that the Fed was in Chicago liquidating Citadel. He switched to a cab rather than discuss it on a crowded train. By the time he reached his desk three more callers had relayed the same story; COO Gerald Beeson had been fielding European desks since 5 a.m. They suspected journalists were bouncing the rumor off every bank in London.
Then James Forese, Citigroup's head of capital markets, rang Griffin, having heard he was in Washington seeking a bailout. Citadel's CDS were pricing more distress than Lehman's on the eve of bankruptcy. Griffin guessed Forese was checking whether he was alive or had money; Forese said he was offering to handle any liquidation discreetly. Griffin replied that Citadel was losing a lot of money but had a lot of liquidity: locked-in funding meant no margin calls, the back office kept marks honest, and it had sold assets more proactively than Bear or Lehman. But the rumors were persuading markets that a convertible fire sale was coming, which drove prices down further. Griffin did not want a conference call that would confirm the worst; Bear and Lehman had held calls, to no avail. Forese suggested a routine call for bondholders: unremarkable in form, and credible because misleading bondholders could land Griffin in trouble with regulators.
At about 10:30 a.m. the inner circle listed six points on a whiteboard, chose Beeson as speaker and added "Speak slowly." A big company like Coca-Cola arranged about 250 lines for a results call; Citadel had only a handful of bondholders, and someone proposed 500 lines; Beeson said 1,000. The 3:30 p.m. call could not start on time: well over 1,000 traders, investors and reporters tried to dial in, some trading floors piped it over loudspeakers, and after a 25-minute delay those who got through heard grating techno hold music. Beeson admitted the two flagships were down 35% but said financing was secured and an $8 billion credit line untouched; general counsel Adam Cooper held up a sign telling him to slow down. Griffin added that Citadel would prosper in the new era of finance. After twelve minutes it was over. Griffin held a staff town hall, Beeson repeated the message to reporters all afternoon, and next day the CNBC truck was gone.
The reckoning, and what it proved
Losses continued for weeks. By year-end Citadel's two flagships were down 55%, and the $9 billion that evaporated equalled at least two LTCMs. Yet Mallaby calls it a model of how finance should work: investors who had risked capital and been rewarded for years took extraordinary losses, as they should; the system was not destabilized; and taxpayers threw no lifeline. Leveraged trading firms with billions under management do not necessarily need rescues when markets go berserk, because careful liquidity management can substitute for the Fed's safety net. The investment banks built castles of leverage on short-term loans; sharing the paranoid culture of hedge funds, Griffin leveraged more cautiously and borrowed longer. So an upstart Goldman imitator proved better for the system than the real Goldman, and incomparably better than Bear or Lehman.
Citadel's fate mirrored the industry's. Funds had danced through the minefield until September, then were whipsawed: their strategies relied on short selling, which was banned; their portfolios relied on leverage, which dried up as brokers hoarded capital. By the end of 2008 most funds had lost money, almost 1,500 had gone bust and many titans were justly deflated. Even so, in the worst year of their history they proved their worth: the industry fell 19% while the S&P 500 fell about twice as much, and unlike banks, investment banks and home lenders, hedge funds imposed no costs on taxpayers or society.
Titans fall
That policy lesson was lost in the spectacle of the fall; the titans had flown so high that their descent was mesmerizing.
| Titan | High point | 2008 fall |
|---|---|---|
| Phil Falcone (Harbinger) | +43% in first half 2008; stalking the Times | Lost over $1 billion in one September week (Lehman plus shorting ban) |
| Fortress | Floated Feb 2007: $10.7bn paper wealth for its chiefs | Assets down almost a third by Dec 2008; two dozen portfolio managers fired |
| Meriwether and Scholes | New funds from 1999 did well for years | Both near the precipice by end-2008 |
| Chris Hohn (TCI) | Wins at Deutsche Börse, ABN AMRO, CSX | -42%; admitted, as if surprised, that activism is hard |
Redemptions and gates
Investors stampeded for the exit. Hedge funds had fallen far less than the market, but clients expected them to make money in any market, as if they had abolished risk rather than managed it. Even macro funds, which gave back only a sliver of earlier gains after Lehman, faced $31 billion of net redemptions because everyone was pulling cash from everyone. Funds answered as Tudor had, suspending quarterly withdrawals and gating money. Sometimes the gates helped investors, averting ruinous fire sales and coming with suspended management fees. Sometimes they were an outrage: one prominent fund reportedly charged a departing investor an early-redemption fee, then blocked the redemption, kept the fee and went on charging management fees.
- -12%. Funds with under $1 billion in assets.
- -16%. Funds with $1-10 billion.
- -27%. Funds with more than $10 billion.
- $36bn to $20bn. Tom Steyer's Farallon, start of 2008 to 2009, a big alpha factory stuck in losing positions when liquidity vanished.
The bigger the fund, Mallaby observes, the more peremptory and arrogant its managers tended to be, and frequently the worse its performance; nimble boutiques, closer in spirit to the Steyer of a decade earlier, often escaped with scratches.
More money than God
The mystique survived. Hedge funds remained repellent and alluring, wizards of capitalism's favourite pastime, the unabashed pursuit of money. In November 2008 five barons (George Soros, Jim Simons, John Paulson, Ken Griffin and Phil Falcone) testified to the House in what promised to be a show trial. Instead Representative Elijah Cummings (Democrat, Maryland) relayed his neighbour's awed question about how it felt to face five people with "more money than God."
"No costs on taxpayers" holds directly but not indirectly: the funds' own prime brokers (Morgan Stanley, Goldman) survived only through the Fed's backstop and the AIG rescue, and Citadel rode out October in markets that the Fed and Treasury were propping up with trillions (Mallaby himself notes the Fed was phoning banks not to pull loans). Griffin's tree as described implies about 52-53% survival (50% + 50% x 5%), a bit below "around 55%." The short-selling ban Mallaby attacks was later called a mistake by Cox himself, and academic studies found such bans hurt liquidity without supporting prices, which supports his view. The conspiracy charge is dismissed mainly on the absence of prosecutions, which proves less than it seems given the evidentiary bar.
