What Is Placebo test?

A placebo test is a negative-control test in which the claimed source of predictive power is deliberately replaced, shuffled, or time-reversed. In quantitative research, examples include permuting labels within suitable time blocks, testing a signal against backward-looking returns, or replacing a proposed feature with a random feature while keeping the modelling and selection procedure unchanged.

The expected result is no meaningful predictive or economic performance. A placebo failure can reveal data leakage, autocorrelation mistaken for prediction, or an overly flexible selection process. Passing a placebo test is necessary evidence against these failure modes, but it does not by itself prove a strategy is valid or profitable.

See also

• Placebo label

• Look-ahead bias

• Out-of-sample testing

Research and literature

• David H. Bailey, Jonathan M. Borwein, Marcos López de Prado, and Qiji Jim Zhu, Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance, Notices of the American Mathematical Society 61(5), 2014.