The Sortino ratio is the Sharpe ratio with one sensible fix. Sharpe treats all volatility as risk, upside included — a strategy gets marked down for a big winning month exactly as if it had lost. Sortino counts only the downside. It divides return by how much the losing periods scatter, so a system that jumps around on the way up but rarely hurts you on the way down finally gets the credit it is due.
What it is
The Sortino ratio measures return per unit of downside deviation. Downside deviation is like standard deviation, but calculated only from returns that fall below a target — usually zero, or the risk-free rate. Winning periods are left out of the denominator entirely. The logic is plain: you do not lie awake over an unusually good month. Risk is losing money, so measure only that. The metric is named after Frank Sortino, who set it out in the 1980s as a more honest denominator than total volatility.
How it works
Sortino ratio = (average return − risk-free rate) ÷ downside deviation
The numerator is identical to Sharpe. The only change is the denominator: instead of the spread of all returns, you use the spread of the ones that came in below your target. Everything above the target is treated as zero deviation. Fewer and milder downside moves mean a smaller denominator and a higher ratio — which is exactly how it rewards strategies that keep their losses tidy while letting their winners run.
A worked example
Take an EA with six monthly returns: +2%, +6%, −1%, +3%, −2%, +9%. The average is (2 + 6 − 1 + 3 − 2 + 9) ÷ 6 ≈ 2.83% a month. The risk-free rate is 0.3%.
For Sharpe, the standard deviation of all six returns is about 3.8% — the +9% and +6% months inflate it. Monthly Sharpe ≈ (2.83 − 0.3) ÷ 3.8 ≈ 0.67. For Sortino, only the negative months count in the denominator: −1% and −2%. Their downside deviation is far smaller, about 1.6%. Monthly Sortino ≈ (2.83 − 0.3) ÷ 1.6 ≈ 1.58. Same strategy, same returns, but Sortino is more than double, and the whole difference is the two big winners that Sharpe held against it.
Why it matters
Trend-following and breakout EAs live off rare large winners between many small losses. Sharpe punishes the very thing that makes them work. Sortino is fairer because it judges a strategy on how badly it loses, not how dramatically it wins. For that kind of EA the two numbers can disagree sharply, and Sortino is usually the one closer to how the account actually felt to run. Set against Sharpe:
- Sortino is always at least as high as Sharpe on the same data, because downside deviation can never exceed total deviation.
- A wide gap between them tells you the volatility is mostly upside — usually a good sign.
- A narrow gap means the swings are roughly symmetric, so the two metrics largely agree.
Common misconceptions
- It is not immune to the blow-up problem. Like Sharpe, it needs enough losing observations to mean anything, and a strategy that simply has not had its bad month yet will flatter both.
- The target changes the number. Downside is measured against a threshold; move the threshold and the ratio moves. Always check what target a quoted Sortino uses.
- It settles slowly. Because only losing periods feed the denominator, you need a decent run of them before the figure is stable — you learn most about downside risk from downside you would rather not have had.
- A high figure is not a free pass. A great Sortino on 20 trades is still built on 20 trades.
The Karnek note
Karnek computes the Sortino ratio from your live terminal returns over a labelled window, on the same trade population as the return shown beside it. Reading read-only means the downside it measures is the real one your account felt, and it can never trade the account.
Written and reviewed by the Karnek Research team. Last updated August 2026.
Educational content only - not financial advice. Past performance does not predict future results. Trading carries significant risk.