Sortino Ratio - Returns Adjusted Only for Downside
By EC Assets Research Team, Performance Analytics · Published
Sortino Ratio: The Sortino ratio is a risk-adjusted return measure that penalises only downside volatility, treating upside variation as desirable rather than risky. It modifies the Sharpe ratio by replacing total standard deviation with downside deviation below a target return.
What the Sortino Ratio Measures
The Sortino ratio is the Sharpe ratio with a different idea of risk in the denominator. Both divide a strategy's return above a reference rate by a measure of how much it moved around. Sharpe uses total volatility, so a month that beats expectations by four points counts against the strategy exactly as much as a month that misses by four. Sortino counts only the misses: it divides by downside deviation, the dispersion of returns below a stated target.
That one change matters whenever a return stream is lopsided. A strategy whose volatility comes mostly from large gains is penalised by Sharpe for being surprised in the direction the investor wanted; Sortino does not penalise it. A strategy that earns steadily and occasionally gives back a lot looks better on Sharpe than its history deserves, and Sortino, which looks only at the losing side, is harder to flatter.
The measure grew out of Frank Sortino's work on downside risk in the 1980s and took its standard form in his papers with Robert van der Meer (1991) and Lee Price (1994).
How It Works
$$\text{Sortino} = \frac{\bar{R} - T}{DD}, \qquad DD = \sqrt{\frac{1}{N}\sum_{t=1}^{N} \min(R_t - T,, 0)^2}$$
Three inputs decide the number:
| Input | What it is | Common choices |
|---|---|---|
| Target T | The return below which a period counts as a shortfall | Zero, the cash rate, or a hurdle rate |
| Downside deviation | Root mean square of the shortfalls below T, averaged over all periods | Monthly data, annualised with √12 |
| Numerator | Mean return minus the same T | Same window as the denominator |
The target appears twice, in the numerator and inside the denominator, so it has to be stated with every published figure.
The averaging step is where implementations differ. Downside deviation divides the summed squared shortfalls by the number of all periods, N, not by the number of losing periods. Over 60 months with 15 below target, dividing by 15 instead of 60 doubles the downside deviation (√(60/15) = 2) and halves the ratio. Both versions circulate in practice; only the first matches the original definition.
Annualising follows the square root of time rule: the monthly mean excess return times 12, divided by the monthly downside deviation times √12, which is simply the monthly ratio times √12.
Worked Example
Two strategies over 24 months. Cash earns 0.3 percent a month, and that risk-free rate serves as the target for both ratios, so the two measure against the same reference point:
| Monthly statistic | Strategy A | Strategy B |
|---|---|---|
| Mean return | 0.8% | 0.8% |
| Volatility | 2.0% | 2.5% |
| Downside deviation (T = 0.3%) | 1.2% | 0.9% |
| Sharpe, annualised | 0.87 | 0.69 |
| Sortino, annualised | 1.44 | 1.92 |
Same mean, opposite verdicts: B is more volatile, but less of its volatility sits below the target.
Sharpe for A is (0.8 - 0.3) / 2.0 × √12 = 0.87; for B, (0.8 - 0.3) / 2.5 × √12 = 0.69. Sortino for A is 0.5 / 1.2 × √12 = 1.44; for B, 0.5 / 0.9 × √12 = 1.92.
The ratio of downside deviation to volatility is the useful diagnostic. A normal distribution with A's mean and volatility would put downside deviation at about 0.57 of volatility; A's actual 0.60 is close to that, so A is roughly symmetric. For B the symmetric benchmark is about 0.60 and the actual figure is 0.36: B's dispersion is concentrated in its good months. Whether that reflects skill or a few fortunate months is a question no single ratio answers.
