Around lunchtime on Sunday, Jeff Larson placed a call to Ken Griffin.22 Larson recalled how Citadel had bought Amaranth’s trading book the previous year. He asked whether Griffin might want to do the same for Sowood’s portfolio. The sooner Sowood could find a buyer, the sooner it could stop predators from targeting its positions. Larson needed a deal before the markets opened the next morning.
Griffin got on the phone to his lieutenants. Gerald Beeson, one of the two executives who had parachuted into Amaranth, had just started his own vacation on a beach near Chicago. He drove back at top speed, dialing several colleagues on the way; he stopped off at home to throw on some long pants and raced to the airport. By around seven o’clock that evening, half a dozen Citadel officials had taken over a conference room at Sowood’s offices in Boston, where they discovered that Sowood had also summoned Morgan Stanley. As the two teams examined Sowood’s portfolio, it quickly became clear that many of the positions were difficult to value; they consisted of derivatives that were traded “over the counter” between companies, rather than on a transparent, centralized exchange, so only a firm that traded all these instruments itself had a hope of figuring out the going rate for them. Sowood’s tangle of legal arrangements with brokers and trading partners had to be assessed. The data that described its trades had to be uploaded into the buyer’s systems.
At around 9:00 P.M., the head of the Morgan team called Griffin in Chicago.
“Ken, we’ll pick this up in the morning.”
“We’ll get this done by then,” Griffin answered. He heard a noise on the line. He wasn’t sure if the other guy was laughing at him.
By 7:00 A.M. on Monday, Griffin had done what he had promised. He and his team had worked through the night and bought Sowood’s entire trading book. Jeff Larson explained to his investors that “Citadel offered the only immediate and comprehensive solution.” Sowood’s two funds were down 57 percent and 53 percent for the month; Harvard’s endowment had taken a $350 million hit; but at least the nightmare had now ended.23 The deal was announced publicly, calming the fear that Sowood might dump its positions. The bond market recovered more than 4 percent that day, and the panic was over. One hedge fund had imploded, threatening to start a systemic fire. Another hedge fund had swooped in, acting as the fireman.
Almost immediately, a new fire started.
THE NEXT FRIDAY, AUGUST 3, A RATINGS AGENCY ANNOUNCED that Bear Stearns’s debt might be downgraded. It was the first time a Wall Street firm’s financial health had been questioned in the crisis, and Bear Stearns’s stock fell so hard that its bosses convened a conference call in an attempt to calm investors. More than two thousand people dialed in, but there was no calming effect at all. Bear’s chief financial officer blurted out that the credit markets were behaving in the most extreme manner he had witnessed in his long career; “he fucking blew the market up,” Bear’s treasurer said sweetly.24 On CNBC a few hours afterward, the financial pundit Jim Cramer fanned the flames. “It is time to get on the Bear Stearns call!” he ranted, excoriating the Fed for sitting on its hands. “We have Armageddon.”25
Cramer could not guess where the next fire would come from. But away on the sidelines of the subprime drama, quantitative hedge funds were starting to sense trouble. In the second half of July, computerized systems that traded equities were no longer performing well, and some were even losing money. As the quants analyzed the problem, they discovered something disturbing. It was not that a new risk was swamping the buy and sell signals that had been profitable for years; the signals themselves were no longer working. The quants had programs that bet on momentum in stock prices; programs that bet that momentum would reverse; and programs that bet that cheap stocks with low price-to-earnings ratios would outperform expensive ones.26 All these bets were fizzling at once. Somewhere out there in the trading universe, one or maybe several quants were liquidating their holdings, perhaps because they had lost money on their mortgage bets and needed to raise cash. Their forced selling was driving prices against anyone who had a comparable portfolio.27
At the end of July, Mike Mendelson, a hard-charging ex-Goldman quant, decided it was time to cut the risk in his trading book. Mendelson now worked for AQR, an investment company set up in 1998 that managed $10 billion of capital in hedge funds and another $28 billion in traditional ones. AQR’s chief founder, Cliff Asness, had contributed to the academic literature on pricing anomalies in stocks; having programmed his computers to milk these effects, he delivered steady, uncorrelated returns while also sleeping soundly. Although AQR’s losses in late July had been too modest to disturb anybody’s rest, Mendelson had heard that another big quantitative fund had suffered a bad loss. Erring on the side of prudence, he trimmed leverage. Then, in the first days of August, AQR’s models started to work again. Whoever had been liquidating quant positions must now have stopped. The trouble seemed to be over.
On Monday, August 6, Mendelson sat through some routine meetings at AQR’s office, a utilitarian suite in a featureless building just by Greenwich station. Around midmorning, he strolled out to the local Subway sandwich store, and as he waited in line he checked his funds’ performance on his BlackBerry. He peered for a few seconds at the screen. The numbers were all red, and they were not small, either. In the past three hours, AQR had lost tens of millions of dollars.28
“Oh, God, this is ridiculous,” Mendelson thought to himself. Some quant somewhere must be deleveraging, but on a monstrous scale. Or maybe several quants were bailing all at once? How long could this go on?
