A positive return can still fall below a positive target. Sortino describes those target shortfalls rather than variability on both sides of a mean. Its value depends on the chosen target, observation frequency and downside-deviation convention. This article keeps a hypothetical monthly example and explains what its result can and cannot show.

What the Sortino ratio is and how it differs from Sharpe

Sortino divides mean return above a specified target by target downside deviation. In the discrete convention used here, downside deviation is √{Σ[min(0, Rt−T)]² / N}, using all N periods. It is not the ordinary standard deviation of only below-target returns around their own mean. State the target, frequency, sample and denominator convention.

Sharpe uses variability of differential returns relative to a benchmark; Sortino uses shortfalls against a stated target. Their numerators need not match unless benchmark and target match. Neither describes the full risk profile. Compare compatible periods and data, then inspect exposure and drawdown separately.

Monthly return does not show the path

A fictional account has 100 units at the start and end of a month, but falls to 90 in between. Monthly return is 0%, while the intramonth decline is 10%. With monthly target 0%, this month contributes zero to squared shortfalls. It does not mean there was no loss anywhere along the path.

Scroll sideways to see the full diagram

Equal month endpoints, a decline betweenStart 100, intramonth 90, end 100. Start-to-end return is 0%, but value fell 10% along the way.Account value [units]10090StartBetweenEndMonthly report: 100 → 100 = 0%Path: 100 → 90 = −10%Recovery: 90 → 100 needs +11.11%
The dashed line joins reporting dates; gold shows the fictional intervening path.

Changing frequency does not add missing data to an old report. To see an intervening decline, you need observations from that interval. Store frequency beside the ratio and inspect equity and drawdown separately. This example demonstrates aggregation, not a new Sortino value.

Assumptions: One arbitrary monthly observation, target 0%, constant currency and no deposits or withdrawals. The invented 100→90→100 path is not a representative strategy sample. For the single monthly return of 0%, downside deviation is 0 and Sortino 0/0 is undefined. Formula or data reference: CFA Institute — The Sortino Ratio (2012).

The formula and the calculation procedure

The classical formula reads as follows:

Sortino Ratio = (Rp − T) / DD

The numerator and denominator must use compatible periods and units. For this monthly calculation, the numerator is mean monthly return minus the monthly target. The denominator is the square root of the mean squared shortfall across all monthly observations, including zero contributions from periods at or above target.

For a discrete monthly illustration, list returns and specify a monthly target, for example 0%. Square each negative shortfall from that target, retain zero for other periods, divide by all observations and take the square root. Divide mean monthly excess return by this monthly downside deviation. Do not automatically multiply by √12 to claim an annual Sortino: aggregation, serial dependence and target conversion can alter downside risk. The cited CFA discussion explicitly cautions against this shortcut.

How to compute the ratio step by step

Hypothetical net monthly returns for an invented system are 3%, 5%, −2%, 6%, 10%, −1%, 2%, 8%, −3%, 4%, 12% and 1%. Their sum is 45 percentage points, so the arithmetic mean is 3.75% per month. The sum is not the compounded return on an investment. Set the monthly target to zero for this calculation.

With target 0%, the squared monthly shortfalls are 4, 1 and 9 percentage-points squared, totaling 14. Dividing by all 12 periods gives 14/12, and the square root is about 1.0801 percentage points. The monthly ratio is 3.75 / 1.0801 ≈ 3.4718. This constructed short series does not establish skill, and multiplying by √12 does not automatically produce a comparable annual downside-risk measure.

Reference points and how to read the result

Interpret the ratio through its inputs. The following ranges are descriptive examples, not verified industry grades. No minimum of 36 months makes a result automatically credible, and a very high number can reflect a small downside denominator.

Reading a Sortino value under one stated convention
Below 1.0Excess return is less than downside deviation under the stated convention; this does not measure maximum drawdown.
1.0 to 2.0A numerical range for comparison only; no verified industry average or universal acceptance threshold is established here.
2.0 to 3.0Compare only with ratios calculated at the same frequency, target, window and downside-deviation convention.
Above 3.0Inspect observation count, downside observations and calculation conventions; a high value does not establish institutional quality.
Reading caveatA small downside denominator can create a very large ratio; inspect its inputs and sample limits

State the return frequency, target and downside-deviation convention when comparing ratios. Downside deviation is measured from a target, whereas standard deviation is measured around the sample mean; one denominator need not always be smaller. The numerical ordering of Sortino and Sharpe cannot be assumed.

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 target changes both parts of the ratio. Use a target expressed in the same periods as the return series. Zero, a stated fixed monthly hurdle and a matching benchmark answer different questions. A policy rate is not automatically an investable return, and an annual target must not silently be treated as a monthly one.

A short or unrepresentative sample can understate downside variation, especially if few returns fall below target. More observations do not by themselves remove selection, dependence or missing-regime problems. There is no universal thirty-six-month threshold that guarantees a credible Sortino estimate.

Sortino does not cap future loss. The discrete historical formula does not require a normal distribution, but its sample can omit extreme outcomes and understate downside exposure. Review tail scenarios, liquidity and drawdown separately; neither a high Sortino nor another historical statistic guarantees protection.

Sortino alongside other risk metrics

Use several views of results rather than a single score. Sharpe ratio describes variability of differential returns; the Sortino versus Sharpe head-to-head explains the comparison with target shortfalls. Calmar ratio relates a stated annualised return convention to maximum drawdown over its specified window. maximum drawdown describes the peak-to-trough decline itself, and expectancy formula describes average payoff per trade. These measures are not interchangeable; the risk management section covers exposure procedures.

What to do tomorrow

  1. Collect a consistent return series. Define complete calendar periods and adjust for deposits and withdrawals. State the observation count and downside observations; two years is a chosen window, not a threshold at which the ratio stops being noisy.
  2. Compare two explicitly stated targets. Calculate the ratio at T=0 and at a hypothetical T=0.33% per month. The latter is an assumption for this exercise, not today's risk-free rate; use a frequency-consistent target for any actual comparison.
  3. Record the denominator as well as the ratio. Keep the target, mean return, downside deviation, shortfall count and window in the report. When a value changes, inspect those inputs rather than assuming that a larger gap from Sharpe proves a particular return shape.
  4. Define review actions before results arrive. Record how a breach of your chosen loss or exposure limit affects new orders and existing positions. A Sortino score below 1 for two quarters is not a universal reason to halt or resume trading; use financial capacity and the complete risk picture.