Einhorn was vindicated in March 2010, when the bankruptcy examiner's report revealed "Repo 105," which Lehman used to shift about $50 billion off its balance sheet at quarter-ends. The SEC adopted an alternative uptick rule (Rule 201) in 2010. Citadel recovered fully and became the archetypal multistrategy platform (see chapter 14's update).
Scarier Than What?
Mallaby's verdict is a policy argument: hedge funds are no saints, but their incentives and culture are less flawed than those of other financial firms, and above all they are small enough to fail. Since too-big-to-fail banks cannot be left to die and regulation is essential but fallible, governments should push risk into institutions that cost taxpayers nothing; the prescription for hedge funds boils down to two words, "Don't regulate." The exception is size: a few giant, leveraged or listed funds should be supervised like the investment banks they resemble.
Not angels: from Douglas Aircraft to Galleon
Douglas Aircraft, 1966
Early in the first hedge-fund boom, in 1966, bankers at Merrill Lynch, Pierce, Fenner & Smith were underwriting a convertible debt issue for Douglas Aircraft. They learned that Douglas was cutting its earnings forecast to $3.50 a share, about a quarter below projections; then that earnings might come in under $3; then that they would be zero. As inside advisers they were strictly barred from passing this to Merrill's brokerage clients, but, not for the last time at an investment bank, internal controls failed. On 21 June the news reached Lawrence Zicklin, the Merrill broker for hedge-fund clients.
Zicklin had a direct wire to Banks Adams, a segment manager at A. W. Jones, installed a year earlier on the understanding that he would pass on everything the moment he knew it; the Jones men paid generous commissions and expected service. According to an SEC administrative proceeding, Adams at once shorted 4,000 Douglas shares; Richard Radcliffe, the ex-Jones man who had started a fund with Barton Biggs, shorted 900; and John Hartwell, a mutual-fund star running a model portfolio for Jones, dumped 1,600.
Next morning a bullish Wall Street Journal piece on aircraft makers kept ordinary investors calm, and Douglas actually rose a dollar. But that day the best-connected money managers lunched with Bob Brimberg, a legendary broker as wide as a truck (a scale his partners installed by his desk did not help), who plied guests with meatballs, corned beef and cocktails and then demanded to know what they were buying and selling; lying meant no further invitations. On 22 June someone mentioned Merrill and Douglas, and the guests rushed to phone Merrill for confirmation. By the end of the next day 13 Merrill clients had sold 175,800 Douglas shares; six were hedge funds, a striking tally for a still-obscure industry. A long SEC inquiry pushed several funds into expensive settlements.
Galleon, 2009
Almost half a century later the pattern recurred around Raj Rajaratnam, the Sri Lankan-born head of the Galleon Group, a man of Brimberg-like girth whom the SEC's enforcement chief called a master of the Rolodex rather than of the universe: no special sauce, many special sources. A 2009 criminal complaint alleged that his contacts tipped him off about:
- Polycom. Unexpectedly good earnings at the technology maker, allegedly worth a quick half-million-dollar profit.
- Hilton. Blackstone's coming bid for Hilton Hotels, allegedly worth $4 million.
- Google. Disappointing earnings known in advance, allegedly worth $9 million.
The network reportedly ran to a senior Intel executive and a McKinsey director who leaked in return for a share of the takings. Prosecutors said the conspirators kept discarding mobile phones like a drug gang; one allegedly bit a SIM card in half after a trade. Rajaratnam pleaded innocent.
Judge against rivals, not perfection
Mallaby concedes that hedge-fund managers are not angels. The record includes Michael Steinhardt's collusive block trading and David Askin's nonexistent prepayment model. Regulators in the Douglas era worried that fund investors included rainmakers and senior executives of public companies who might leak secrets to their money managers; Galleon seems to vindicate that fear two generations on. Douglas also showed how the huge commissions funds pay brokers create room for abuse, which very likely continues. Every industry has its criminals and charlatans.
But the right benchmark is rivals, not perfection. The case for the industry is that its incentives and culture are less flawed than those of other financial firms. There is no evidence hedge funds commit fraud more often: a 2003 SEC inquiry looked and found none. A freestanding fund is arguably less likely to receive illegal tips than an asset manager inside a big bank, which sits amid profitable information flows from corporate clients and trading partners. Inside a conglomerate, a secret need only pierce the Chinese walls between underwriters or merger advisers and proprietary traders; to reach a hedge fund it must also leave the building.
Most charges against hedge funds, Mallaby says, would be better aimed at other players. The heart of his case is one phrase: while much of the financial system has proved too big to fail, hedge funds are generally small enough to fail, and when they blow up they cost taxpayers nothing.
The vicious cycle of the safety net
After the bust of 2007-2009 and the meandering regulatory debate that followed, this advantage is hard to overstate. The implosion of Lehman and AIG froze global credit and caused the steepest recession since the 1930s. Bailouts deepened the rich world's fiscal crisis and sped the shift of economic power to emerging economies.
- 43% to 70%. US national debt as a share of GDP, before the crisis versus projected for 2010, while public debt in China, India, Russia and Brazil stayed roughly flat; Greece and Ireland teetered on the brink.
- $10 trillion. IMF estimate of cash injections, debt guarantees and other support for too-big-to-fail institutions in the big advanced economies.
- $13,000. That support per citizen of those countries.
- December 2009. Barack Obama said he had not run for office to help "fat cat bankers"; help them he did, and populist anger followed.
The sums spent rescuing well-heeled financiers damaged the legitimacy of capitalism. Worse, no leader knows how to stop the behemoths fleecing the public again: Mallaby's darkest point is that the crisis is likely to be repeated.
