A Bayesian Approach to Measurement of 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 overfitting such as out-sample backtesting turn out be unreliable in situations when the selection is based results too many holdout sample. There an ongoing discussion how estimate probability backtest adjust expected performance indicators Sharpe ratio order reflect properly effect multiple testing. We propose consistent Bayesian approach that yields desired robust estimates basis Markov chain Monte Carlo (MCMC) simulation. The technical trading where seemingly profitable strategy can naïve approach.