CHAPTER 1
Quake

"Take aggressive risks, but manage losses."

On the morning of August 6, 2007, a Monday, I arrived early at WorldQuant's office in Old Greenwich, Connecticut. I had a lot on my mind: I was in the middle of moving, my head filled with the logistical details of movers, schedules, and the kids. By 10 a.m., however, I knew something was wrong. We had been hit, seemingly out of nowhere, by a wave of losses on our statistical-arbitrage trades - a strategy, common to a hedge fund firm like WorldQuant, that takes advantage of pricing differentials between related financial securities.

As the hours ticked by, anxiety quietly gripped the office. Because our trading is automated, the atmosphere at a quantitative investment management firm like WorldQuant resembles a library far more than it does a frantic trading floor. Nobody's screaming or rushing around. But that Monday you could feel the tension. There was little laughter, and the portfolio managers, clearly nervous, drifted in to discuss their exposures. The next day it got worse.

WorldQuant had been in existence for only six months, although I had been engaged in quantitative trading, which involves using sophisticated math and large amounts of data to identify trading opportunities, since 1995. At WorldQuant we had poured resources into developing about a hundred predictive algorithms we call alphas: mathematical expressions and computer source code that we rigorously back-test before putting them into production in live investment strategies. All that effort went into ensuring that we wouldn't take a hit like the one we were suffering. We knew that individual alphas regularly weaken or fail, and we were no strangers to drawdowns - we experienced significant declines roughly once a year back then. But our alphas were not supposed to fail collectively. This was bad.

You know what they say: When the CEO moves into a new house, it's a signal to sell. What we didn't know immediately was that similar losses were hitting our competitors at other quant firms. Renaissance Technologies, D.E. Shaw, AQR, and Highbridge Capital Management all saw their finely honed strategies take a sudden nosedive. Goldman Sachs, which at the time had one of the largest quant books - $165 billion - eventually lost more than 30%. Just like us, our rivals must have been struggling to figure out what had happened and why it seemed to be happening just to quant firms.

There had been some ominous signs in the surrounding financial world. For much of the summer, fallout from the unfolding subprime mortgage crisis had been sending shock waves through the markets. Bear Stearns was forced to close two mortgage-backed credit funds, and there were signs that European banks were growing wary of lending to one another. But our investment strategies were designed to be market neutral - that is, uncorrelated with the broader market. Those subprime issues, in theory, should not have affected the quantitative strategies we employed at WorldQuant. But then, nearly every quant shop probably thought the same way.

Quant firms are only a slice of the hedge fund world, which in turn is only part of the investing universe. Though firms like WorldQuant were hit hard on August 6, 2007, there were no signs of a broader collapse. The next day the Federal Reserve decided to leave interest rates unchanged. Stocks fell after the announcement, then recovered; that week the S&P 500 edged down only very slightly.

As we tried to figure out what had happened, all we really knew was that our relative-value and statistical-arbitrage alphas were not working, as if their plugs had been pulled. We suspected that someone out there had taken a hit and was liquidating, setting off a chain reaction of selling, but we lacked the time, the distance, and the data to comprehend fully what was going on. We watched nervously as the problem spread from the U.S. to Japan.

Over my trading career I'd learned a number of lessons that had served me well: Don't get emotional about your trades. React instantly to bad news. If it's scary, run. Take aggressive risks, but manage losses. Back in August 1998, when I was just building my trading portfolio, the Russian government suddenly devalued the ruble and defaulted on its debt. In the resulting violent drawdown, I saw my entire year's gains evaporate in a few days. A month after that, hedge fund firm Long-Term Capital Management needed a bailout by major banks to avoid causing damage to the American financial system. Now, almost nine years to the day later, that chaotic time was on my mind.

The problem of looking ahead, of course, is that you can't know how big or how long the declines will be. After the first losses on Monday, I made the decision to start liquidating the entire portfolio on Tuesday, giving up all the year-to-date profits. Some of this was my memory of the Russian default, when I held on too long, and some was intuition - observing the fear in people's eyes. Liquidating was difficult to swallow, but on Wednesday the carnage deepened, and we felt lucky to be out of it. On Thursday I came into the office early and made a decision to jump back in with 50% of our capital. I was aware that the market could sweep lower, but once again I was relying on intuition - not just on instinct, but on instinct shaped by experience.

In fact, the markets righted themselves as suddenly as they had declined. Just like that, most of the participants were making money again, though we took a few months to get back to 100% invested. We ended up having a pretty solid year. But those who hesitated to sell, had trouble liquidating, or sold into the recovery doubled their pain.

That August 2007 episode became known as "the quant quake," and it contained a number of lessons: There are risks that you've never thought about, and there are uncertainties. Sometimes you have to act quickly with too few data points. At WorldQuant we may practice quantitative trading, but we also know when to rely on intuition born of experience.

The firm went on to generate stable returns again, and as we accumulated the alphas that we use to build strategies, we experienced fewer significant drawdowns. In the industry the quant quake triggered a rethinking of investment models and a considerable amount of debate. Were too many quantitative hedge funds chasing the same strategies and eliminating the profits? What did happen in early August 2007?

To this day the evidence remains circumstantial and no one really knows for sure what set off the quake. But in the subsequent years, we've developed a better idea thanks to academic research. A month or so after the quake, two finance academics, MIT's Andrew Lo and Amir Khandani, tried to unravel what had happened by building quant portfolios and simulating the episode - in a sense, running the history backward. They concluded that somewhere in the markets a large player - Lo and Khandani thought it was a bank, but Bob Litterman, who ran Goldman's quant fund at the time, later argued it was a multistrategy hedge fund - may have taken a hit and quickly sold a large relative-value position to respond to credit-related margin calls or to take risk-reduction measures. Given what was going on at the time, there may have been a link to the growing subprime mortgage problem. Liquidating positions in turn put pressure on quant firms with similar positions heavily invested in equities, made worse by leverage, which magnifies gains in rising markets and losses in falling ones.

Then a contagion effect developed, with the stress in one part of the market spreading to others. Prices fell, and the more they fell, the worse it got. The fact that the quant quake seemed to target relative-value trades may have been a coincidence, but it did suggest that unrelated markets had inadvertently grown more correlated, creating a so-called crowded trade without realizing it, and raising the risk for everyone.

We would see far broader and more dangerous correlations emerge when the global financial crisis broke upon us all. When Lehman Brothers collapsed in September 2008, WorldQuant had another scare: Lehman was our prime broker in Asia and Europe, and its failure meant we couldn't trade our overseas portfolios for several days. But in this case, at least, we knew what the problem was. We quickly negotiated a new prime brokerage relationship and got back into the market in about a week.

As the world struggled to recover from the financial crisis, WorldQuant continued to perform and grow. Today we believe our greatest growth is still ahead of us. We have seen remarkable increases in people and data, computing power and market experience. In fact, it has become clear to me that we are part of an exponential revolution in quantitative finance.

What does that mean? I believe that nearly all aspects of WorldQuant's business, and perhaps our broader business lives, are undergoing not linear but exponential growth. As a result, goals that seem shocking today will look normal tomorrow and useless the day after. Exponential thinking requires audacity, not complacency. It means not believing in limits, which are temporary and meant to be broken. It calls for risk-taking as a way of life. In exponential thinking the terrain ahead is always unknown. In unknown terrain there are always bumps; it's a world of turbulence...