The one thing Mendelson knew was that he would have to cut leverage quickly. If a fund has $100 of capital to support $800 of positions, a 5 percent loss will leave it with $60 in capital and $760 worth of positions: Its leverage rockets up from eight to one to more than twelve to one.29 If there is another 5 percent loss the next day, the leverage will not merely rise another 50 percent; it will practically triple, from twelve to one to thirty-three to one. A third 5 percent setback will drive leverage to infinity and beyond, since the fund’s capital will be negative. Back in AQR’s offices, Mendelson and his colleagues sketched a hyperbolic curve on a notepad, showing how leverage could accelerate upward. The only way to survive was to keep leverage on the flattish, left-hand portion of the curve. If AQR’s funds were down 5 percent, they might have to sell almost two fifths of their positions to keep leverage stable. Otherwise they would begin a death spiral.
Meanwhile, versions of this drama were playing out at other quantitative hedge funds. Most were not like Jim Simons’s Medallion: They were trading well-known price anomalies, not esoteric secrets; “there is no E=MC2 under the hood,” as Asness put it.30 Even Simons confessed that his large fund for institutional investors traded on signals that were understood by others; everyone had read the same academic papers, had looked at the same data, and was making the same types of bets, especially on stocks with momentum and value.31 In normal times, this didn’t matter: Even if an army of funds was chasing “value,” there were dozens of ways to measure this phenomenon, so crowding was limited. But as with all investment strategies, crowding did turn out to matter during a panic: Selling by one big fund caused losses at other ones, especially since quantitative strategies had grown large enough to shove prices around.32 Once rival funds started to incur losses, the logic of the hyperbolic curve would force them to sell too. By Monday afternoon, Mendelson began to see how bad things could get. Not only would AQR have to sell a huge chunk of its positions to keep its leverage stable in the face of initial losses. As its competitors sold too, it would have to sprint to stay still, racing other quants to cut the size of its portfolio.
The next day was even more brutal than Mendelson had expected. AQR’s models lost money twice as quickly as on Monday, and the firm instructed its computers to dump billions of dollars’ worth of positions. The good news was that technology made it possible to liquidate a portfolio much faster than in earlier years; the bad news was that AQR’s rivals could liquida
te just as quickly. Every quant was firing off torpedoes at every other quant; there were rumors of funds that were down 10 percent or worse. At the Renaissance campus in Long Island, Jim Simons huddled with his top lieutenants in front of the computer screens, tweaking the parameters on his models like a pilot navigating a hurricane. Cliff Asness, AQR’s founder, blew up in the office and smashed computer screens.33 He took a call from Ken Griffin, who by now had a reputation for buying the corpses out of car wrecks. “I looked up and saw the Valkyries coming and heard the grim reaper’s scythe knocking on my door. I did my best to run to the light,” Asness said later.34
While this chaos unfolded, policy makers appeared to occupy a parallel universe. On Tuesday, August 7, the second day of the quant quake, the Fed’s interest-rate committee issued a warning about the risks of inflation. The next day President Bush visited the Treasury to meet with his economic advisers. “[I]f the market functions normally, it will lead to a soft landing,” he said hopefully. On Thursday the tone from Washington began to change, but less because of the carnage at quantitative funds than because of trouble from Europe: The giant French bank BNP Paribas had suspended redemptions from three internal money-market funds, citing “the complete evaporation of liquidity.” Subprime losses were clearly scaring the markets, and the European Central Bank responded with $131 billion in emergency liquidity. By Thursday afternoon, the Fed’s chairman, Ben Bernanke, had turned his office into a makeshift war room, and his chief lieutenants dialed in from various vacation locations. Early the next morning, the Fed reversed its earlier emphasis on inflation, pledging to provide enough cash “to facilitate the orderly functioning of financial markets.”
Meanwhile in Greenwich, Mendelson was starting to see light at the end of the tunnel. It had nothing to do with the Fed’s U-turn and everything to do with other hedge funds. Starting on Tuesday, Mendelson had begun to call brokers and friends in the markets—anybody who might know anything about how other leveraged quant funds were positioned. The business was dominated by a handful of firms. There was Jim Simons’s new fund, which ran more than $25 billion of institutional money. There was Highbridge Capital, a subsidiary of J.P. Morgan. There were D. E. Shaw, Barclays Global Investors, and Goldman Sachs Asset Management. Mendelson wanted to know how much the big players had cut leverage so far: If they got themselves down into the flat part of the hyperbolic curve, the selling pressure would end and the storm would be over. After working the phones all day Tuesday, Mendelson reached a buddy just after midnight. The guy had rushed back early from vacation and was falling apart: He was exhausted, blabbering, at the end of his tether. Mendelson could tell he was about to liquidate his whole book. He ticked one firm off his list and hit the phones again the next morning.
