Though it may seem unusual, not all market crashes are obvious at first sight. True, when investors bet a particular investment will appreciate and come to find out they are wrong, the result is usually plainly visible in the price of that investment. However, not all investment strategies are bets that a particular security, or a market as a whole, will move in a certain direction.
Especially since the rise of quantitative investment strategies, some investors bet on the relationship between assets, or more specifically the outperformance or underperformance of certain characteristics of investments, rather than in specific securities. When these strategies fail, the results may be invisible to those simply looking at prices in the market as a whole but they will be painfully apparent to those well versed and invested in such strategies. This is exactly what happened during a market event known as the ‘Quant Quake’.
Factors
Quantitative investing can be grouped into a set of distinct strategies. One of those that experienced the most pain during the August 2007 ‘Quant Quake’ was statistical arbitrage, or ‘stat arb’. This ‘quant’ strategy can be defined as trading based on very short-term movements in technical indicators of a security’s performance, such as price and trading volumes, often with an eye to profit from short-term reversion to the mean. Other strategies, distinct from stat arb, may be based on similar principles but look for longer-term opportunities.
In either case, quantitative investing often attempts to be ‘market neutral’ or contain little correlation with the direction of broader markets. For example, a market neutral equity fund may bet on certain stocks and against others so to profit from any divergence or convergence but be unaffected by movement affecting both stocks equally. In other words, whether the stock market as a whole goes up or down should be irrelevant under this investment approach.
Further, quant strategies generally seek to profit from what are relatively small movements in security prices. As a result, they tend to do better in more volatile markets. That said, markets do not tend to be very volatile most of the time and in these less volatile markets, quant funds generally use debt, or leverage, to maximize returns. In essence, they borrow money to amplify the return of each dollar invested. Borrowing is intrinsic to many quant strategies because they involve simultaneously establishing ‘long’ and ‘short’ positions in different securities and ‘shorting’ a security is a form of borrowing. In addition, their long positions themselves may be financed by a lender.
To understand quantitative strategies, it is important to note that they usually segment market performance into ‘factors’ rather than individual securities or types of securities. Factors include the likes of growth, value, large-cap, small-cap, momentum, liquidity, and volatility, among others. So, instead of attributing a market rise to a handful of ‘hot’ technology stocks, say Cisco or Microsoft in an age past, a quant might attribute it to the ‘growth’ or ‘momentum’ factors.
Rather than invest in the stock of a particular company to profit from this trend, a quant fund might decide to go ‘long growth’ by simultaneously betting on growth companies and against value companies, slower growing firms generally trading at cheaper valuations. Doing so allows the fund to make a bet on growth stocks specifically rather than on the movement of the overall market. In other words, this strategy can succeed even if growth stocks fall, so long as they fall by less than value stocks, making the fund manager indifferent as to whether stocks rise or fall overall, at least in theory.
Alpha
What quant funds look for is termed ‘alpha’, a strategy’s ability to deliver returns without incremental market risk. In searching for alpha, funds would often back-test a particular strategy using historical return data. Once an alpha-generating strategy became public knowledge, it could no longer be relied upon to generate alpha because whatever mispricing that allowed it to arise would disappear as other investors sought to profit from it.
It would seem that quant funds would do anything to protect their secret formulas. However, their strategies were rarely proprietary in reality, at least not in the sense of being truly unique to a particular firm. Quantitative investors hired employees with similar backgrounds, employed similar models, and were always chasing the most reliable sources of alpha. This is perhaps the most important fact in setting the context for the Quant Quake.
While on this topic, the typical staff of a quant fund is quite different from that of other investment managers. Quant firms employed and were often run by mathematicians and scientists; a typical fund counted Ph.D.s among its researchers and managers. One large quantitative hedge fund, Renaissance Technologies, was founded by a Jim Simons, a mathematician previously employed by the U.S. National Security Agency as a codebreaker. For this kind of talent, there was a lot of competition on the part of firms like Renaissance, D.E. Shaw, AQR, Highbridge Capital Management, and others running quant investment strategies, whether stat arb or other approaches.
