Sortino Ratio — Sharpe Counting Only the Downside

Picture a trend-following system that produces the occasional fat month of fifteen percent alongside a handful of barely scratched losing months of one or two percent. The Sharpe ratio, which makes no distinction between upward and downward swings, literally penalises you for those big winners. The Sortino ratio, developed by Frank Sortino at the turn of the 1980s and 1990s, fixes that flaw by looking only at downside deviations, since those are the moves that genuinely keep traders up at night.

What the Sortino ratio is and how it differs from Sharpe

The Sortino ratio measures the excess return above a chosen target (most often the risk-free rate, sometimes plain zero) divided by the downside deviation, meaning the standard deviation computed only from periods in which the result fell below that target. Frank Sortino published the concept jointly with Robert van der Meer in Downside Risk in the Journal of Portfolio Management in 1991, and developed it further a decade later in the book Managing Downside Risk in Financial Markets co-authored with Stephen Satchell.

The difference with Sharpe boils down to a single question. Sharpe treats every swing identically, whether it is a pleasant jump upward or a painful slide downward. Sortino notes that for someone managing capital, only drawdowns hurt, while jumps upward are the reason we accept risk in the first place. For symmetric strategies, such as a typical mean-reversion system where wins and losses are roughly the same size, both ratios paint a similar picture. For asymmetric strategies, like classic trend-followers, the difference can be dramatic, and only Sortino reflects the true risk profile.

The formula and the calculation procedure

The classical formula reads as follows:

Sortino Ratio = (R_p − T) / DD

The numerator holds the difference between the portfolio's average annual return and the chosen target, which we denote as T. The denominator holds the downside deviation, which is the square root of the average squared shortfall in periods where the result fell below the target. Periods with a positive result are treated as zero in the denominator calculation, and that is precisely the core distinction from the classical standard deviation.

The spreadsheet procedure runs as follows. List monthly returns in one column, covering at least twenty-four months. Set a target value — for most purposes the monthly equivalent of the risk-free rate works well, and at the start of 2026 that lands around four percent a year, or about 0.33 percent a month. For each month below target, square the shortfall; months above target you set to zero. Then divide the sum of squares by the total number of observations, not just the negative ones — a critical detail many people get wrong. The square root of that quotient gives the downside deviation, and dividing the average excess return by it gives the monthly ratio. To annualise, multiply by the square root of twelve.

Worked example — step by step

An illustrative example. Suppose a trend-following system on EUR/USD with twelve monthly results: three percent, five percent, minus two percent, six percent, ten percent, minus one percent, two percent, eight percent, minus three percent, four percent, twelve percent and one percent. The average monthly return here is roughly four and a half percent.

The three negative periods are minus two, minus one and minus three percent. With a target of zero, the squared shortfalls come out to four, one and nine, which sum to fourteen. We divide by twelve months (not by three — a common mistake) and obtain 1.17. The square root gives a downside deviation of roughly 1.08 percent a month. The average excess above target is 4.58 percent, so the monthly ratio is about 4.24, and after annualising by the square root of twelve we end up around fourteen — a figure that strongly suggests either an exceptionally good strategy or, more plausibly, that our sample of twelve months is too short and happened to cover a lucky stretch.

Reference points and how to read the result

Across both academic literature and institutional practice, the following bands are used for annualised Sortino values computed from at least three years of data:

Bear in mind that Sortino benchmarks tend to be higher than the matching Sharpe benchmarks, because the denominator is smaller. Comparing the two numbers directly without keeping this distinction in mind leads straight to faulty conclusions.

Honest warnings and limitations

Sortino, like any statistic, comes with a few traps that are worth knowing before letting the number drive real-money decisions.

The choice of target T changes the result. A ratio computed against zero looks different from one computed against the annual risk-free rate, and different again from one computed against the return on a benchmark index. When reporting, always state the assumed target, because without that information the number is not comparable to anyone else's.

A small sample hides asymmetry. With only twelve months of data it is easy to land in a situation where the strategy enjoyed a lucky run without deeper drawdowns, and the downside deviation came out artificially low. The longer the data series, the more credible the result. The institutional standard is a minimum of thirty-six months.

Sortino does not protect against the black swan. The ratio is built from historical data and, just like Sharpe, implicitly assumes roughly normal tails. A single extreme event — the abrupt unpegging of the Swiss franc in January 2015 or the turbulence of March 2020 — can wipe out an account regardless of how flattering the Sortino number looked a week earlier. For evaluating fat-tail exposure, measures such as conditional value at risk (CVaR) or a straight reading of maximum drawdown serve far better.

"Investors define risk differently from statisticians. They fear the loss of capital, not the variability of its gains. Downside deviation answers that distinction in a way the classical standard deviation simply cannot." — Frank A. Sortino and Robert van der Meer, Downside Risk, Journal of Portfolio Management, 1991.

Sortino alongside other risk metrics

A single ratio never tells the full story of an account. Professional performance reports usually combine several mutually complementary measures. The classical Sharpe ratio gives a synthetic picture of two-way volatility and remains the comparison standard between funds. The closely related Sortino versus Sharpe head-to-head shows when each of these measures fits better. The Calmar ratio divides the annual return by the worst historical drawdown and is more empirical than statistical. The maximum drawdown itself is a raw number that every trader should know by heart from their own journal. In turn, the expectancy formula translates everything back into the language of an individual trade, which is often the most practical viewpoint early in the journey. For broader treatment of how risk measures sit inside a complete framework, see the section on risk management at ForexMechanics.

What to do tomorrow

1. Gather at least two years of your own monthly returns in a spreadsheet. Export trade history from your platform and split it into complete calendar months, because only at this sample size does Sortino stop being random noise and start carrying real information about the risk profile of your strategy.
2. Compute Sortino for two different targets T and compare the results. Run the calculation once with a target of zero and once with the monthly risk-free rate, roughly 0.33 percent a month, so you can see directly how sensitive the final number is to a parameter chosen somewhat arbitrarily.
3. Place Sortino next to two other measures in your monthly report. Add columns to your journal showing Sharpe, Sortino and Calmar on a rolling twelve-month window, because divergence between these three numbers is often the first sign that some material feature of the strategy has shifted and deserves attention.
4. Write a concrete intervention threshold into your risk plan. Specify that if rolling Sortino falls below 1.0 for two consecutive quarters, you halt live trading and conduct an out-of-sample review of the strategy, rather than telling yourself the dip was just bad luck and pressing on in the hope of mean reversion.