The Trouble with Quants
08/23/2007 12:00 am EST
Nicholas Vardy, editor of the Global Guru, explains why Wall Street's sophisticated quantitative-trading models went haywire-yet again.
Quantitative trading has been a terrific-and profitable-boon to Wall Street. But recent market turmoil has revealed some startling weaknesses in this real world video game style of trading.
Quantitative trading uses complex computer models to make trading decisions based on historical models that extrapolate past patterns into the future. Sophisticated computer algorithms spot opportunities to buy and sell securities-exploiting very small price differences. These differences are then leveraged up, leading to often eye-popping returns.
Recently, none other than Goldman Sachs, one of the savviest players in finance, revealed huge losses in its quant-driven Global Equity Opportunities funds. The collapse was startling in its quickness. [Within a week], it had lost more than 30% of its value.
Goldman committed $2 billion of its own capital to bail out the fund. The failure of such a large fund would have been a huge embarrassment to Goldman. And among the losers would have been Goldman partners, a significant number of whom are thought to have been investors in the fund.
Highly respected Renaissance Technologies suffered similar woes. Founder James Simons wrote investors that "we cannot predict the duration of the current environment," even as he revealed startling losses in its funds, including Medallion, which has had an annual return of 30% since 1988. As Lehman Brothers noted: "Models (ours including) are behaving in the opposite way we would predict and have seen and tested for over very long time periods,"
Quant models work very well-until they don't. At that point, they collapse, with devastating consequences. Here's why.
First, for all their bells and whistles, quant models are inherently flawed. It turns out that "rare" events aren't so "rare" after all. The collapse of the Goldman fund is just a reprise of the collapse of Long Term Capital Management (LTCM) in 1998. Recall that LTCM's whiz kids, who included some Nobel prize-winning economists, had devised model-based trading strategies that also went south.
Second, market players assume a level of liquidity that just isn't there in times of panic. When someone yells "fire" in the global financial markets, too many players head for the exit doors at once. And the low-interest-rate, low-return environment of the past few years prodded many investors to go into less liquid assets to generate higher returns.
Third, good times breed greed or-in financial terms-an appetite for leverage. Goldman's fund was leveraged six times before Monday's bailout. As other highly leveraged investors cut debt, they unwind positions, forcing them to sell stocks at fire sale prices.
The elephant in the room is investor psychology-that elusive variable that can never make it into even the most sophisticated of quant models. Yet plain old common sense tells us that when investors get frightened, they sell indiscriminately. The only real surprise is that this surprises the quants.