Nobody ever lost sleep over a trade that made too much money. Yet the Sharpe ratio penalizes exactly that. The Sortino ratio fixes the mistake by measuring only the risk that actually hurts.
The flaw in Sharpe that Sortino fixes
The Sharpe ratio uses standard deviation as its measure of risk, and standard deviation is symmetric — it treats a surprise +6R winner as just as “risky” as a −3R loser. Both widen the spread of your returns, both drag Sharpe down.
For a trader this is upside-down. Volatility to the upside is the entire point. A strategy that occasionally catches enormous runners is better, not riskier, yet Sharpe marks it down for the privilege. Sortino corrects this by ignoring upside entirely and scoring you only on the returns that fell short. It answers a sharper question: for each unit of bad surprise, how much return did I earn?
Downside deviation: only returns below your target
The engine of the Sortino ratio is downside deviation — a version of standard deviation that counts only the returns below a threshold you choose, called the minimum acceptable return (MAR) or target return.
The mechanics:
- For each period, compute how far the return fell below the target. If it met or beat the target, that period contributes zero — not a negative, not a small positive, exactly zero.
- Square those shortfalls, average them across all periods (including the good ones, which count as zeros), and take the square root.
That last detail trips people up. You divide by the total number of periods, not just the losing ones. Losing rarely is supposed to be rewarded, and averaging over the full series is what captures it. The result is a number that grows only when your returns dip below what you consider acceptable — a clean measure of downside pain.
The Sortino formula step by step
Putting it together:
Sortino = (mean return − target return) / downside deviation
Step by step, from a series of daily or per-trade returns:
- Pick your target return (the MAR). More on choosing it below.
- Compute the mean of your actual returns.
- Subtract the target from the mean — that is your excess return over the bar you set.
- Compute the downside deviation against that same target, as described above.
- Divide. Annualize the same way as Sharpe, multiplying by √N where N matches your sampling frequency.
Keep the target consistent between the numerator and the downside-deviation calculation. Mixing a zero target in one and a risk-free rate in the other produces a number that means nothing.
Choosing a minimum acceptable return (MAR)
The MAR is where Sortino becomes personal, and picking it honestly matters more than the arithmetic.
- Zero is the most common choice: any losing period counts as downside, any flat-or-up period is fine. Clean and intuitive for active traders.
- The risk-free rate treats “worse than parking cash in T-bills” as the failure line. Fine for longer horizons, overkill for intraday.
- A required daily rate ties the target to a goal — for example, the pace you need to pass a challenge or clear a payout window. This makes Sortino answer a prop-specific question: how often, and how badly, do I fall behind the pace I actually need?
Whatever you pick, write it down and never move it between measurements. A MAR you quietly lower after a bad month is just Sharpe with extra steps.
Reading Sortino vs Sharpe as a pair
Sortino is most powerful next to Sharpe, not instead of it. The gap between the two tells you about the shape of your returns.
| Pattern | What it means |
|---|---|
| Sortino ≈ Sharpe | Returns are roughly symmetric — winners and losers are similar in size and spread. |
| Sortino ≫ Sharpe | Positive skew — your big surprises are winners; Sharpe was unfairly punishing your upside. |
| Sortino < Sharpe | Negative skew — small frequent gains, rare large losses. The classic hidden-risk profile. |
That last row is the one that ends funded accounts. A high Sharpe with a lower Sortino is the statistical fingerprint of a strategy that grinds out steady wins and then hands them all back in one violent trade — the martingale trap in disguise. Watching Sortino fall below Sharpe is an early warning that your left tail is getting fat.
This is where Shibiki’s approach pays off. Because it auto-journals every fill and grades live edge health with a Wilson confidence interval, the downside profile is measured continuously rather than reconstructed after a blowup. And because it pushes hard risk limits down to the broker-side EA, the single catastrophic loss that wrecks your Sortino — and your trailing drawdown — is capped before it lands, no matter how convinced you are that price is about to come back. Pair that with a drawdown calculator to see how much downside your firm’s rules actually tolerate before a breach.
Where Sortino still falls short
Sortino is better than Sharpe, not perfect.
- It ignores drawdown clustering. Downside deviation treats scattered losing days the same as a run of them stacked back to back — but consecutive losses are what breach a trailing drawdown. For that, Calmar (return over max drawdown) speaks more directly.
- It needs a decent sample. On a handful of trades the downside deviation is dominated by one or two bad prints and the ratio swings wildly. Treat any Sortino from a short sample as provisional.
- It is target-sensitive. Move the MAR and the whole number moves. That is a feature when you set it honestly and a bug when you fish for a flattering figure.
Read Sortino as one instrument on the panel — beside Sharpe, expectancy, and max drawdown. Together they describe risk-adjusted return from every angle; alone, each one lies a little. An expectancy calculator anchors the whole picture in per-trade edge before you start scoring the smoothness of the ride.
Related: Trailing drawdown · Drawdown calculator · Expectancy calculator