Inflows
Quant strategies generally performed well in the 2000s. This was a period in which hedge funds were successful overall and received large inflows of investor money. In the case of quant funds, some new investors sent money their way despite finding quantitative investing as comprehensible as magic. In response to this frenzy, some firms started new quant funds pursuing strategies that were entirely new to them. It’s important to recall that this was during a period when markets were not particularly volatile. Therefore, these funds, old and new alike, used larger amounts of leverage to amplify their returns.
Quant Quake
After some years of this trend, quantitative investing faced its biggest challenge to date. The Quant Quake struck on the week of August 6, 2007. The consensus is that it began when a firm or some group of firms began to liquidate their equity quant strategy portfolios. This meant buying the securities they were previously short and selling those they were previously long in an effort to unwind positions. The exact reasons, or even the identity of the firm involved in the initial liquidation was never discovered.
Nonetheless, events accelerated on August 8th and 9th. As some firms unwound positions, the act of covering short positions and selling long positions caused prices in the underlying securities to move against other quant investors pursuing similar strategies. This is because selling a long position causes that security’s price to fall while unwinding a short position causes the security’s price to rise. Thus, when one investor unwinds a position, it actually makes losses worse for other investors with the same exposure.
Not surprisingly, the price movements caused funds with similar exposures to unwind positions to reduce risk or raise cash. Funds would have needed to reduce leverage in order to avoid a margin call from their lender or in response to one already received. The breadth of the turmoil widened as strategies initially unaffected began to experience losses also, the product of continued liquidations by multi-strategy funds. Making matters worse, some firms had made ‘portfolio insurance’ deals with investors to liquidate and return funds if losses rose to some threshold. In short, their leverage and investor fear made quant funds forced sellers during the Quant Quake, aggravating the turmoil.
It didn’t help matters that liquidity disappeared just as it was most needed, as it often does. Measures of liquidity in equity markets fell over the course of the week. Market-makers and dealers, which typically add liquidity to markets, may have reduced their activities to reduce risk, buying fewer securities others wanted to sell and selling fewer that were widely sought.
Further, some quant funds also employed mean-reversion strategies that made them de-facto market-makers by frequently enter into contrarian trades (e.g. buying what others are selling to profit from reversion to the mean). However, these groups likely reduced or even ceased their activities as other parts of their firms suffered losses. The lack of liquidity exasperated the swings in particular stocks.
During the Quant Quake, the more overlap that existed between funds’ positions, the worse those trades held in common performed. It was the most popular trades that did worse during the Quant Quake and not for exogenous reasons but precisely because of their popularity. The volatile combination of overlapping positions, evaporating liquidity, and sudden deleveraging became the nightmare of the industry.
Nonetheless, the stock market was essentially flat on the week of Quant Quake. Indeed, the turmoil at quantitative hedge funds may not have even registered in the minds of discerning investors elsewhere. This market crash was purely cross-sectional, as old relationships between stocks melted down. As funds with both short and long positions liquidated, it caused some stocks to fall and others to rise but no massive overall change.
Precedent
Quantitative hedge funds had reason to worry about ‘crowded trades’, positions held by many firms simultaneously. Similar events transpired a year earlier when hedge fund Amaranth Advisors liquidated, affecting other quant funds but having little wider market impact. However, what was happening in 2007 was on a much larger scale. The Amaranth fiasco involved a liquidation of $5-10 billion in assets at most; Quant Quake may have seen $500 billion in assets liquidated in a week according to some. Further back, in August 1998, funds involved in fixed-income arbitrage strategies suffered losses as the industry attempted to deleverage in the wake of the failure of Long Term Capital Management. There had to be a bailout then to prevent wider damage.