The catch-22
Government support for finance is trapped. Many institutions really are too big to fail: refuse to rescue them to protect taxpayers, and the meltdown costs taxpayers more, as post-Lehman showed. Yet every time government pays for financiers' risks, it lowers the cost of risk and weakens the incentive to curb it. Without deposit insurance, depositors would shun risk-hungry banks or demand higher rates; without governments acting as lender of last resort to banks (and now investment banks), bank bondholders would monitor soundness harder. Mallaby does not argue for abolishing either, since letting big firms fail is too costly; but insurance breeds risk, and risk forces more insurance.
Banking history shows the cycle. Over the past century governments have widened last-resort lending, eased its terms and extended deposit insurance to more customers, and risk taking has grown in step. Since 1900 US banks have tripled leverage from about 4 to 12, taken more liquidity risk by funding long-term assets with short-term borrowing, and shifted resources into high-risk proprietary trading. In 2007-2009 governments stretched deposit insurance further, guaranteed money-market funds that were supposedly uninsured and pumped emergency liquidity into every corner of the markets, all of which will invite more recklessness. The cycle, Mallaby warns, runs until governments are bankrupt.
Two standard answers, both flawed
Let them fail. Make financiers pay for their own risks and they will be prudent: had investors absorbed Bear Stearns's bankruptcy instead of a Fed-subsidized rescue, they might have braced better for Lehman. But this denies the problem exists: some firms truly are too big to fail, and they and their lenders know the state will blink whatever it says. Even the soft version (save the giants, let medium-sized firms like Bear go) is unrealistic, because regulators lean toward intervention to avoid a disaster on their watch: human nature, which cannot be changed.
Regulate harder. Capital requirements, liquidity monitoring and limits on proprietary trading can break the arms race, and as Mallaby finished the book rich-world governments were about to try. But no regulatory system is foolproof: 2008 humbled many agencies applying many rule books in many countries at once.
The roll call of regulatory failure in 2008 is Mallaby's evidence.
| Institutions | Overseer and rules | Outcome in 2008 |
|---|---|---|
| US deposit-taking banks | Fed, FDIC and two smaller bodies; Basel I capital standards | Did miserably |
| US investment banks | SEC, its own risk limits | Two failed, one sold to avoid failure, two rescued |
| Fannie Mae, Freddie Mac | A dedicated government department | Nationalized |
| AIG | Slipped through the net | Crashed; $85bn rescue |
| Money-market funds | Nominally the SEC | Needed an emergency government guarantee |
| UK banks | FSA, seen as a model regulator | A string of costly disasters |
| Continental European banks | Updated Basel rules (Basel II) | Made no difference |
When so many regulators fail together, it is hard to believe the fix is simply to manage, staff or de-ideologize some key agency differently. Regulation is genuinely difficult, for reasons Mallaby lists:
- Capital against what? Measuring capital against raw assets is meaningless, since assets range from scary mortgage bonds to hedged government bonds; it must be measured against risk-weighted assets, which means defining risk and fighting over the definition.
- Funding structure. Capital alone does not decide resilience: short-term borrowing makes a firm more vulnerable to a loss of confidence than long-term borrowing.
- Liquidity of holdings. Trading illiquid instruments (complex mortgage securities, loans to Kazakh banks) is riskier than trading on an organized exchange, another judgment call.
- Asymmetric politics. Regulators are like air-traffic controllers, ignored when all goes well and excoriated after a disaster; at every step their desire for safety collides with firms' appetite for risk, and given big firms' brainpower and political influence, they win some of the arguments. Regulation will be softer than it should be.
The third option: encourage hedge funds
If behemoths cannot be allowed to fail and regulation is both essential and fallible, policy makers should pursue a third course: drive financial risk into institutions that cost taxpayers less. That means encouraging firms that are not too big to fail, shrinking the share of risk the state must backstop, and favouring institutions whose own incentives to control risk are strong enough that regulators carry less of the load. Part of the answer, Mallaby says, is hiding in plain sight: governments must encourage hedge funds.
They are no cure-all: they will not take deposits, underwrite securities or lend to small companies. But at managing money without endangering the system, they have proved themselves. Between 2000 and 2009 about 5,000 hedge funds went out of business and not one needed a taxpayer bailout. Because they mark every asset to market and live in fear of margin calls, they generally monitor risk better and recognize setbacks faster than rivals; badly hit, they tend to liquidate and close before secondary effects reach the system. So governments should let them thrive, multiply and absorb more risk from too-big-to-fail rivals; and since the aim is more of them, burdening them with oversight is counterproductive. Hence the two words: Don't regulate.
Objection one: do they rip off their clients?
Since the 1960s the 20% performance fee has excited envy and alarm as a heads-I-win, tails-you-lose invitation to wild punts. Academics add a subtler charge: fees reward smooth returns that hide blowup risk. A fund takes $100, puts it in the S&P 500, and earns, say, $5 a year selling options to investors who want insurance against a crash. Over five or ten years a crash probably will not come, so the manager beats the index by 5 points a year and is hailed a genius, until the crash wipes the fund out. If fees reward such trickery, rogues will try it, clients who deserve consumer protection will be hurt, and blowups will destabilize markets.
Plausible, says Mallaby, until you look at the alternatives.
Versus buying bank shares
Owning a hedge fund is safer than owning an investment bank's stock, which excites no controversy. Most hedge-fund managers have their own wealth in the fund alongside clients. And bank "performance fees" are far larger: investment banks have paid out fully 50% of net revenues as salary and bonuses. The comparison is imperfect but telling: bank pay creates a bigger incentive to shoot for the moon, damaging stability when they miss. A hedge-fund fee formula is fixed in advance; banks decide the payout each year on management's say-so, without consulting ordinary shareholders.
