Sortino Ratio: Sharpe That Ignores Upside Swings

The Sortino ratio is a risk-adjusted return measure that only counts downside volatility, the moves that hurt an account, and ignores upside volatility entirely. It answers a narrower question than the Sharpe ratio: how much return did a strategy generate per unit of harmful risk, rather than per unit of total price movement. Because it drops the upside from the penalty side of the equation, a strategy with big winning months can post a much higher Sortino than Sharpe number, sometimes flattering results that a Sharpe ratio would rate as merely average.

What is the sortino ratio?

The Sortino ratio divides excess return by downside deviation. Excess return is the strategy's return above some minimum acceptable return, often zero or a chosen benchmark rate. Downside deviation is calculated only from periods where the return fell below that minimum, squaring those shortfalls, averaging them, and taking the square root.

The formula looks like this: Sortino = (return minus minimum acceptable return) divided by downside deviation. Every input above the minimum acceptable return is simply left out of the denominator. A month that returns plus 8% contributes nothing to the risk figure, even though it was a large, volatile swing.

How it differs from the Sharpe ratio

The Sharpe ratio uses standard deviation, which treats every deviation from the mean as risk, whether the account went up 5% or down 5%. A strategy that swings wildly to the upside gets punished by Sharpe exactly as hard as one that swings wildly to the downside, as long as the total variance is the same.

Sortino only measures the deviations below the minimum acceptable return. Two strategies can have identical Sharpe ratios and very different Sortino ratios, if one strategy's volatility comes mostly from strong up months and the other's comes from a mix of ups and downs. The math rewards the strategy whose bad months are smaller and rarer, regardless of how large its good months get.

A metric that only counts losses will always look kinder than one that counts every swing.

A worked example

Say two strategies each average a 1.5% monthly return with the same standard deviation of 4%, giving them identical Sharpe ratios of roughly 0.35 on a monthly basis. Strategy A's volatility is driven by a mix: some months at plus 6%, some at minus 5%. Strategy B's volatility comes mostly from occasional plus 10% months, with small, consistent losing months around minus 1.5%.

Standard deviation treats both the same, because it does not care about direction. Downside deviation does. Strategy B's losing months are small and tightly clustered, so its downside deviation is low, maybe 1.8%, compared to Strategy A's 4.2%. Strategy B's Sortino ratio comes out several times higher than Strategy A's, even though Sharpe called them equal.

Why it can flatter a strategy

This is where the ratio needs a careful reader. A strategy can post a strong Sortino number simply by having a few outsized winning months, even if the underlying edge is thin or the sample size is small. One large lucky month drags the metric upward without changing anything about how the strategy handles risk.

Sortino also depends heavily on the choice of minimum acceptable return and on how many losing periods exist in the sample. With a short track record and only a handful of down months, downside deviation is calculated from a tiny dataset. Swap out one bad month for a slightly better one and the ratio can move a lot, more than most readers expect from a metric with two decimal places.

When sortino ratio is useful, and its limits

Sortino earns its place when the return stream is asymmetric in a way that matters, meaning the strategy's upside actually looks different from its downside. Trend-following systems, option-selling strategies, and anything with a defined stop-loss but open-ended profit targets often fit this pattern, and Sortino describes them more honestly than Sharpe does.

It says less about tail risk than people assume. A strategy can have a fine Sortino ratio and still carry a small number of very large losing months buried in the sample, because downside deviation averages the shortfalls rather than flagging the worst one. Reading Sortino next to maximum drawdown and the size of the single worst month gives a fuller picture than either number alone.