New Challenges in the Quantitative Space

I was having lunch with a friend of mine (from back in high school) who is now a portfolio manager on the west coast, and naturally we shared views on the “interesting times,” we live in. One of the things that came up was how quantitative asset managers have lost some of their pre-crisis shine in the wake of the credit crunch cum liquidity freeze cum mortgage implosion cum banking crisis cum recession and how faith in “financial academics” has plummeted along with it. It was only about two years ago in February 2007 when articles like Empire of the Quants in Bloomberg Magazine and others were documenting the extraordinary success of quant firms like BGI and Goldman Sach’s Global Alpha quantitative fund. In June 2007, the New York Society of Security Analysts hosted a discussion on the blending of Quantitative and Fundamental research where the quants present were visibly tripping over themselves trying not to say “we just have a better investment record” in front of their more fundamental-oriented panelists.

Part of the irony of this situation is that we now live in a time where the rigorous review of assumptions and critical analysis that is the hallmark of a good academic is needed now more than ever. I casually joked once that the removal of Lew         Sanders as CEO of Alliance Bernstein was motivated by the the idea that “our research was outstanding, but our performance was terrible, so to change things going forward, we will now be investing without research.” This, incidentally, is not intended as a potshot at current CEO Peter Kraus, who thus far shows no indication of disproportionately devaluing research, but it does use humor to highlight the fallacy of suggesting that too much rigorous thinking is somehow to blame for the mess that we all find ourselves in. When market conditions change - as they have recently - one needs to adjust investment strategy, or maybe change it dramatically, but that is different from abandoning having discipline at all and investing “by the seat of your pants.”

Indeed, in my view, the current crisis is not so much about the failure of investment theory, quantitative or otherwise, but rather the outgrowth of a set of sociological problems - groupthink, the nature of business evolution in a competitive environment which gradually eliminates conservative practices over an extended period of low volatility, lax regulatory oversight and the excess risk-taking it encouraged, and many other things. The problem has not been that strategy design has been “too academic,” but that non-academics simply took academic models “off the shelf” and used them without the awareness and criticism that those who designed them require.

For quantitative funds, one of the biggest challenges has been the unwinding of positions across all asset classes as funds adjust to new leverage requirements constraining their ability to borrow. Even some funds that have performed well and would appear to deserve rewards and client loyalty are faced with redemptions or even closures as clients attempt to sell winners to offset losses and rebalance to target weights in their now-smaller portfolios. This has hit many quant funds: the larger even harder than the smaller.

Even market neutral quant funds have felt the pinch. In these cases, the unwinding and redemptions have pushed down the prices of commonly held long positions and pushed up the prices of short positions, resulting in a simultaneous collapse of long positions plus a “short squeeze” on the balancing short portfolio. As long as different funds are unwinding statistically unrelated positions, the market neutral strategy should hold its own in a liquidity crisis. But if the unwinding is rapid enough, or deep enough, what securities get sold and what securities get covered will inevitably be determined by common factors that are by necessity included in most asset pricing models. When deleveraging prompted by the liquidity freeze on lendable assets is more than just a little, most fund buying and selling necessarily becomes highly correlated with everyone else’s.

Challenges ahead: regime change, and do we have representative data anymore?

I see two big challenges on the road ahead for the quantitative community. To some extent, these are also opportunities, but it will require work to take advantage of them. The challenges are related. The first is whether the investment rules underlying the post-crisis world will look like the investment rules and strategies that worked in the pre-crisis world. My expectation (which I don’t claim to be unique) is that they will change on a number of dimensions with respect to financial regulation, the cost of leverage, the availability of derivative strategies, and the degree to which the financial markets and the real economy are linked. This crisis is big enough that there will almost certainly be a “before” a “during” and and “after” set of both formal and informal rules.

The second challenge is - assuming that there is in fact some kind of shift in underlying rules - how long will we have to wait until there is enough data on the post crisis regime to develop reliable quantitative strategies. For example, a typical rule-of-thumb is that one should seek to have 10 years of monthly returns, 5 years of weekly returns, or 3 years of daily returns before feeling confident that statistical estimates are representative. Two additional complexities arise here: first, quantitative data should generally try to cover a full business cycle, or 4-8 years, in order to smooth out effects that may be cyclical; second, more complex models, including regime shifts and models with interaction terms generally require larger datasets than simpler models.

The changes in “market behavior” from pre-crisis to post-crisis are still a little difficult to specify with any precision. The Obama administration and the US Congress has clearly stated an intent to introduce more regulation of systemic risk, although the details of those plans have not yet come to light. There will be differences in risk appetite, as many investors lick their financial wounds and re-evaluate their situation. Risk taking behavior will likely differ from asset class to asset class, and jittery investors will likely mean that technical indicators and momentum will prove more influential than they had been. These changes thus include both formal changes in the regulatory rules, as well as informal changes in investor behavior and risk appetite.

I don’t see that quantitative asset management cannot recover from the current challenges, but it will take some time to see how the rules of quantitative management have changed, and - for some strategies - even more time to re-estimate parameters to reflect structural shifts in the economy and financial markets. On the other hand, it is also a time where creative and insightful modelers who are attuned to how real investor behavior and financial markets are related can adapt and capitalize on what systematic models can best offer - a way to separate the emotional swings of discretionary trading from the actual practice of investment management.