Stock Portfolio Design and Backtest Overfitting

Abstract

We demonstrate a computer program that designs a portfolio consisting of common securities, such as the constituents of the S&P 500 index, that achieves any desired profile via in-sample backtest optimization. Unfortunately, the program also shows that these portfolios typically perform erratically on more recent, out-of-sample data, which is symptomatic of selection bias. One implication of these results is that so-called smart beta funds, which are designed in-sample to deliver a desirable performance pro file, are likely to disappoint out-of-sample.

Keywords: backtest, historical simulation, probability of backtest overfitting, investment strategy, optimization, Sharpe ratio, minimum backtest length, performance degradation

JEL Classification: G0, G1, G2, G15, G24, E44