The Sharpe ratio is annual excess return divided by annual volatility. It's the simplest risk-adjusted return metric, the one every quant book teaches first, and the one every retail blog reaches for when comparing strategies. It does honest work as a sanity check.
It does dishonest work as an optimization target.
Sharpe assumes returns are roughly normally distributed. Real return distributions aren't. The strategies that generate the highest Sharpe ratios are often the ones that quietly accumulate small wins by selling tail risk — picking up pennies in front of the steamroller, in the cliché. A high-Sharpe strategy that sells volatility looks brilliant for years and then gives back its Sharpe in one bad month. Optimizing for Sharpe selects for exactly this failure mode.
The fix isn't to abandon Sharpe; it's to read it as one signal among several. I look at Sharpe alongside max drawdown, the worst rolling 90-day return, and the distribution of monthly outcomes. A strategy with a Sharpe of 1.4 and a max drawdown of 8% is meaningfully different from a strategy with a Sharpe of 1.4 and a max drawdown of 35%. Sharpe alone can't tell you which.
The number is honest about what it measures. The number lies about what it doesn't.