Versus mutual funds
At first sight hedge funds are scandalously expensive: mutual funds charge about 1%, hedge funds 1-2% plus about 20% of profits. But the right comparison separates alpha (returns from skill) from beta (returns from market exposure). Index exposure costs about 10 basis points, so an active mutual fund charging 1% is effectively charging 90 basis points for alpha. Study after study finds active managers as a group do not beat the market. They charge 90 basis points and deliver nothing, so their fee per unit of alpha is infinite.
Do hedge funds actually deliver alpha?
The book has shown many sources of hedge-fund profit: trading against central banks that are not in markets for profit; buying from price-insensitive forced sellers; supplying liquidity to big institutions; spotting asymmetric bets; even simple stock picking like the Tiger cubs (see the appendix). But a history of pioneers is biased toward winners; perhaps the average fund loses money, or fees swallow its alpha.
Mallaby grants that a precise answer is elusive, because funds earn partly by taking exotic risks that are hard to measure. Funds-of-funds report lower returns than the funds they invest in, by more than fees explain, suggesting that reported returns are often exaggerated. Still, the tentative bottom line is surprisingly positive.
Roger Ibbotson (Yale School of Management), Peng Chen (Ibbotson Associates) and Kevin Zhu (Hong Kong Polytechnic University) took performance data for 8,400 hedge funds from January 1995 to December 2009. They corrected for survivorship bias (counting only funds alive at the end leaves out those that blew up, overstating returns) and backfill bias (funds usually start reporting after a year of excellent profits, so including those bonanzas flatters the industry). After adjustment the average fund returned 11.4% a year gross and 7.7% after fees, and that net return included 3 percentage points of alpha: profits beyond market exposure, with alpha left over for clients despite the fees.
Versus private equity
Both are "alternatives": private partnerships, leveraged, charging performance fees, and increasingly running each other's products. But the promise differs. Hedge funds bet that the market has mispriced securities or currencies, a game of numbers and psychology; private equity promises to improve companies, installing new chiefs and revamping everything from advertising budgets to middle-management incentives. Here too beta (exposure to corporate profits, like an index) must be separated from alpha. Three influential studies find venture capital generates alpha of about 4-5% a year, while buyout funds do little better than the S&P 500. And private equity locks up investors' money for as long as a decade.
The customers' yachts
Hedge funds are no substitute for other vehicles: for ordinary savers, cheap index funds remain best. But for large investors they compare well with most rivals, are no more prone to insider trading or fraud, and deliver real value. To Fred Schwed's old question about where the customers' yachts are, Mallaby answers: ask Harvard and Yale, whose endowments returned 8.9% and 11.8% a year from 1999 to 2009, credit crisis included. Hedge funds help pay for university science buildings and merit scholarships and swell philanthropies from the Open Society Institute to Robin Hood. If they also widen the gap between the super-rich and the rest, the remedy is progressive taxation, not smothering their trading.
Objection two: do they drive prices to extremes?
A long-standing charge is that trend-following funds push prices to illogical extremes and destabilize economies. Mallaby's first answer is that most hedge funds push prices toward rational levels:
- Arbitrageurs. Including the much-maligned fast statistical arbitrage funds, they close gaps between related prices.
- Stock pickers. A Julian Robertson-style fund buying cheap and shorting dear moves shares toward fair value and steers capital to firms that will use it best.
- Commodity traders. Reacting quickly to gluts and shortages sends price signals that force healthy adjustment: bidding up oil on news of an African coup tells motorists to economize before unsustainable consumption drives prices even higher.
He concedes markets overshoot and trend followers can contribute: if hundreds of traders pile on and push oil needlessly far, they hurt consumers and companies and set up a destabilizing correction.
Rational. Oil's 2007 rise from $61 to $96 a barrel was probably a fair response to booming emerging-market demand. The Soros-Druckenmiller sterling trade was rational, since Germany's post-unification high rates made the peg untenable. Thailand's 1997 devaluation was too: its growing trade deficit was incompatible with the peg.
Overshoot. Oil's climb to $145 in the first half of 2008 probably exceeded the fundamentals. Indonesia in 1997, with a small trade deficit and a flexible exchange rate, suffered a far bigger devaluation than Thailand because political instability caused panic.
Would regulation tame the overshoots? Restrictions would curb contrarian trading as well as trend following, and there are no data on the net effect. But Mallaby judges that, despite the trend-following tradition from Commodities Corporation to Paul Tudor Jones, regulation would probably worsen overshooting, because hedge funds are structurally more likely to buck trends than other investors.
- They can short. Unlike some institutional investors, they are free to go short as well as long.
- Absolute returns. They are judged on money made, whereas mutual-fund managers measured against consensus-reflecting indices must be wary of bucking conventional wisdom.
- Performance fees. Contrarian conviction needs heavy research, and fees supply the money and motive: John Paulson bought the best house-price database, paid a technology firm to warehouse it and hired extra analysts before taking on the mortgage-industrial complex.
Even the trend follower breaks trends
Jones follows trends because, contrary to efficient-market theory, investors absorb news gradually: pension funds, insurers, mutual funds and individuals each move on their own timescales, so prices adjust in steps. Precisely because he understands momentum, he senses when it has gone too far and looks to profit from the reversal. Hence his short before the 1987 crash, his bet against Tokyo in 1990, and his warning in summer 2008, weeks before the bubble burst, that oil was a huge mania that would end badly. Once a trend loses touch with fundamentals, the most famous trend follower becomes a trend breaker.
The same holds for the industry. In Europe's currency crisis Soros and Druckenmiller did not attack every currency: they made money betting that the French franc would hold. In Asia, funds helped force Thailand's devaluation because the peg was illogical, but the rupiah's crazy crash was driven by Indonesians rushing money abroad; the Soros team pushed back, unsuccessfully, and Tom Steyer's Farallon helped start the turnaround by buying Bank Central Asia. Funds do sometimes herd: in 1993 Michael Steinhardt rode a red-hot bond market into a bubble and paid heavily in 1994. But their structure makes them more contrarian than other investors, so the regulatory lesson is contrarian too: the best way to damp trend following is to let hedge funds go about their business.
