When Systematic Strategies Decay
47 Pages Posted: 18 May 2021 Last revised: 1 Jul 2021
Date Written: May 14, 2021
In this paper, we ask which ex-ante characteristics empirically predict the out-of-sample drop in risk-adjusted performance of published stock anomalies. Our sample is a large cross- section of anomalies published in finance and academic journals, and we define out-of-sample as the post-publication period. Our set of predictors of OOS decay is inspired by two standard hypotheses: arbitrage capital flowing into newly published strategies, and in-sample overfitting linked to multiple hypothesis testing. The year of publication alone – compatible with both hypotheses – explains 30% of the variance of Sharpe decay across predictors: Every year, the Sharpe decay of newly-published factors increases by 5ppt. The other important variables are directly related to overfitting. One of our two “complexity” variables, the number of operations required to calculate the signal, predicts a significant drop of performance. Also, both measures of in-sample “sensitivity to outliers” are shown to matter. Together, these three overfitting- related variables add another 15% of explanatory power. Some arbitrage-related variables are statistically significant, but their predictive power is marginal.
Keywords: factor returns, arbitrage, overfitting
JEL Classification: G14, G15, G40
Suggested Citation: Suggested Citation