Calmar Ratio Calculator

The Calmar ratio divides an annualised return by the worst peak-to-trough loss recorded over the same period. It answers how much annual return a strategy delivered for each unit of the deepest loss it put its holder through. A ratio of 1 means the annual return matched the worst loss in magnitude; above 1, the return was the larger of the two. The measure is commonly cited in managed-futures and hedge-fund performance reporting, where a single deep drawdown matters more to an investor than the average scatter of monthly returns.

A worked example

Suppose a trading strategy generated an annualised return of 18% over five years, and experienced a maximum drawdown of 12% during that period. Enter 18 for annualised return and 12 for maximum drawdown. The calculator returns 1.50. This indicates the strategy returned 1.50 units of gain for every unit of worst-case loss absorbed. By contrast, a strategy with 12% annualised return and the same 12% drawdown would return a ratio of 1.0, meaning return and maximum loss were equal in magnitude.

Which input moves the number more

On a proportional basis the two inputs are near-mirrors. Scaling the return up by 1% scales the ratio by 1.01. Scaling the drawdown up by 1% scales it by 1/1.01, a factor of 0.990, and scaling the drawdown down by 1% scales it by 1.010. The slight asymmetry is the reciprocal at work: dividing by a larger number costs a little less than dividing by a smaller one gains. Because these are scalings rather than additions, they hold whatever the sign of the return and at any drawdown: on a negative ratio the same factors move the figure further from or closer to zero, rather than up or down.

On a percentage-point basis, which is how the What-If cards move each input, the picture depends on where the ratio sits. Adding one point to the return changes the ratio by 1/d, whatever the return's sign. Taking one point off the drawdown changes it by r/(d(d−1)), and adding one point to it changes it by −r/(d(d+1)); both scale with the return, so it is the return's magnitude that settles which lever is larger. The two are equal when that magnitude sits one percentage point inside the drawdown for a reduction, or one point outside it for an increase. On the ratio scale that puts an equality band 1/d either side of 1.00 and, mirrored about zero, 1/d either side of −1.00: 0.96 to 1.04 and −1.04 to −0.96 at a 25% drawdown, but 0.80 to 1.20 and −1.20 to −0.80 at a 5% one. Between the two bands, where the ratio's magnitude is the smaller, the return is the larger lever; beyond them the drawdown is; inside either band the answer depends on which way the drawdown is moved. At the 15 and 25 defaults, one point of return is worth 0.040 and one point off the drawdown is worth 0.025, so the return dominates. The What-If cards round to two decimals, so that second figure renders there as 0.03.

The two point-basis expressions carry different domains. The reduction term, r/(d(d−1)), assumes a drawdown above one point: it is undefined at exactly 1 and changes sign below it. The increase term, −r/(d(d+1)), holds at every drawdown the input accepts, from the 0.1 floor to the 100 ceiling. The panel imposes a separate limit on the reduction: because the drawdown floors at 0.1, a full one-point cut is only available from a drawdown of 1.1 upward. Between 1.0 and 1.1 the What-If card relabels itself to the smaller move it actually applied, and below 1.0 the card is dropped rather than shown against a figure the panel cannot reach. The proportional relationships in the paragraph above are properties of the ratio rather than moves the panel offers, so they hold at any drawdown and any sign, though scaling the 25 default up by 1% gives 25.25, which sits off the 0.1 step grid and can only be entered by typing.

Reading the band

The result card carries an illustrative band alongside the number, describing where the ratio falls in plain terms. The five bands are mutually exclusive and read: Return was negative over the period, Return below half the drawdown, Return at least half the drawdown but below all of it, Return at least equal to the drawdown but below three times it, and Return at least three times the drawdown. Every boundary is lower-inclusive, so a ratio of exactly 0, 0.5, 1.0 or 3.0 falls into the higher of the two bands it divides; a flat return against a real drawdown therefore reads as Return below half the drawdown rather than as a negative one. They are descriptive shorthand for the arithmetic, not thresholds carrying any external standing.

Common scenarios where this metric matters

• Evaluating managed funds or trading strategies where you have historical return and drawdown data
• Comparing two strategies with different risk profiles: one high-return, high-drawdown versus one moderate on both
• Stress-testing a portfolio model under past market conditions to see how the ratio behaves
• Screening a list of strategies by band, since the two lowest bands cover strategies whose return was negative, and those whose return was zero or positive but came to less than half the worst decline

What this captures and what it does not

The Calmar ratio shows the relationship between total return earned and the largest peak-to-trough decline observed in the historical record. It does not account for:

• How often drawdowns occurred or their duration
• Volatility between the worst drawdown and average performance
• Fees, taxes, or transaction costs applied over the holding period
• Whether past drawdowns will repeat or how likely future ones may be
• Investor behaviour during steep declines, where actual outcomes often differ from modelled ones

One structural point is easy to miss. The numerator is a rate per year; the denominator is a single historical magnitude with no time dimension at all. The quotient therefore carries units of "per year", and two Calmar figures are only comparable when both were measured over the same window. A three-year figure and a ten-year figure are different quantities wearing the same name.

The output is for educational illustration and modelling purposes only, not a prediction of future results.

Using this alongside other measures

A high Calmar ratio in isolation can mask other risks. The Sharpe ratio divides excess return by total volatility, so it responds to the whole distribution of returns rather than to one deep decline. The Sortino ratio divides by downside deviation only, which sits between the two. Value at risk describes a loss threshold at a chosen confidence level rather than the worst observed outcome. Each divides return by a different definition of risk, and the Calmar figure is the one that keys on a single historical low point.