The upward bias in a reported Sharpe ratio caused by testing many strategies and reporting the best, or by using returns that hide their own risk.
Test 100 random strategies with zero true edge over five years and the best one will show a Sharpe near 1.0 through luck alone. Report only that one and you have manufactured evidence without lying about any single number. This is the trading-specific face of multiple testing, and it is p-hacking by another name.
The correction is a known one: the deflated Sharpe ratio adjusts the reported figure for the number of trials, the length of the sample, and the skew and kurtosis of the returns. Applying it typically cuts backtest Sharpes by half or more, and it routinely converts a headline 1.5 from a wide parameter search into something statistically indistinguishable from zero.
Sharpe is also inflated structurally by illiquid or smoothed marks, by monthly rather than daily data, and by strategies with negative return-skew, whose risk simply is not in the volatility figure. Always ask three questions: how many variants were tried, how long is the sample, and what does the loss distribution look like.
Original diagrams for the ideas on this page. Illustrative, not real market data.
The volatility smile. Options on the same stock and the same expiry are not priced off one volatility. Strikes near the money carry the lowest implied volatility, and it rises towards both ends — usually faster on the downside, which tilts the smile into a skew.Expectancy: the average trade. Forty trades sorted by outcome: 24 small losses and 16 larger wins. Weighting each side by how often it happens gives the average result per trade, marked here by the dashed line at +$120.
Educational only, not advice. Spotted an error? Post in Site Feedback.