By Thursday evening, Mendelson had figured out that only one big player had yet to cut leverage. He guessed it might be one of the hedge-fund subsidiaries of his old firm, Goldman Sachs: the $5 billion Global Equity Opportunities Fund. Goldman’s executives had decided that the fund’s positions were too big to sell, so its leverage had rocketed up as its bets got hammered. When the market opened on Friday, one of two things would happen. Either the fund would liquidate, hitting other quant funds for a fifth straight day. Or it would be recapitalized by its parent company.35
When Friday morning came, Mendelson could not care less about the Fed’s stance on inflation. He was looking at his own trading model. Within a few minutes, it was generating profits—its performance practically screamed out that the Goldman subsidiary had been rescued and that the quant liquidations had ended. Acting on that signal, AQR began to releverage as quickly as possible; the more it could catch the upswing as money flooded back, the more it could make up for the past four days of carnage. The following Monday, Goldman Sachs announced publicly what Mendelson had already guessed. It had recapitalized the Global Equity fund with $3 billion in fresh money.36
The quant quake of August 2007 ended as abruptly as it had started. Friday and Monday were great days for the models, and most of the quants recouped at least part of their losses. But the drama prompted a new round of debate about hedge funds, and the agonizing outlasted the disruption in the markets. Granted, Amaranth and Sowood had been rescued by a fellow hedge fund, making it hard to argue that the sector was destabilizing. Granted, hedge funds—or at least, nearly all freestanding hedge funds—had dodged the mortgage bullet, suggesting that they were better money managers than their banking rivals. But the storm in the equity market was surely a warning. The most sophisticated hedge funds had lost control of their models. The rocket scientists had blown up their rockets.
The most persuasive critics came from within the hedge-fund establishment. Andrew Lo, an MIT professor who ran his own hedge fund, published a widely cited postmortem on the quant quake; and Richard Bookstaber, an MIT alumnus who had worked at several major funds, pressed the warnings he had recently published in a pessimistic book on finance.37 Lo and Bookstaber contended that the rise of leveraged hedge funds created a new threat: Trouble in the mortgage or credit markets could saddle a multistrategy fund with losses, forcing it to liquidate equity holdings; distress could leap from one sector to the next; the financial system as a whole was riskier. Lo and Bookstaber linked this warning to a gloomy view of hedge funds’ investment performance. There were too many quant funds chasing small market anomalies. This crowding diminished investment returns, which in turn drove hedge funds to use dangerous amounts of leverage to maintain profits.
There was some truth in all this pessimism. During the LTCM crisis, credit markets had been in turmoil but quantitative equity funds had been fine; the rise of leveraged traders helped to explain why the fire had jumped the fire wall this time.38 But it was one thing to say that crowding was a problem in moments of turmoil, quite another to assert that it was forcing down returns in the good times and making dangerous leverage inevitable. Lo’s paper presented the returns from one quantitative strategy—buy stocks that are doing badly and sell ones that are doing well—and made much of the fact that profits from this simple contrarian model had deteriorated since 1995, apparently substantiating the case that hedge funds had no choice but to employ more scary leverage. But basic contrarian strategies were not central to the quant quake, because the quants themselves could see that there were limited profits in them. The big money in quant funds was in other strategies—for example, sell expensive growth stocks with high price-to-earnings ratios and buy cheap, dowdy ones with low ratios. The price gap between growth stocks and value stocks had shown little sign of narrowing in the years leading up to 2007; as Asness put it, “We are fond of saying that if these strategies are truly horribly overcrowded, then someone has apparently forgotten to tell the prices.”39 Other quants made versions of the same point. Marek Fludzinski, the head of a hedge-fund company called Thales, tested virtually identical trading strategies to see if the presence of one of them would erode the other one’s returns; remarkably, it did not, suggesting that crowding was not actually a problem.40 It was true that uncrowded bets could suddenly feel intensely crowded in a moment of turmoil; but this held for virtually all trading strategies, not just quantitative ones. “Of course, ‘good investing gets clocked as some investors rush for exits’ is not as catchy a story as ‘quant brainiacs follow their computers to a well-deserved doom,’ so I’m probably not going to win this battle in the media,” Asness lamented.41
If quant strategies were not as crowded as the critics suggested, and if scary amounts of leverage were therefore not inevitable, what of the charge that hedge funds increased the risk of broad systemic blowups? Here too the Lo-Bookstaber critique needed to be qualified. During the first three days of the quant quake, the signals in the traders’ models performed abysmally, but the broader market remained calm—the average American household with its nest egg in an index fund would have noticed nothing, undermining the notion that this was a crisis for the whole financial system. On the fourth day of the quant quake, the market did su
ffer a hard fall, but this reflected the crisis in the credit markets more than the tremors in quant land. The great thing about liquidating so-called market-neutral strategies was that the effect on the overall market was neutral: For every stock the quants sold, they covered a short position. It was true that if Goldman Sachs had not rescued its subsidiary, the deleveraging would have lasted longer, potentially forcing multistrategy funds to dump positions in other markets and spreading the trouble. But the way things actually turned out, the market punished overleveraged traders while the broader system suffered little harm. This was how capitalism was meant to discipline its children. No regulator could have done better.
In the final analysis, it was hard to disagree with Mike Mendelson’s verdict on the quant quake: “A bunch of us lost a bunch of money and had a really tiring week, which sucked, believe me. But I don’t think it was a big public policy issue.”42
More Money Than God_Hedge Funds and the Making of a New Elite Page 40
