What is the Sharpe Ratio?

The Sharpe ratio asks whether the return was worth the ride. It takes the return above a risk-free rate and divides it by how much the returns bounced around. Two EAs can both post 20% a year; the one whose equity crept up steadily has the higher Sharpe, and the one that lurched to get there has the lower. A bigger number means more reward for each unit of volatility you had to stomach.

What it is

The Sharpe ratio measures return per unit of total volatility. Volatility here is the standard deviation of the strategy's returns — a measure of how widely they scatter around their average. A system that returns a steady 1% a month has low volatility; one that swings between +8% and −6% has high volatility, even if it ends the year in the same place. The ratio rewards the steady one. It was introduced by William Sharpe in 1966 and remains the default way of comparing returns on a like-for-like risk basis.

How it works

Sharpe ratio = (average return − risk-free rate) ÷ standard deviation of returns

Three pieces feed it. The average return is the mean over your chosen period, say monthly. The risk-free rate is what you could earn for no risk over the same period — from short-dated government bills, for instance. It is the hurdle: return you would earn just for showing up should not count as skill. The standard deviation measures how much the returns scatter. The figure is usually annualised so numbers can be compared; to annualise a monthly Sharpe, multiply by the square root of 12.

A worked example

An EA returns an average of 1.5% a month. The risk-free rate is 0.3% a month, and the standard deviation of its monthly returns is 2.0%.

Monthly Sharpe = (1.5 − 0.3) ÷ 2.0 = 1.2 ÷ 2.0 = 0.6. Annualised, that is 0.6 × √12 ≈ 2.08 — strong on paper. Now take a second EA with the same 1.5% average monthly return but a standard deviation of 4.5%. Its monthly Sharpe is (1.5 − 0.3) ÷ 4.5 = 0.27, or about 0.92 annualised. Same return, less than half the Sharpe, because it put you through more than twice the turbulence to get there.

Why it matters

As a rough guide, using annualised figures:

Below 1 — the return is not paying you well for the volatility.
Above 1 — decent.
Above 2 — strong.
Above 3 — exceptional, and worth checking whether it is real or the sample is simply too short.

For judging an EA, the Sharpe is a fair first filter: it puts two systems on the same footing instead of letting the flashier return win by default. But it is a filter, not a verdict, and it should sit beside drawdown and the trade count rather than stand alone.

Common misconceptions

It punishes upside as if it were risk. Standard deviation counts a big winning month exactly like a big losing one, so a strategy with occasional large wins can score a mediocre Sharpe despite never hurting you on the downside. The Sortino ratio was built to fix this.
It assumes returns are roughly normal. Grid and martingale EAs produce long calm stretches followed by a rare disaster. Sharpe reads the calm as low risk and flatters them right up until the blow-up.
It needs enough data. A Sharpe from six trades or two months is noise dressed up as a statistic.
It can be gamed by the return period. Measuring monthly instead of daily smooths the numbers and can lift the figure without anything changing in the strategy. Always check what period and window a quoted Sharpe uses.

The Karnek note

Karnek works out the Sharpe ratio from your live terminal returns, over a labelled window, on the same trades as the figures shown beside it. Because it reads read-only, the number reflects real fills rather than a tidy backtest, and it can never place a trade or touch the account.

See it in Karnek: Karnek's live metrics dashboard tracks the figures in this guide, straight from your terminal.