The persuasive case: size
In sum, hedge funds are not especially prone to fraud, offer a partial answer to too big to fail, deliver value to investors and are more likely to blunt trends. Regulators should encourage them. One argument for regulating some of them persuades Mallaby: they are growing. His case is for entrepreneurial boutiques. Once a fund stops being small enough to fail, or goes public and gives up the private-partnership structure that has controlled risk so well, regulation is warranted. LTCM is the warning: no public money paid for its burial, but the Fed was worried enough to convene the undertakers.
There is no simple size threshold, because systemic impact depends on timing: LTCM failed amid post-Russia panic, while Amaranth collapsed when Wall Street was awash in easy money. But three clues predict systemic danger: size of capital, extent of leverage and the liquidity of the markets traded.
| Case | Capital | Leverage and book | Markets traded | Systemic? |
|---|---|---|---|---|
| LTCM 1998 | just under $5bn (half Amaranth's) | 25x: $120bn positions plus $1.2tn derivatives exposure | esoteric, illiquid | Yes: Fed's Peter Fisher organized the burial |
| Amaranth 2006 | $9bn | normal leverage | gas, nearly all on exchanges | No: more headlines than shocks |
| Sowood 2007 | $3bn | normal | fairly liquid corporate bonds | No |
| Quant funds 2007 | $100bn+ combined | about 8x: $800bn+ superportfolio | superliquid equities; cut leverage fast | No |
| Citadel 2008 | $13bn | 11x: $145bn excluding derivatives | big OTC book, backed by medium and long-term borrowing | No |
The quant quake and Citadel are the most revealing cases: both failed the first two tests yet proved harmless because they passed the third. Citadel managed liquidity as LTCM had tried to, but better: by backing positions that could not be sold instantly with loans that could not be pulled instantly, it avoided a death spiral of forced selling. The lesson: portfolios above some threshold deserve scrutiny, but if liquidity is well managed, regulators should let the firm be.
- Tier 1: above about $120 billion of total leveraged assets. Regulators cross-examine the fund's derivatives positions and liquidity management. The bar is LTCM's 1998 portfolio, cautious given how much markets have grown since.
- Tier 2: above about $200 billion of total assets. Oversight extends to leverage, and if risk-weighted calculations show too thin a capital buffer, the fund must add extra capital. Still cautious: Bear Stearns held $350 billion of assets in 2006.
- Tier 3: any listed hedge fund. The presumption of competent risk management is softened, and examiners visit more often and more insistently.
The regime would deliberately leave nearly all funds outside the net. In January 2009 Institutional Investor listed only 39 hedge funds worldwide with capital above $10 billion. The other 9,000 or so, holding a bit over half the sector's capital, would need no registration or oversight unless suspected of insider trading or other violations, or unless unusual leverage pushed them over $120 billion. Free of compliance costs, they could grow and take over risk now held by behemoths, while the few dangerous funds would be treated as the investment banks they would in effect be.
The wheel turns
In 1949, when Alfred Winslow Jones set up his hedged fund, the merchant banks that became Goldman Sachs, Morgan Stanley and Lehman Brothers were neither global nor public. They were private partnerships risking their own capital flexibly, managing risk ferociously and largely unregulated, much like today's hedge funds. Over the next half century they listed their shares and spread worldwide, less because sprawling public firms manage risk better than because the rewards were irresistible: each flotation turned partners' illiquid stakes into instant liquid wealth, and each new market offered a fresh chance to risk shareholders' capital and collect a 50% quasi performance fee on wins. Those public-company incentives pushed them to take ever more risk until they paid the price; becoming bank holding companies in late 2008 was Goldman's and Morgan's admission that they could survive only with the Fed behind them.
Today's hedge funds are the Goldmans and Morgans of half a century ago: equally ferocious about risk, equally lightly regulated. The same temptation will lead some to go public, and a handful already have. When that happens, Mallaby warns, hedge funds will become the threat they were wrongly accused of being. The wheel of Wall Street turns; greed and risk never leave.
Mallaby errs on Galleon's prosecutor: the criminal case was brought by the US Attorney for the Southern District of New York, not the Manhattan district attorney. More substantively, "small enough to fail" judges funds one at a time; the quant quake and the March 2020 dash for cash, when crowded, leveraged Treasury basis trades unwound and the Fed bought Treasuries in bulk, show hedge funds acting together as a systemic force that his per-fund thresholds would miss. The Ibbotson study's 3 points of alpha rest on self-reported databases and a period that ends before a decade of weaker returns. "Don't regulate" also sits uneasily with the insider-trading record he recounts, and his thresholds ignore unregistered family offices.
Regulation. Dodd-Frank (July 2010) did the opposite of "Don't regulate": from 2012 advisers with over $150 million in private funds had to register with the SEC and file confidential Form PF. The Volcker Rule (final December 2013) barred most bank proprietary trading and capped bank stakes in hedge funds at 3%, pushing risk toward the funds as Mallaby wanted. No hedge fund was ever designated systemically important, and in June 2024 a federal appeals court vacated the SEC's new private-fund adviser rules. Crime. Rajaratnam was convicted in May 2011 and got 11 years; SAC Capital pleaded guilty in 2013 and paid $1.8 billion. Archegos. In March 2021 Bill Hwang's family office, exempt from registration, collapsed from a peak of $36 billion, leaving banks about $10 billion of losses (Credit Suisse $5.5 billion, before its 2023 forced sale to UBS); Hwang got 18 years in November 2024. Size. HFR put industry capital at a record $5.6 trillion in Q2 2026, about three times its size when the book appeared.
Do the Tiger Funds Generate Alpha?
