Sharpe Ratio

The Sharpe ratio divides a strategy's excess return by the volatility of those returns:

Sharpe = (R − Rf) / σ

where R is the strategy's average return, Rf the risk-free rate, and σ the standard deviation of returns. Computed on per-period returns, it is annualized by scaling with the square root of periods per year:

annualized Sharpe = mean(r) / std(r) × √N

(N = 365 for daily crypto returns, 252 for equities.)

Reading the number

As a rough retail yardstick on out-of-sample results: below 0.5 is noise, 0.5 to 1 is workable, 1 to 2 is good. Sustained numbers above 2 deserve suspicion of overfitting before celebration, because in-sample Sharpe is easy to manufacture.

Limitations

Sharpe treats upside and downside volatility identically, and it says nothing about the shape of losses: two strategies with equal Sharpe can have very different max drawdowns. It also assumes returns are roughly normal. Strategies with rare catastrophic losses (the short-vol shape) post flattering Sharpes right up until they don't, so read it alongside the drawdown and the equity curve rather than on its own.

On AlphaProve

Because a single headline Sharpe hides whether the edge was steady or came from one hot streak, the tearsheet also plots a rolling 30-trade Sharpe: the ratio recomputed over a moving window of the most recent trades. A line that stays near the headline value throughout is a different signal from one that spikes early and then flatlines, even when both runs report the same overall number. It sits next to Sortino, Calmar, and Omega, which reweight the same returns to penalize the downside and tail losses plain Sharpe treats as ordinary volatility.