Most quant models are based on historical data. Even those using analysts' forecasts or other "sentiment" signals turn out to depend heavily on the past because sentiment usually is biased in the direction of historical trends. Regardless of the type of model, quants use past relationships and behavior to develop theories and build models to help predict the future. If markets have behaved in a particular way for awhile, quants will come to depend on that behavior persisting. If there is a regime change, the quant will typically suffer because the relationships and behavior he is counting on are altered, at least temporarily.

Dependence on the past is certainly one of the more interesting problems to consider in analyzing quant strategies and determining how to use them. In some cases, dependence on the persistence of historical behavior is blatant, as in the case of trend-following strategies. Note that this isn't necessarily an indictment of these strategies. Indeed, such strategies have made money for decades and have exhibited better risk-adjusted returns than the stock market by far. However, if an established trend reverses, the trend follower will almost certainly lose money. Ironically, mean reversion-focused quants may also suffer during a large trend reversal, particularly if they are engaged in a relative mean reversion strategy. We might expect that if a reversal of trend occurs, this should be good for the mean reversion trader, since he bets ...

Get Inside the Black Box: The Simple Truth About Quantitative Trading now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.