Mallaby tests whether Julian Robertson's record was skill or luck by asking whether his protégés kept beating the market after leaving him. A Tiger Index built for him by the Hennessee Group gained 11.9% a year from 2000 to 2008 while the S&P 500 lost 5.3%, with a Sharpe ratio of 1.42: strong evidence that stock picking, the discipline where alpha is said not to exist, can be taught.
Robertson's record and the luck test
From its launch in May 1980 to its peak in August 1998, Tiger earned 31.7% a year after fees against 12.7% for the S&P 500; counting the 1999-2000 collapse, about 26%. It was up in 17 of the 21 calendar years Robertson ran it, all the more striking because its staple was stock selection, where academic studies find consistent outperformance does not exist.
Pure luck is possible but unlikely: if 1,000 people each flip 21 coins, about four will get heads 17 or more times, leaving 996 in 1,000 odds of skill. Borrowing Warren Buffett's argument (chapter 5), Mallaby adds a sharper test: if the managers Robertson trained also excelled, they learned something from him, and his own record can be put down to skill with near certainty.
Building the Tiger Index
The Hennessee Group, which has collected hedge-fund data since 1987, held monthly results for 18 of the 36 "Tiger cub" funds run by people who worked for Robertson before 2000 (distinct from the post-2000 "Tiger seeds" he financed). Senior vice president Samuel Norvell built the index. Two data biases are limited:
- Survivorship bias. The sample includes two funds that blew up, so it does not count only winners.
- Smoothing bias. The cubs trade liquid equities, not loans or off-exchange derivatives, so every wiggle is marked promptly rather than smoothed by infrequent valuation.
All hedge-fund databases still rely on voluntary self-reporting, which Mallaby concedes.
| Hennessee index, 2000-2008 | Annual return | Std. dev. | Sharpe | Tiger alpha vs it | Correlation with Tiger | Months up | $1,000 became | Worst run of losses |
|---|---|---|---|---|---|---|---|---|
| Tiger Index | 11.89% | 7.42% | 1.42 | n/a | n/a | 79 of 108 (73%) | $2,749 | -19.85% (Dec 07-Dec 08) |
| Hedge Fund Index | 4.78% | 6.66% | 0.59 | 8.95 pts | 0.54 | 70 (65%) | $1,523 | -20.87% (Oct 07-Nov 08) |
| Long/Short Equity | 4.44% | 7.76% | 0.47 | 9.95 pts | 0.46 | 67 (62%) | $1,478 | -19.74% (Oct 07-Nov 08) |
| International | 7.63% | 7.83% | 0.85 | 6.55 pts | 0.72 | 75 (69%) | $1,938 | -24.48% (Oct 07-Nov 08) |
| S&P 500 | -5.26% | 15.20% | -0.34 | 12.79 pts | 0.26 | 59 (55%) | $615 | -46.28% (Aug 00-Sep 02) |
- 2000: +33.0%. 2001: +21.3%. 2002: +6.0%. Strong gains while the S&P 500 fell for three straight years.
- 2003: +12.5%. 2004: +7.2%. 2005: +16.8%. 2006: +10.9%.
- 2007: +28.5%. The best year after 2000.
- 2008: -19.9%. Nearly a fifth lost, which still left the nine-year average near 12%.
Three objections, three answers
- Extra risk? Higher returns could just reflect more risk. Dividing return by volatility, as William Sharpe taught, gives the cubs a Sharpe ratio of 1.42, far above every benchmark (0.59 for hedge funds generally).
- Hidden insurance selling? A fund can flatter its Sharpe ratio by acting like an undercapitalized insurer, writing options against extreme swings and collecting steady premiums until the swing arrives and bankrupts it. Mallaby calls this vanishingly unlikely here: option writing is not part of Tiger culture, and the data run through late 2008, when such quasi-insurers were wiped out (some were).
- Still luck? Treating the index as a Buffett coin-flipper: the cubs beat the S&P 500 in 62 of 108 months, and the chance of doing that by luck is 7.43%, so roughly twelve in thirteen that the skill is real.
Alpha is slippery and imperfect data give imperfect conclusions, but Mallaby judges the weight of evidence overwhelming: both Robertson and his protégés had real skill, in a branch of investing where alpha is often said to be impossible.
The data are self-reported, half the cubs are missing (and missing funds tend to be worse), and the window, a falling S&P 500, flatters any long/short book with a short side. The 7.43% luck test asks whether the cubs beat the index more often than chance, a weaker claim than skill after risk. Mallaby's arithmetic checks out (four in 1,000 for 17-plus heads in 21 flips; 12/13 for the cubs). The cubs' later record is mixed: Tiger Global's public hedge fund lost nearly 56% in 2022. Robertson died in August 2022, aged 90.
Master Takeaways
Mallaby's history is a sixty-year argument against two myths: that markets are too efficient to beat, and that hedge funds are the system's most dangerous players. Hedge funds beat markets by finding edges that fade, and they fail without costing taxpayers, which makes them a safer home for financial risk than the too-big-to-fail banks that blew up in 2008.
- Edges are real but perishable. Every winning method (hedging, block trading, trend following, reflexivity, code-breaking) eroded once rivals copied it or the fund grew too big.
- Small enough to fail is the case. About 5,000 hedge funds closed from 2000 to 2009 and none needed a bailout; banks needed an IMF-estimated $10 trillion of support.
- Official mistakes pay best. Soros's sterling trade and Paulson's subprime short (about $15 billion for his firm in 2007) both exploited policy blunders, not innocent markets.
- Ask payouts, not odds. Paulson's Treasury, like the Bank of England in 1992, judged Lehman's failure a fair gamble; traders saw a free bet on collapse and made it certain.
- Leverage makes overreaction mandatory. From the 1994 bond rout to LTCM and the quant quake, borrowed money turned small losses into forced selling.
- Liquidity shrinks as you grow. Tiger's US Airways stake, LTCM's illiquid books and Amaranth's 40% of gas open interest all assumed exits their own size had closed.
