Sortino Ratio

What is the Sortino Ratio?

The Sharpe ratio divides excess return by the standard deviation of all returns, treating upside surprises as if they were just as undesirable as downside ones. That treatment makes mathematical sense (variance is symmetric around the mean) but offends intuition (no investor regrets a 20 percent month). Frank Sortino's contribution in the 1980s and 1990s was to rebuild the ratio around a single change: replace total volatility in the denominator with downside deviation, which measures dispersion only below a target return. Strategies with asymmetric distributions (more upside than downside in size) get a higher Sortino than Sharpe; strategies with the opposite shape get a lower one.

The threshold is called the Minimum Acceptable Return, or MAR. It is the line below which deviations are counted as risk. Common choices are zero (any loss is risk), the risk-free rate (anything below cash is risk), or an investor-specific hurdle (anything below a 5 percent goal is risk). The choice meaningfully changes the result. Sortino computed against MAR equal to zero is the most common default in backtest reporting, because it isolates the downside-vs-upside shape from any debate about the right hurdle.

As rough calibration, a Sortino ratio above 1 is generally considered good, above 2 very good, and above 3 unusual for an equity strategy across a multi-year window. Long-only equity strategies typically post Sortino ratios from about 0.6 to 1.5 in normal windows. Carry, mean-reversion, and short-volatility strategies often produce higher Sortino than Sharpe (small frequent gains, occasional larger losses), while trend-following strategies tend to show similar Sortino and Sharpe because their winning months are not asymmetrically larger than their losing ones on a per-month basis.

Formula

The denominator looks like a one-sided standard deviation but with one critical detail: the sum of squared shortfalls is divided by the total number of observations N, not by the count of below-MAR observations alone. This convention follows Sortino's original methodology and is the version used in CFA Institute curriculum material. Dividing by N preserves the interpretation of the denominator as a downside-only second moment of the full return distribution; dividing only by below-MAR observations would inflate the figure during calm periods with few losses, which is the opposite of what the ratio should do.

The denominator is then annualized like any other volatility figure, typically by multiplying by √12 for monthly returns or √252 for daily returns. The numerator is the annualized excess return over the same MAR. When MAR equals the risk-free rate, the Sortino numerator matches the Sharpe numerator exactly, which makes the side-by-side comparison interpretable: any difference between the two ratios comes from the denominator only.

Why the Sortino Ratio Matters in Backtesting

Sortino exists because Sharpe answers a slightly wrong question. A strategy with monthly returns of plus 5, plus 5, plus 5, plus 5, minus 1 has lower Sharpe than a strategy with returns of plus 2, plus 2, plus 2, plus 2, minus 0.5, simply because the first has higher total volatility. But for an investor, the first strategy is plainly better: it compounds faster and the upside variability is a feature, not a bug. Sortino captures that intuition. Sharpe does not.

The practical use of Sortino is comparison across strategies with differently shaped distributions. Two strategies with similar Sharpe ratios but very different Sortino ratios are telling you that one of them has more asymmetry than the other. A Sortino notably higher than the Sharpe (sometimes 1.5x, sometimes 2x) signals upside-skewed variability, which is a desirable shape. A Sortino roughly equal to the Sharpe signals symmetric variability around the mean, which Sharpe is already pricing correctly.

Sortino is also useful for evaluating option-selling, carry, and mean-reversion strategies, where return distributions naturally tilt toward many small gains and occasional larger losses. Sharpe undercounts the steady-gain component by treating it as risk; Sortino corrects that. The flip side is that Sortino does not catch tail risk, since rare extreme losses get absorbed into the downside deviation without special weighting. For tail-heavy strategies, Sortino can look attractive right up until the bad month arrives. Reading Sortino alongside Max Drawdown is the standard pairing for that reason.

How SledgeKey Implements the Sortino Ratio

Sortino appears on the backtest results page as a single annualized number alongside Sharpe, computed on the same monthly returns the rest of the results use. The Minimum Acceptable Return is set to the risk-free rate for the test window, sourced from live Treasury data so the figure reflects the actual rate environment during the test rather than a hardcoded baseline. The denominator divides the sum of squared shortfalls by N (the total number of monthly observations) following Sortino's original convention, then annualizes by multiplying by the square root of 12.

The benchmark's Sortino is computed the same way on the benchmark's monthly returns against the same MAR over the identical window. Reading the two side by side is the most useful framing: a strategy with a Sortino notably higher than the benchmark's, even when Sharpes are comparable, is one with more upside-skewed variability than the index. A strategy with a lower Sortino than the benchmark's despite similar Sharpe is taking more downside-skewed risk in exchange for the same volatility-adjusted return.

Common Pitfalls

The first pitfall is interpreting Sortino as a full risk-adjusted measure. Sortino corrects Sharpe's symmetric treatment of variance, but it does not correct Sharpe's blind spot to tail risk. A strategy can post a high Sortino across a calm decade by never producing a large downside month, then produce one terrible month that erases years of cumulative return. Sortino looked great right up to that month and looked terrible immediately after. Sortino is an improvement over Sharpe, not a substitute for examining drawdowns and tail months.

The second pitfall is comparing Sortino ratios computed with different MAR thresholds. A Sortino against MAR equal to zero will generally differ from a Sortino against MAR equal to the risk-free rate or MAR equal to 5 percent. The gap matters especially in higher-rate environments, where the choice of MAR can shift the ratio by 0.3 or more. Whenever two Sortino figures are compared, confirm they share the same MAR convention before drawing conclusions.

The third pitfall is small-sample noise in the downside-deviation estimate. Sortino's denominator is driven by the subset of returns below MAR. In a calm three-year window of monthly data, that subset might contain only six or eight observations, and a few outlier months drive the figure unstably. Sortino computed on short windows is wobblier than Sharpe computed on the same window, simply because the effective input set is smaller. For honest comparisons, use windows long enough to provide at least 24 to 36 below-MAR observations, which usually means at least ten years of monthly returns for typical equity strategies.

Related Concepts