Risk-Adjusted Returns: The Arithmetic and the Catch

A return on its own is not comparable to anything. Two funds that both returned 12% are not equivalent if one did it with the steadiness of a bond ladder and the other by nearly halving twice. Risk-adjusted return is the attempt to put both on one scale, and every version of it is the same shape: excess return on top, some measure of variability underneath.

Sharpe ratio

Sharpe = (return − risk-free rate) ÷ volatility

The numerator is what you earned above cash, because earning 4% when Treasury bills pay 4% is not skill. The denominator is the annualised standard deviation of returns — the same figure described in the volatility article.

Worked through: a fund returning 10% against a 4% risk-free rate with 13.0% volatility has a Sharpe of (10 − 4) ÷ 13.0 ≈ 0.46. The same 10% earned with the volatility of a single volatile megacap — 37.8% — gives 0.16. Identical return, very different quality.

Sortino ratio

Sharpe's denominator punishes upside and downside equally, which is odd: nobody complains about a violent gain. Sortino replaces total volatility with downside deviation, computed only from returns below a threshold.

The practical consequence is that Sortino flatters anything whose big moves are mostly upward and treats a steady grind with occasional collapses far more harshly than Sharpe does. Which you prefer depends on whether your worry is variability or loss.

What they cannot see

• Order. Shuffle a return series into any sequence and Sharpe is unchanged — but "down 30% then up 50%" and "up 50% then down 30%" are not the same experience, and if you were withdrawing they are not the same outcome either.
• Fat tails. Standard deviation assumes moves cluster the way a normal distribution says. Real returns have far more extreme days than that, so a strategy that sells insurance looks superb right up until it doesn't.
• The measurement window. A ratio over three years and the same ratio over ten routinely disagree, and the shorter one is quoted more often precisely because it can be chosen after the fact.
• Correlation with what you already hold. A high Sharpe on its own says nothing about whether adding it diversifies you or concentrates you further. That question needs the correlation, not the ratio.

Maximum drawdown, the one people actually feel

Drawdown is the largest peak-to-trough fall over a period. It is not risk-adjusted anything, and it correlates poorly with Sharpe — but it is the number that decides whether a strategy gets abandoned. A portfolio held through a 50% fall and a portfolio sold at the bottom have identical statistics and completely different results.

Using them honestly

These ratios are for comparing things of the same kind over the same window: two funds with the same mandate, or one strategy against its benchmark. Comparing a bond fund's Sharpe with an equity fund's rewards the asset class, not the manager. And every one of them is computed on the past, with no clause obliging the future to cooperate.

Figures on this page are measured from daily returns over the year to 2026-09-22, across 5,262 liquid US securities, and are rebuilt nightly.

Related: volatility · Backtest errors · Back to all articles