Metrics

What is Calmar Ratio?

Reviewed by Karthikeyan Anbarasan (Founder, TradeLyser) · Updated 19 Oct 2026 · Editorial policy

DefinitionThe Calmar ratio divides annualised return by maximum drawdown, conventionally measured over the last 36 months. If ₹10,00,000 grew to ₹15,00,000 over three years, about 14.5% a year, and the worst peak-to-trough fall was 18%, Calmar is 14.5 ÷ 18 = 0.80.
Formula
Calmar ratio = Annualised return % ÷ Max drawdown % (usually over the trailing 36 months)
Example
(₹15,00,000 ÷ ₹10,00,000)^(1/3) − 1 = 14.5% a year; 14.5 ÷ 18 = 0.80
0%−6%−12%Max drawdown −11%
Figure 1. Illustrative underwater chart with a deepest drawdown of −11%. If the same 36-month window returned 22% a year, Calmar would be 22 ÷ 11 = 2.0.

Why the 36-month window

The ratio comes from the managed futures industry and is conventionally measured over the trailing 36 months, updated monthly. A shorter window may not contain the kind of drawdown a longer record would, and annualising a few strong months exaggerates the return.

Some tools use the full history instead. The closely related MAR ratio uses compound annual return and max drawdown since the start of the record. Check which window a figure uses before comparing two of them.

Working it out

  1. Take the equity curve for the window, ideally 36 months, using net P&L after charges.
  2. Annualised return = (Ending ÷ Starting)^(1 ÷ years) − 1. For ₹10,00,000 to ₹15,00,000 over 3 years, that is 14.5%.
  3. Max drawdown % = the largest (peak − trough) ÷ peak in the window: 18% in this example.
  4. Divide: 14.5 ÷ 18 = 0.80.
Common mistake
Deposits and withdrawals change the equity curve without any trading. Use returns that strip them out, or both the annualised return and the drawdown will be wrong.

Worked example: a ₹20,00,000 F&O book

Illustrative. A positional F&O strategy runs on ₹20,00,000 of capital set aside for it, with no deposits or withdrawals, and ends 36 months later at ₹28,00,000 after charges. Its highest point before the worst fall was ₹24,50,000 and the lowest point after it was ₹20,58,000.

Annualised return: (28 ÷ 20)^(1/3) − 1
11.9%
Max drawdown: ₹3,92,000 ÷ ₹24,50,000
16.0%
Calmar ratio: 11.9 ÷ 16.0
0.74

A second strategy with the same 11.9% a year but an 8% max drawdown would have a Calmar of about 1.5: the same return for half the worst fall. At 0.74, a repeat of the 16% drawdown would take roughly 16 months to earn back at the average pace, ignoring compounding.

Calmar next to Sharpe and Sortino

Sharpe

Return per unit of all volatility, up and down.
Counts as risk: every swing

Sortino

Return per unit of downside deviation.
Counts as risk: falls below a target

Calmar

Return per unit of the single worst fall.
Counts as risk: the deepest drawdown

Because Calmar rests on one event, a single bad month can move it a long way, and a short record can look good simply because its worst drawdown has not happened yet. Read it alongside the length of the record.

MeasureReturn usedDrawdown usedUsual window
Calmar ratioAnnualised return (%)Max drawdown (%)Trailing 36 months
MAR ratioCompound annual return since the start (%)Max drawdown since the start (%)Full record
Recovery factorNet profit (₹)Max drawdown (₹)Any stated period, not annualised

Using it in your monthly review

  1. Keep a month-end equity figure for each strategy, after charges, with deposits and withdrawals noted separately.
  2. Each month, recompute annualised return and max drawdown over the trailing 36 months, or the full record if it is shorter, and write the window next to the ratio.
  3. Log the date and cause of the max drawdown. If it came from a rule break or a size increase, note it; the ratio cannot tell those apart from normal losses.
  4. Watch the direction over several months rather than one reading. A falling Calmar with a steady return means drawdowns are getting deeper.

Tag strategies consistently in the journal, by setup and by instrument, so that the Calmar for a Bank Nifty option-selling strategy is not mixed with trades from a different approach. A ratio for the whole account blends strategies whose drawdowns may come at different times.

For traders in India
Measure returns on the capital set aside for a strategy, not on margin blocked, and after brokerage, STT and exchange charges. F&O income is taxed at slab rates on the year’s total, so Calmar is usually worked out before income tax.

Reference guide

ContextValueReading
Calmar of 1.0A year’s return equals the worst fallA repeat of that drawdown would cost about a full year of gains
Below 1.0Worst fall bigger than a year’s returnClimbing back from a repeat would take more than a year at the same pace
Computed on under a year of dataNot comparableAnnualising a few good months inflates the return; the 36-month convention exists for this reason

Questions

What is a good Calmar ratio?

Above 1 means the annual return exceeded the worst drawdown in the window. There is no fixed standard, so compare strategies over the same window and the same data.

How is the Calmar ratio calculated?

Divide the annualised return by the maximum drawdown, both in percent, conventionally over the last 36 months. 14.5% a year with an 18% max drawdown gives 0.80.

What is the difference between the Calmar and Sharpe ratios?

Sharpe divides excess return by the standard deviation of all returns. Calmar divides annualised return by the single deepest drawdown, so it focuses on the worst loss rather than everyday swings.

Can I calculate Calmar for less than three years?

Yes, and many tools do, but state the window you used. Annualising a short record can make the return look larger than it is likely to be over time.

What is the MAR ratio?

A close relative of Calmar: compound annual return divided by max drawdown, both measured from the start of the record rather than over the last 36 months. On a three-year record the two can match; on longer records they usually differ.

Is a higher Calmar ratio always better?

Only when the windows and data match. A high Calmar on a short or quiet record can simply mean the deepest drawdown has not happened yet, so check the record length and the size of the drawdown itself.

Related terms