LEARN · EN · easyquanttrading.com
Sortino vs Sharpe: which volatility should count against you?
Both ratios divide a return by a measure of how much that return bounced around. They differ in one respect: Sharpe counts every large move against you, and Sortino counts only the large downward ones. That single change reorders the rankings.
- Sharpe uses total standard deviation; Sortino uses downside deviation
- Sortino is kinder to strategies with occasional large wins
- Both ignore the sequence of returns
- Choose which one you use before looking at the numbers
The difference in one line
Sharpe compares average return to the standard deviation of all returns. Sortino compares average return to the standard deviation of returns below a target, usually zero.
The consequence is that a strategy with occasional very good months is penalised by Sharpe and barely affected by Sortino. For trend-following systems, whose returns genuinely are asymmetric in that way, the two ratios can disagree substantially.
Side by side
Neither is more correct. They answer different questions about what constitutes bad news.
| Sharpe | Sortino | |
|---|---|---|
| Denominator | Standard deviation of all returns | Downside deviation only |
| Penalises upside | Yes | No |
| Best suited to | Roughly symmetric return streams | Asymmetric, positively skewed returns |
| Common in | Academic work, fund reporting | Retail and systematic trading discussion |
| Blind to sequence | Yes | Yes |
| Sensitive to sample length | Yes | Yes, and more so when downside events are rare |
The last row matters: with few large losses in the sample, downside deviation is poorly estimated.
Why the choice must be made in advance
If you compute both and report whichever looks better, you have added a selection step to your evaluation. With two ratios there is a fifty percent chance of picking the friendlier number by accident, and the correction for that is the same multiple-testing problem that affects strategy search.
Decide in advance based on the shape of your return distribution. If your strategy's returns are roughly symmetrical, Sharpe is the natural choice. If it relies on a few large wins, Sortino describes the risk you actually care about more accurately.
What neither of them tells you
Neither ratio captures sequence. A strategy that loses steadily for a year and then recovers can have an excellent Sharpe and a terrible experience. Maximum drawdown and drawdown duration are the measures that capture what you will actually feel.
Both are also sensitive to the sample. Computed over three months they are close to meaningless, because the denominator is estimated from a handful of observations. Reporting either without the sample length is reporting half a number.
A worked comparison
Suppose two strategies both average 1% per month over three years. The first returns a steady 1% with small variation in both directions. The second returns nothing for two months and then 3% in the third, repeatedly.
Their Sharpe ratios will differ modestly, because the second has more total volatility. Their Sortino ratios will differ much more, because most of the second strategy's variation is on the upside. Selecting on Sharpe would prefer the first; on Sortino the second might win.
Neither answer is wrong. The question is whether you consider a month of +3% followed by two flat months to be risky. If you do — because it tempts you to increase size — Sharpe is capturing something real. If you do not, Sortino is the more honest description of your risk.
The practical recommendation
Report both, decide in advance which one your selection rule uses, and pair it with drawdown regardless. The ratio tells you about the shape of the return distribution; the drawdown tells you whether you can survive it. Choosing a strategy needs both pieces, and neither substitutes for the other.
FAQ
- Is Sortino always higher than Sharpe?
- Usually, because downside deviation is smaller than total deviation whenever there are any gains at all. That means the two are not directly comparable across strategies, and a Sortino of 2.0 is not equivalent to a Sharpe of 2.0.
- Which do professional investors use?
- Sharpe is far more common in institutional reporting, largely for historical reasons and because it is better understood. Sortino appears more often in systematic trading literature, where asymmetric return distributions are the norm rather than the exception.
- Can either be computed on a backtest with few trades?
- They can be computed, but they should not be trusted. With fewer than a few dozen observations, both ratios are dominated by estimation error, and a single unusual trade can move them substantially.
More guides
- How EasyQuant validates strategies — evidence you can filter
- Honest backtesting, not pretty curves
- Gold strategy research that stays honest
- Overfitting detection: catch it before you deploy
- System Forge: design, then prove
- Walk-forward analysis: the only backtest that fights overfitting
- MT5 export without custody
- Glass box, not black box AI signals
Not investment advice. Historical results do not guarantee future performance. EasyQuant is a research factory — you execute on accounts you control.