- Owners' money disciplines risk. Managers with their savings in the fund, high-water marks and fear of margin calls behave more prudently than bank traders paid with shareholders' capital.
- Funding structure decides survival. Citadel lost 55% in 2008 yet lived, because it borrowed long against illiquid assets; Bear and Lehman borrowed overnight and died.
- Risk models are brittle. LTCM's partners ran VaR and stress tests and still failed, because in a crisis the correlations are created by the traders themselves.
- Pay for measured performance. A. W. Jones's real edge was not market timing but a system that paid stock pickers on results and made them invest alongside clients.
- Liquidity provision is a service. Steinhardt profited by buying blocks institutions needed to dump; markets were thinner when block desks retreated after 1987.
- Trend following works because people adjust slowly. Commodities Corporation, Tudor Jones and their followers profited from gradual absorption of news, contradicting random-walk theory.
- Reflexivity: prices change fundamentals. Soros's theory holds that biased perceptions move prices, which then alter the realities they are supposed to reflect.
- Hedge funds are not superpredators. In Asia and Russia they were often cautious or wrong; local capital flight, weak banks and fixed pegs drove the crashes.
- Fighting a bubble can be ruinous. Robertson closed Tiger in 2000 after refusing tech stocks, weeks before he was proved right.
- Smooth returns can hide illiquidity. Event-driven and endowment-style portfolios understate risk in good times and concentrate it in a crisis.
- Computers find what intuition misses. Renaissance's Medallion averaged about 39% a year from 1989 to 2006 by trading nonintuitive short-term patterns.
- Factories delegate risk. Multistrategy alpha factories let bosses back stars like Brian Hunter whose bets they could not fully understand.
- Banks did worse for four reasons. Capital rules as a crutch, weak personal stakes, fee-business distraction and complacent culture led them to buy subprime that hedge funds shorted.
- Crowding is harmless until a panic. Quant funds shared factor bets that only bit when one fund had to deleverage in August 2007.
- Short sellers are messengers. Chanos and Einhorn exposed Bear and Lehman's weakness; the September 2008 ban hurt hedge funds without saving the banks it was meant to protect.
- Mutual funds charge infinitely for alpha. An active fund charging 90bp above an index fund and delivering no excess return has an infinite fee per unit of alpha.
- Hedge funds do deliver some alpha. Correcting for survivorship and backfill, Ibbotson's team found 3 points a year of alpha in a 7.7% net return, 1995-2009.
- Regulate by size, leverage and liquidity. Mallaby's tiers: scrutiny above $120 billion of leveraged assets, capital rules above $200 billion, extra oversight for listed funds.
- Today's hedge funds are yesterday's merchant banks. Goldman and Morgan were private, prudent partnerships until going public rewarded risk with shareholders' money.
- Bank-sponsored funds carry bank culture. Bear Stearns's internal funds, run by ex-salesman Ralph Cioffi at 35 to 1 leverage and rescued by the parent, failed for bank reasons, not hedge-fund ones.
- Hedge funds can be firefighters. Citadel bought Amaranth's energy book in about two days and Sowood's in one night; once no dump loomed, prices recovered and the panics ended.
- Flat is enough. Pellegrini showed house prices need not fall: if they merely stall, refinancing stops, defaults pass 7%, and a 1%-thick BBB tranche goes to zero.
- Go for the jugular when the odds are lopsided. Soros pushed Druckenmiller to bet the maximum on sterling; Paulson recalled the advice when the ABX let him buy $2 billion of protection in two days.
- Disclosure can make you prey. Amaranth reported its strategies monthly and rivals could read its positions from widening spreads, which undercuts the claim that hedge-fund secrecy was the problem.
- In a fire sale the best assets go first. Sowood was long safe senior bonds, but deleveraging funds dumped what was easiest to sell, so quality offered no protection.
- A hedge-fund bubble shows up as a talent shortage. By 2005 over 8,000 funds, Hedge Funds for Dummies and celebrity founders signalled that capital was outrunning skill, and delegated factories gave novices room to blow up.
Frameworks & mental models
| Name | What it says | Use it when |
|---|---|---|
| Hedged fund (Jones) | Long good stocks, short bad ones, set leverage separately; split returns into selection and market | judging whether a manager adds skill or just market exposure |
| Alpha versus beta | Pay index fees for market return; pay up only for skill-based excess return | comparing fees across mutual, hedge and private-equity funds |
| Reflexivity (Soros) | Perceptions move prices and prices change fundamentals, creating boom-bust sequences | a trend is feeding on itself through credit or confidence |
| Script for the market (Tudor Jones) | Map how others are positioned and who must sell; bet when the odds are skewed | a crowded trade or forced seller is visible |
| Odds versus payouts | Ask what each outcome pays, not only how likely it is | a policy choice offers traders a one-way bet |
| Leverage arithmetic | A 5% loss on 8:1 leverage pushes leverage to 12.7:1 and forces selling | sizing borrowed positions or stress-testing deleveraging |
| Liquidity matching (Citadel) | Fund assets with borrowing of matching maturity; overnight money only for overnight-sellable assets | designing funding for any leveraged book |
| Three systemic tests | Capital, leverage and the liquidity of markets traded decide whether a failure spreads | deciding which institutions need supervision |
| Vicious cycle of safety nets | Insurance lowers the cost of risk, risk grows, insurance must widen | evaluating bailouts and deposit guarantees |
| Contrarian structure | Shorting, absolute-return goals and performance fees make funds trend breakers | arguing about whether speculation destabilizes prices |
