The Factor Zoo Problem
The proliferation of factor strategies has attracted serious academic scrutiny. Campbell Harvey, Yan Liu, and Heqing Zhu's 2016 paper found that over 300 distinct factors had been published in academic finance literature. With enough data mining — testing hundreds of variables across enough time periods — some factors will appear statistically significant by pure chance.
The concern is not that all factors are spurious but that the bar for publication has historically been too low, and many published factors reflect historical data mining rather than genuine, persistent return premiums. Researchers now argue that newly proposed factors should pass a much higher statistical bar (a t-statistic of 3.0 rather than the traditional 2.0) to be considered credible.