The same logic, annualised, for three managers with identical Sharpe ratios:
| Manager | Return | Volatility | Downside deviation | Sharpe | Sortino |
|---|---|---|---|---|---|
| A, symmetric | 10% | 12.5% | 7.6% | 0.64 | 1.05 |
| B, positive skew | 10% | 12.5% | 5.0% | 0.64 | 1.60 |
| C, negative skew | 10% | 12.5% | 10.5% | 0.64 | 0.76 |
Cash and target at 2 percent throughout. Sharpe cannot tell the three apart; Sortino ranks them by the shape of their losses. The 7.6 percent for A is what a normal distribution with these inputs produces from monthly data.
When It Applies (and Limitations)
It needs enough losing periods. The denominator is estimated from shortfalls only. Five years of monthly data with ten months below target is a small sample: if those ten shortfalls are of similar size, one more month three times as deep raises the downside deviation by more than a third.
It is blind to risk that has not happened yet. A strategy that sells far out-of-the-money puts can run for years without a meaningful shortfall. Its downside deviation is tiny and its Sortino spectacular until the month that defines the strategy arrives. A low downside deviation is evidence about the sample, not about tail risk.
The target must match across comparisons. A Sortino computed against zero is not comparable to one computed against cash or a hurdle. The same data give materially different values; a figure without its target is not a figure.
Upside is not automatically free. Treating gains as riskless assumes they are repeatable. A backtest with one exceptional month earns a high Sortino without earning any confidence.
Why It Matters for Institutional Investors
Liabilities define the target. A pension scheme discounting its liabilities at 5 percent cares about returns below 5 percent; returns above it are not a risk to the scheme. Sortino with T set to the liability rate measures exactly that, which is why it suits liability-driven portfolios better than Sharpe.
Asymmetric strategies need it. Long-volatility and trend-following programmes earn much of their return in a few violent months. Sharpe reads that as risk; Sortino reads it as the purpose of the strategy. An allocator who ranks such managers on Sharpe alone will underrate them, and rate short-volatility programmes too highly.
The gap between the two ratios is information. A manager with a Sharpe of 0.8 and a Sortino of 1.8 is showing where the volatility lives. Track the gap across periods: a stable gap describes the strategy, a sudden one describes a handful of months.
It belongs in a set, not alone. Sharpe for dispersion, Sortino for the shape of losses, the Calmar ratio and maximum drawdown for the worst path actually travelled. The ratio calculator computes all three from the same inputs.
References
- Sortino, F. A., & Price, L. N. (1994). Performance Measurement in a Downside Risk Framework. Journal of Investing, 3(3), 59-64.
- Sortino, F. A., & van der Meer, R. (1991). Downside Risk. Journal of Portfolio Management, 17(4), 27-31.
- Sharpe, W. F. (1994). The Sharpe Ratio. Journal of Portfolio Management, 21(1), 49-58.
- Bacon, C. R. (2008). Practical Portfolio Performance Measurement and Attribution (2nd ed.). Wiley.
Frequently asked questions
Is a higher Sortino ratio always better?
Only when the target and the period match. A Sortino of 2.5 measured against zero is not comparable with 1.8 measured against the cash rate, because a higher target raises the downside deviation and lowers the numerator at the same time. Check the target before comparing the numbers.
What happens if no return falls below the target?
The downside deviation is zero and the ratio is undefined. Over any realistic window that only happens when the sample is too short or the returns are too smooth, for example appraisal-based valuations of illiquid assets. Treat it as a warning about the data, not as a perfect score.
Why not simply use maximum drawdown?
Maximum drawdown is a single observation: the worst peak-to-trough path. Downside deviation uses every shortfall in the sample. The first describes the worst case actually lived through, the second the typical shape of losses, and a sound assessment reads both.
Which target should be used?
The return below which the investor is genuinely worse off: the liability discount rate for a pension scheme, the cash rate for capital that could sit in bills, zero for capital whose first job is not to lose money. Whatever the choice, it belongs next to the figure.
Does Sortino fix Sharpe's problems for hedge funds?
Partly. It handles positive skew correctly, but it shares Sharpe's blind spots for smoothed or illiquid returns, which understate both volatility and downside deviation, and for tail risk that has not yet shown up in the sample.
Stay informed
Market commentary, firm news and research from EC Assets - direct to your inbox.