What Is the Calmar Ratio? Return Measured Against Maximum Drawdown

The Calmar ratio measures how much annualised return a strategy produced for each unit of its worst peak-to-trough decline over the same period. It is a pain-adjusted measure: instead of asking how bumpy the ride was, it asks how deep the deepest hole was, and whether the return earned justified sitting in it.

That framing is closer to how investors actually behave. Very few people abandon a strategy because its standard deviation rose. A great many abandon one after watching a third of their capital disappear and stay gone for two years.

What it measures

The formula in words:

Calmar ratio = annualised return over the period, divided by the maximum drawdown over the same period, taken as a positive number.

Two inputs, both worth care.

Annualised return is the compound rate of growth over the window, expressed per year. This is the same idea as CAGR: the constant yearly rate that would have carried the starting value to the ending value. Note that unlike Sharpe and Sortino, the standard Calmar formula does not subtract a risk-free rate. Some variants do; if a provider is subtracting one, they should say so.

Maximum drawdown is the largest fall from any peak in the portfolio’s value to the lowest point that followed, before a new peak was set. To compute it, you track the running high-water mark of the portfolio, measure the percentage decline from that mark at every point in time, and take the worst one. It is a single number describing the single worst episode, not an average of bad episodes. For the mechanics and the traps, see what is maximum drawdown.

Maximum drawdown is normally quoted as a negative percentage. The Calmar ratio uses its absolute value so the resulting ratio is positive when returns are positive.

How to read it

The intuition is unusually clean for a risk metric. A Calmar ratio of 1 means the strategy earned, per year, roughly what its worst drawdown cost at the trough. A ratio of 0.5 means the worst fall was about twice the annual return, so recovering from it consumed roughly two years of typical progress. A ratio of 2 means the annual return was about double the worst fall.

Some hypothetical arithmetic: a strategy that compounded at 18 percent a year over three years while suffering a worst peak-to-trough decline of 30 percent has a Calmar ratio of 18 divided by 30, or 0.6. A second strategy compounding at 12 percent with a worst fall of 12 percent has a ratio of 1.0. The second earned less but was far less punishing to hold, and the Calmar ratio ranks it higher. Whether that is the right ranking depends entirely on whose money it is and how long it can stay invested.

A few reading habits.

• Insist on the window. The original convention was 36 months, and three years remains the most common default. This is not cosmetic. Maximum drawdown can only grow as the window lengthens, so a Calmar ratio over ten years is almost always lower than one over three for the same strategy. Comparing across different windows is meaningless.
• Check whether the window contains a real market fall. A Calmar ratio computed across an uninterrupted rise divides by a small denominator and looks superb. It has not been tested.
• Look at the drawdown itself, not just the ratio. The same ratio can come from a modest return with a shallow fall or a large return with a deep one. Those are different products.
• Ask about recovery time. Depth is only half of drawdown pain. A 25 percent fall that recovered in four months and a 25 percent fall that stayed underwater for three years have the same denominator and very different lived experience. See drawdown recovery analysis.

Calmar answers a question the volatility ratios cannot: was the return large enough to justify the worst thing that happened, rather than the average thing.

Why depth deserves its own ratio

Volatility and drawdown measure genuinely different risks, and a portfolio can score well on one and badly on the other.

Consider a hypothetical strategy whose monthly returns are small and tightly clustered, giving it low standard deviation and a flattering Sharpe ratio, but which drifts persistently downward for eighteen months during a particular market regime before recovering. Standard deviation barely registers a long slow bleed, because no individual month is extreme. Maximum drawdown captures it completely, because it accumulates.

The reverse also happens. A strategy with several very large positive months and a couple of sharp but quickly reversed falls will have high standard deviation, a mediocre Sharpe ratio, and a perfectly respectable Calmar ratio.

Neither metric is wrong. They are measuring different things, and reporting only one is how a real risk gets hidden in plain sight.

What it does not tell you

It depends on one single observation. The denominator is a single historical episode: the worst one that happened to occur in the window. Every other loss in the sample is ignored entirely. That makes the ratio unstable. Extend the window by a month that happens to include a crash and the ratio can halve, without anything about the strategy changing.

It says nothing about frequency of losses. A strategy that lost money in eight of twelve months but never fell deeply can post a better Calmar ratio than one that rose steadily except for a single sharp episode. Frequency of disappointment is the domain of downside deviation, win rate and the Sortino ratio.

It ignores time underwater. As noted above, the ratio measures depth only. Two strategies with identical Calmar ratios can differ enormously in how long an investor spent below their previous high.

It is highly sensitive to the return frequency used. Maximum drawdown computed on monthly closing values is almost always shallower than the same drawdown computed on daily values, because the intra-month low is invisible. A provider using monthly data will report better Calmar ratios than one using daily data for exactly the same portfolio.

It does not adjust for market exposure. A strategy that simply held a rising market gets full credit for the market’s return and is charged only for the market’s fall. Separating market exposure from active decisions requires beta, the Treynor ratio or the information ratio.

It has no view on whether the worst is behind you. The maximum drawdown in a sample is the worst that did happen, which is not the worst that can happen. Treating a historical maximum as a bound is one of the most common and most expensive errors in reading risk metrics. Stress and scenario work exist precisely because history is a small sample. See stress testing a portfolio.

It inherits any flaw in the return series. If the underlying test used data that was not knowable at the time, the drawdown path is fictional. See why point-in-time data matters.

Using it in practice

Calmar earns its place in a report as the depth counterweight to the volatility ratios. The productive way to use it is comparative and contextual: same window, same data frequency, same cost treatment, read next to the raw maximum drawdown figure, the time-to-recovery figure, and at least one measure of loss frequency.

When Calmar and Sharpe disagree sharply, that disagreement is the finding. It usually means the strategy’s risk is concentrated in a shape one of the two metrics is structurally unable to see, and that is worth understanding before anything else in the report.

Related reading

• Portfolio metrics explained: the hub that maps every risk and return metric to the question it answers.
• What is maximum drawdown: the denominator on its own, and how it is computed.
• What is the Sharpe ratio: the volatility-based counterpart.
• Sharpe vs Sortino vs Calmar: which of the three answers which question.
• Drawdown recovery analysis: time underwater, and why depth alone understates the pain.

This article is educational. Altys Labs is not a registered research analyst or investment adviser, and nothing here is investment advice or a recommendation to buy, sell, or hold any security.