Quant Memo
Statistics/●●●●●

The best of fifty backtests looks amazing

Asked at Two Sigma

Your team backtested 5050 candidate strategies, ranked them by Sharpe ratio, and promoted the single best one, which showed p=0.02p = 0.02 against zero. You expect it to perform in production much as it did in the backtest.

Why is the promoted strategy's estimated edge likely overstated, and how should you evaluate it instead?

Show a hint

You did not test one strategy. You tested 50 and picked the maximum. What does taking the maximum do to noise?

Your answer

This one is open-ended. Work it through, then check your reasoning against the full solution.

More Statistics questions