Sharpe vs Sortino vs Calmar: which risk-adjusted ratio to trust

A raw return number tells you nothing useful — a strategy that returned 60% by risking ruin is worse than one that returned 25% calmly. Risk-adjusted ratios exist to make that comparison. The three you will meet are Sharpe, Sortino, and Calmar. They disagree, on purpose, because each defines 'risk' differently. Knowing what each one penalises is the difference between reading a backtest and being fooled by one.

Sharpe — return per unit of total volatility

The Sharpe ratio divides excess return by the standard deviation of returns. It is the default, it is universal, and it has one quietly large flaw: it treats all volatility as bad. A strategy that frequently posts big up-days is punished by Sharpe exactly as if those were big down-days, because standard deviation does not care about sign. For a strategy with a positively skewed return profile — small losses, occasional large wins — Sharpe understates how good it is.

Sortino — return per unit of downside volatility

The Sortino ratio fixes Sharpe's main flaw. It divides excess return by the downside deviation — the volatility of negative returns only. Upside swings no longer count against you. For a strategy that makes its money in occasional sharp rallies, Sortino gives a fairer, usually higher, score. The catch: if a strategy has very few losing periods, the downside-deviation denominator gets tiny and Sortino can balloon to a number that looks spectacular and means little. A huge Sortino on a short track record is a sample-size artefact, not an edge.

Calmar — return per unit of pain

The Calmar ratio takes a completely different angle: annualised return divided by the maximum drawdown. It ignores day-to-day volatility entirely and asks one question — how much return did you get for the worst peak-to-trough loss you had to sit through? That makes Calmar the most psychologically honest of the three, because max drawdown is the number that actually makes people abandon a strategy. Its weakness is the mirror of its strength: it is built on a single historical event, the worst drawdown, so it is noisy and easy for one lucky (or unlucky) period to dominate.

How to use all three

Do not pick a favourite — read them as a panel, and pay attention when they disagree:

• Sharpe high but Sortino much higher — the strategy has positive skew (its volatility is mostly upside). Good news Sharpe alone would have hidden.
• Sharpe decent but Calmar poor — smooth most of the time but with one brutal drawdown. You may not survive that drawdown emotionally even if the average looks fine.
• Sortino enormous on a short backtest — almost certainly too few losing periods to estimate from. Distrust it; get more data.
• All three modest but consistent across walk-forward windows — the most trustworthy result of all. Consistency beats a single big number.

Noon Barbari reports Sharpe, Sortino, and Calmar — plus max drawdown, drawdown duration, and the deflated Sharpe — on every backtest, and shows each ratio across every walk-forward window so you can see whether they hold up out of sample. The free tier includes the full metric panel.

One ratio is a headline. Three ratios, read together and checked across out-of-sample windows, are an actual assessment. Strategies that look good on all three, consistently, are rare — and that rarity is exactly the point.