Sharpe ratio

Measures return per unit of risk: (return − risk-free rate) divided by standard deviation. It lets you compare strategies, with important limitations.

The Sharpe ratio, created by William F. Sharpe, measures how much return above a risk-free investment you earn for each unit of volatility taken. Its formula is (Rp − Rf) / σp, where Rp is the strategy’s average return, Rf the risk-free rate, such as Treasury bills or the short-term interest rate, and σp the standard deviation of returns, all in the same time units.

Example with monthly data: average return of 1.0%, monthly standard deviation of 3.0% and an annual risk-free rate of 3%, which is 0.25% per month. The monthly Sharpe is (1.0 − 0.25) / 3.0 = 0.25. To annualise, multiply by the square root of 12: 0.25 × 3.464 ≈ 0.87. The result matches if you work annually: simple return of 12%, volatility of 3% × 3.464 ≈ 10.39%, and (12 − 3) / 10.39 ≈ 0.87. With daily data you multiply by the square root of 252.

To interpret it, a value of 0 means you are not compensated for risk compared with holding a safe asset; below 0, you earn less than that asset. As a rough guide, above 1 is usually considered good, but there is no universal scale and it depends on strategy type, period and market. Very high Sharpe ratios on little data are usually a warning of overfitting, not an achievement.

The main limitation is estimation. With an annual Sharpe of 0.87 and 3 years of data, the approximate standard error is the square root of (1 + 0.5 × 0.87²) / 3 ≈ 0.68, which implies a very wide confidence interval, from almost zero to much higher figures. A strategy with only a few years of history cannot reliably separate skill from luck.

Other limitations: it uses standard deviation, which penalises unexpected upside as much as downside, so it does not capture downside risk well (the Sortino ratio exists for that); it assumes roughly normal returns and misses the tail risk of strategies that earn a little at a time and lose a lot at once, such as martingales or option selling; and it can be inflated by smoothed valuations or autocorrelated returns.

Practical details: use the same period for return and volatility, calculate on account returns with commissions included, and compare only strategies with comparable horizons and leverage. Leveraging a strategy multiplies return and volatility together, so the Sharpe ratio does not change, although the risk of ruin grows considerably.

In risk management, use it as one metric among several and never the only one: combine it with maximum drawdown, profit factor, expectancy and out-of-sample analysis. An acceptable Sharpe ratio with a drawdown you cannot tolerate is still an unsuitable strategy for you.

More terms