Analysis & metrics

Sharpe Ratio

By Jarosław Wasiński • Published: Apr 2, 2026 • Updated: Oct 11, 2026

A risk-adjusted return measure: the mean return above a stated benchmark, commonly a risk-free return, divided by the standard deviation of that differential return. Specify the observation frequency, sample, costs and annualisation method. It treats upside and downside variation alike. A higher estimate is not proof of a durable edge, and fixed labels such as 'good above 1' are not universal standards. Serial dependence and non-normal returns affect interpretation; compare like-for-like conventions.

Example. Assume a mean monthly return after costs of 0.8% and a benchmark return of 0.2% per month. Mean differential return is 0.6%; with a standard deviation of 2%, monthly Sharpe = 0.6/2 = 0.30. Under a simplified linear-aggregation model with uncorrelated periods, multiplying by √12 gives approximately 1.04 annually. This is neither CAGR nor a promised result.

Calculate differential returns

For each period, subtract the selected benchmark return, commonly a suitable risk-free return, from the strategy return. The historical ratio is the mean of those differences divided by their standard deviation. State currency, frequency, standard-deviation convention and costs. USD P&L per trade is not automatically an account’s periodic return. Definition source: William F. Sharpe, The Sharpe Ratio.

Annualisation needs assumptions

Do not mix a monthly mean with an annual standard deviation. Multiplication by the square root of the period count assumes a specified aggregation model, stable parameters and the relevant absence of autocorrelation. Correlated results, smoothed valuations and compounding complicate the conversion. See Andrew W. Lo, The Statistics of Sharpe Ratios. For irregular trades, do not substitute the trade count for the period count without justification.

Limitations and common mistake

Standard deviation includes favourable and adverse departures from the mean; it is not a loss-only measure. Sharpe does not describe every distribution tail, gap or drawdown depth. A short or selected sample can exaggerate the assessment. Labels such as “above 1 is good” or “above 2 is safe” are not universal standards. With zero standard deviation, the ratio is undefined.

Do not select active periods only

In monthly account-performance analysis, months without new trades still belong to the series: they may contain open positions, interest and charges. Removing them changes the strategy being measured and its frequency. If the benchmark varies between periods, calculate the standard deviation of differential returns rather than mechanically using strategy returns alone. Keep the complete series and explain cash, unrealised valuation and missing observations.

Use it in an audit

Align currency, benchmark, measurement window and valuation method, disclosing uncertainty and open positions. Read the result alongside drawdown, exposure, Sortino and extreme events; do not select a system from a single number. In the course: journals and metrics. Sources checked on 11 October 2026.

FAQ

Does higher Sharpe always mean a better strategy?

No. Check method comparability, uncertainty, hidden risks and the investor’s objective.

Can the benchmark be zero?

Yes, as a disclosed analytical convention. Do not automatically describe it as excess return over an actual risk-free investment.

← Back to the full glossary