Traditional methods for analyzing portfolio returns often rely on multifactor risk assessment, and tests of significance are typically based on variants of the t‐test. This approach has serious limitations when analyzing the returns from dynamically traded portfolios that include derivative positions, because standard tests of significance can be ‘gamed’ using options trading strategies. To deal with this problem we propose a test that assumes nothing about the structure of returns except that they form a martingale difference. Although the test is conservative and corrects for unrealized tail risk, the loss in power is small at high levels of significance.