A Bayesian Approach to Backtest Overfitting

Quantitative investment strategies are often selected from a broad class of candidate models estimated and tested on historical data. Standard statistical techniques to prevent model over-fitting  turn out to be unreliable... [ view full abstract ]

Quantitative investment strategies are often selected from a broad class of candidate models estimated and tested on historical data. Standard statistical techniques to prevent model over-fitting  turn out to be unreliable when selection is based on too many models tested on the holdout sample. There is an ongoing discussion how to estimate the probability of back-test over-fitting and adjust the expected performance indicators in order to reflect properly the effect of multiple testing. We propose a consistent Bayesian MCMC approach that yields the desired robust estimates. The approach is tested on a class of technical trading strategies.