Summer 2007
So, there was precedent to what was happening. Still, it isn’t completely clear what triggered the panic in 2007. This is partially because the identity of ‘patient zero’ remains unknown, unlike was the case with Amaranth and Long Term Capital. There is a widely held theory though. Many suspect that the trouble spilled over from fixed-income markets.
In July 2007, trouble in the mortgage markets was becoming widely apparent. The investment bank Bear Stearns had to shut down two mortgage-focused hedge funds. The speculation is that losses in these credit markets caused a particular firm, ‘patient zero’, to liquidate their quant fund holdings, either to reduce overall leverage and risk exposure or to raise cash. Investors often have to liquidate what they can, namely what is most liquid, rather than what is worth selling on its own merits.
In any case, liquidation of their quant positions would have caused this firm’s pain to spread to other managers pursuing similar strategies. It’s been supposed that the likeliest candidate for patient zero, if there was just one firm to blame, would be a company with assets deployed in both credit markets and quant strategies, perhaps a multi-strategy hedge fund or a proprietary-trading desk at a bank, where a bank’s own capital is invested in an array of markets. In any case, this would seem to suggest that quant strategies, or at least equity stat arb strategies, might not be as uncorrelated as previously thought. They can even be exposed to tail risk or black swan events present in totally different markets.
Scope
Quant Quake was severe; some ‘factors’ and the firms that invested in them lost 30% of their value in just a few days. However, the events were also short-lived. New capital coming into the market looking for bargains steadily reversed the meltdown, starting on August 10th. Nonetheless, some quant firms shut their doors after Quant Quake and those that survived did so in reduced form. Goldman Sachs’ Quantitative Investment Strategies unit, which managed $165 billion, was down 30% during the week and lost even more funds to client redemptions that followed.
Nonetheless, because overlapping positions were central to the meltdown, some distinct strategies were unaffected. Examples includes global macro quant strategies that are based on a top-down analysis of macroeconomic and other trends not endogenous to a particular security. Regardless, whether engaged in stat arb or other strategies, those that were able to withstand the pain and stick to their strategies actually recovered a significant portion of their losses within weeks. At least some recovered back to their old high water marks within months of Quant Quake. However, for a firm to survive, it needed to have had healthy cash levels, trusting investors, and stoic lenders, not exactly things that are obvious to find in a highly leveraged fund delivering mammoth losses in the span of just a few days.
Lesson
The Quant Quake may be one of the most recent large market crashes you’ve never heard of. It isn’t exactly a secret to anyone, but neither did it produce dramatic yet simple charts or easy to comprehend storylines, at least for those not acquainted with investments or financial markets. However, for those who lived through it with financial or reputational interests at stake, it was unforgettable. Regardless, it has wider lessons for investors, especially when it comes to the importance of understanding systemic risks like crowded trades and broad deleveraging.
More from the Tontine Coffee-House
Quantitative investing was enabled by advances in computing. Read about how other technologies changed finance, like the transatlantic telegraph cable and the stock ticker machine. Also learn the effects the internet had on the brokerage business.
Further Reading
1. Asness, Clifford. “The August of Our Discontent: Questions and Answers about the Crash and Subsequent Rebound of Quantitative Stock Selection Strategies.” Sept. 2007.
2. DeMurs, David. “Risk Fast and Slow.” R/Finance 2017. R/Finance Conference, 2017, Chicago.
3. Gorman, William Evan, “Crowding the Exits: Quantitative Equity Market Portfolio Endogenity, and 13F Disclosure as Catalyst for Systemic Liquidity Shock and Contagion” (2016). Law School Student Scholarship. 849.
4. Khandani, Amir E., and Andrew W. Lo. “What Happened to the Quants in August 2007? Evidence from Factors and Transactions Data.” Journal of Financial Markets, vol. 14, no. 1, 2011, pp. 1–46.
5. Tulchinsky, Igor. The Unrules: Success in the Age of Robots and Machine-Learning. John Wiley & Sons, 2018.