| Tranche arithmetic | Junior layers absorb the first losses; a thin tranche goes from whole to zero over a narrow band of defaults | valuing structured credit or any layered capital stack |
| Capital-structure arbitrage | Short the junior debt of weak, cyclical, indebted firms; hedge with their senior debt | a cycle is turning and debt is layered |
| Edge in the platform (multistrategy) | Technology, legal, treasury and risk systems let one firm run many strategies and act faster than banks | judging whether scale is an asset or a risk |
| Four reasons banks fared worse | Capital rules as crutch, weak personal stakes, fee-business distraction, complacent culture | comparing risk-taking across institution types |
Strategies and how they broke
| Strategy | Pioneer (Ch.) | Edge | What went wrong |
|---|---|---|---|
| Long/short equity | A. W. Jones (1), Robertson (5) | stock picking plus paid-for-performance research | imitators dropped the hedge in 1969; Tiger grew too big and shunned tech |
| Block trading | Steinhardt (2) | paid for liquidity by institutions, plus broker tips | consolidated tape ended the opaque edge; 1994 leverage sank him |
| Trend following | Commodities Corp., Tudor Jones (3, 6) | slow absorption of news; strict risk rules | crowding; the trend stops at overextension |
| Global macro | Soros, Druckenmiller (4, 7, 9) | betting against untenable official policies | size and politics; Russia 1998, the dot-com exit |
| Fixed-income arbitrage | LTCM, Askin (8, 10) | tiny mispricings magnified by leverage | crisis correlations, illiquid exits |
| Event-driven | Farallon (12) | mergers, distress, illiquid assets with steady returns | hidden illiquidity in 2008 |
| Quant / stat arb | Simons, Shaw, AQR (13, 15) | computer-found short-term patterns | crowded factor bets in August 2007 |
| Multistrategy | Citadel, Amaranth (14, 16) | platform scale, specialists, back office | delegated risk (Hunter); 2008 convertible losses |
| Credit shorts | Paulson, Bass, FrontPoint (15) | asymmetric bets against subprime tranches | none in 2007; later Paulson losses on gold and banks |
Who's who
| Name | Role | Why they matter |
|---|---|---|
| Alfred Winslow Jones | founder, A. W. Jones & Co. (1949) | invented the hedge, leverage and 20% fee template |
| Michael Steinhardt | Steinhardt, Fine, Berkowitz | block trading and monetary calls; drowned in 1994 |
| Paul Samuelson | economist, Commodities Corp. backer | random-walk champion who bet on beatable markets |
| Bruce Kovner | Commodities Corp., then Caxton | trend follower turned macro trader |
| George Soros | Quantum Fund | reflexivity; Plaza 1985, sterling 1992, Asia 1997 |
| Stan Druckenmiller | Soros's chief strategist, Duquesne | ran the sterling trade; rode then quit the dot-com bubble |
| Julian Robertson | Tiger Management | stock picker whose cubs proved his skill; closed 2000 |
| Paul Tudor Jones | Tudor Investment | trader of positioning; 1987 and Lehman calls |
| David Askin | Askin Capital | mortgage-bond fund destroyed in 1994 |
| John Meriwether | Long-Term Capital Management | convergence arbitrage that needed a Fed-brokered rescue |
| David Swensen | Yale endowment | built the endowment model that funded the industry |
| Tom Steyer | Farallon | event-driven investing with illiquid assets |
| James Simons | Renaissance Technologies | code-breaking quant; Medallion's record |
| David Shaw | D. E. Shaw | computer-science approach to statistical arbitrage |
| Ken Griffin | Citadel | multistrategy factory; rescued Amaranth and Sowood; survived 2008 |
| Brian Hunter | Amaranth | gas trader who lost $6bn in 2006 |
| John Paulson | Paulson & Co. | subprime short, the "new Soros" of 2007 |
| Cliff Asness | AQR | quant whose fund was hit in the August 2007 quake |
| Nick Maounis | Amaranth founder | multistrategy boss who bet the firm on one gas trader |
| Paolo Pellegrini | Paulson's co-manager | mortgage research showing flat prices would wipe out BBB bonds |
| Andrew Lo, Richard Bookstaber | MIT economist; risk writer | warned that crowded, leveraged funds spread distress across markets |
| James Chanos, David Einhorn | short sellers | exposed Enron, Bear and Lehman |
| Alan Greenspan, Hank Paulson | Fed chairman; Treasury secretary | policy makers whose choices hedge funds traded against |
Mallaby had deep access to his subjects (about 150 recorded interviews) and it shows in his generosity: collusion, insider trading and market manipulation (Steinhardt, Galleon, Amaranth) are recorded, then outweighed. His "small enough to fail" test judges funds one at a time and underplays collective risk: crowded quant trades in 2007, the 2020 Treasury basis-trade unwind, and Archegos (2021), which cost banks about $10 billion. Some shorts helped create the products they bet against (Abacus, Magnetar), the alpha evidence rests on self-reported data, and his advice not to regulate was overtaken by Dodd-Frank and the Volcker Rule within months of publication.
Connect it
- Banking: prudential regulation. Mallaby's three tests (capital, leverage, market liquidity) map onto Basel III's capital, leverage ratio and liquidity rules (LCR, NSFR); Citadel's maturity-matched funding is the NSFR idea run by a hedge fund.
- Banking: counterparty risk. Askin, LTCM, Amaranth, Archegos and the 2008 margin fights show prime brokerage as the channel by which hedge-fund losses return to banks; daily valuation of derivatives, as Citadel did, is the defence.
- Risk management. LTCM's brittle VaR and Amaranth's "maximum loss" argue for stress tests that assume your own position sets the exit price, and for position limits as a share of open interest.
- Incentives. High-water marks, managers' own capital and fixed fee formulas are the design features that banks later imitated through bonus deferral, malus and clawback.
About this book
More Money Than God: Hedge Funds and the Making of a New Elite by Sebastian Mallaby, 2010. This is a chapter-by-chapter study of the book, with summaries, maps, tables and diagrams; short quotations keep their page references. It is shared free for personal study, is not for sale, and is not endorsed by the author or publisher. Please read and buy the original.
Insight Editions · rajeshrmahapatra.com/insight-editions