The Sortino ratio measures return above a chosen target per unit of downside deviation relative to that target. Unlike the Sharpe ratio, it does not treat upside variability and downside shortfall as the same type of risk.
The target is part of the definition
Let T be the target return or minimum acceptable return for one measurement period. A common form is Sortino = (mean(R) − T) ÷ DownsideDeviation(T).
The target might be zero, a periodic risk-free return or another stated hurdle. Changing T changes both which observations count as shortfalls and the ratio itself, so the target should always be reported.
Downside deviation measures shortfall relative to the target
One common sample convention calculates downside deviation as sqrt(mean(min(Rt − T, 0)2)) across all observations, with periods at or above the target contributing zero.
Other implementations may use a different denominator convention, such as averaging only below-target observations. Those values are not automatically comparable. The calculation method must be stated.
Sortino and Sharpe answer related but different questions
Sharpe Ratio uses the standard deviation of a differential-return series. Sortino focuses the denominator on downside shortfall relative to a target.
A strategy with large positive swings may therefore look more variable under Sharpe while those same upside observations do not increase downside deviation in the same way.
There is no universal good Sortino threshold
A larger positive Sortino ratio indicates more return above the stated target relative to the measured downside deviation in that sample. But a single cutoff cannot account for different targets, frequencies, calculation conventions, sample sizes and strategy structures.
Before comparing two ratios, make the inputs comparable.
Annualization must match the return convention
As with other volatility-based ratios, practitioners often scale periodic results to annual units. The appropriate method depends on how the numerator and downside deviation are constructed and on assumptions about the return process.
Disclose the source frequency and annualization method instead of treating an annualized Sortino as assumption-free.
A short sample can make downside deviation unstable
If only a small number of observations fall below the target, one additional bad period can materially change the denominator. A strategy with very few downside observations may therefore produce an impressive but fragile ratio.
Trade count, return-period count and the full downside distribution should be visible beside the summary statistic.
Sortino does not replace maximum drawdown
Downside deviation describes the distribution of shortfalls from a target. Maximum Drawdown describes the deepest observed peak-to-trough path loss. A strategy can have a particular Sortino ratio and still experience a drawdown that matters greatly to its capital plan.
The target should reflect the analytical question
Using zero asks how returns behaved relative to non-negative performance. Using a risk-free or required return asks a different question. Neither choice is universally correct in every application.
The important requirement is consistency: choose the target for a reason, state it, and do not change it simply to improve the reported ratio.
Use Sortino as one part of a risk-adjusted review
Read the ratio with return, maximum drawdown, loss distribution, sample period and the strategy's actual risk constraints. If the strategy is highly skewed or path-dependent, those complementary measures become especially important.
The Sortino ratio is useful because it makes the definition of undesirable return explicit. Its credibility depends on an equally explicit target, downside-deviation formula and sample.