Trading concepts

Calmar Ratio: Definition, How to Calculate It, and Why the Window Changes the Answer

The definition used by Morningstar and the PerformanceAnalytics package, the calculation done step by step on 36 months of the S&P 500 ETF, and how the same three years give a ratio of 3.00 on month-end closes and 1.21 on daily closes.

· Founder of PatternsWizard
Checked 7 min read
Why month-end closes understate the drawdown: a daily price path over about three months, with dots on the month-end closes joined by a dashed line. The path peaks, falls to a low in the middle of the second month and recovers before that month ends. The drawdown measured on daily closes, from the peak to the mid-month low, is much deeper than the drawdown measured on month-end closes.
On this page
  1. Definition
  2. A worked example on the S&P 500 ETF
  3. How much the window matters
  4. Calmar, Sterling and Sharpe
  5. Frequently asked questions

The Calmar ratio divides an investment's annual return by its worst loss from a peak, its maximum drawdown, over the same period. A ratio of 2 means the investment earned, each year on average, twice the deepest fall it suffered. It is most commonly used with hedge funds, and it works for any fund, strategy or index with a return history. This page gives the definition, the calculation step by step on the S&P 500 ETF, and how much the result depends on the window and the data you use.

Definition

Morningstar's documentation of its custom calculations describes the Calmar ratio as a variation on the Sterling ratio, used mostly with hedge funds: the annualized return divided by the maximum drawdown over the same time period. The documentation of the PerformanceAnalytics package for R gives the same ratio, annualized return over the absolute value of the maximum drawdown, and notes that it is traditional to compute it on a three-year return series.

Where the name comes from: Wikipedia's entry credits Terry W. Young, owner of a firm called California Managed Accounts, and cites an article of his in Futures magazine dated October 1, 1991; it reads the name as short for California Managed Accounts Reports, the firm and its newsletter. In the definition Wikipedia quotes from that article, the ratio uses the last 36 months and is calculated on a monthly basis. We could not obtain that article, so this page does not quote it and relies on the definitions above.

The three years are a convention, not a rule: PerformanceAnalytics computes the ratio on whatever return series it is given, without checking its length. When you read a Calmar ratio, check the window and the data frequency before comparing it with another.

A worked example on the S&P 500 ETF

Take the monthly closes of the S&P 500 ETF (dividend-adjusted) for the 36 months ending September 2026.

  1. Start and end. The close was 412.26 at the end of September 2023 and 762.63 at the end of September 2026.
  2. Annual return. (762.63 / 412.26) to the power 1/3, minus 1, is 22.76% a year, compounded.
  3. Maximum drawdown. The deepest fall from a month-end high to a later month-end close was from 590.22 in January 2025 to 545.49 in April 2025: 7.58%.
  4. Calmar ratio. 22.76% / 7.58% = 3.00.

The four month-end closes behind a Calmar ratio of 3.00, S&P 500 ETF

The four month-end closes behind a Calmar ratio of 3.00, S&P 500 ETF. Start of the window, September 2023: 412.26; Highest close before the drawdown, January 2025: 590.22; Lowest close after it, April 2025: 545.49, −7.58%; End of the window, September 2026: 762.63. Our computation; the inputs and the method are in the text.
Our computation; the inputs and the method are in the text.Month-end closes, dividend-adjusted. The start and the end give the 22.76% annual return; the fall from January to April 2025 is the 7.58% maximum drawdown.

Run the same window on daily closes, and the drawdown becomes 18.76%, from February 19 to April 8, 2025: the April low fell in the middle of the month, and month-end closes miss it. The ratio drops to 22.76% / 18.76% = 1.21. Same ETF, same three years, a ratio less than half as large. A monthly calculation smooths the drawdown; two Calmar ratios are comparable only when both use the same data frequency.

Why month-end closes understate the drawdown

Why month-end closes understate the drawdown: a daily price path over about three months, with dots on the month-end closes joined by a dashed line. The path peaks, falls to a low in the middle of the second month and recovers before that month ends. The drawdown measured on daily closes, from the peak to the mid-month low, is much deeper than the drawdown measured on month-end closes.
Schematic, not market data. Drawn by hand to show the effect, not computed.

Same 36 months, two data frequencies: the Calmar ratio of the S&P 500 ETF

Same 36 months, two data frequencies: the Calmar ratio of the S&P 500 ETF. 36 months to September 2024: Month-end closes 0.49, Daily closes 0.48; 36 months to September 2025: Month-end closes 2.98, Daily closes 1.32; 36 months to September 2026: Month-end closes 3.00, Daily closes 1.21. Our computation; the inputs and the method are in the text.
  • Month-end closes
  • Daily closes
Our computation; the inputs and the method are in the text.Same annualized return in both series (from the month-end closes that start and end the window); only the drawdown changes: 23.93% against 24.50%, 8.33% against 18.76%, and 7.58% against 18.76%. Dividend-adjusted prices.

The gap depends on where the low falls. In the window ending September 2024, the deepest daily close, October 12, 2022, came a few sessions after the September 2022 month-end, and both drawdowns are about 24%: the ratio is 0.49 on month-end closes and 0.48 on daily closes. When the low falls mid-month, as in April 2025, the month-end ratio is more than twice the daily one.

How much the window matters

36 months endingAnnual returnMax drawdownCalmar
Feb 2009−15.08%50.78%−0.30
Dec 201113.99%18.07%0.77
Dec 202014.01%19.45%0.72
Dec 20227.62%23.93%0.32
Sep 202411.80%23.93%0.49
Sep 202524.79%8.33%2.98
Sep 202622.76%7.58%3.00

Month-end closes of the S&P 500 ETF, dividend-adjusted. Across all 368 windows of 36 months ending between February 1996 and September 2026, the ratio ranged from −0.36 (ending March 2003) to 6.75 (ending July 1997), with a median of 0.99; it was negative in 16.3% of them. The same index, with no change in strategy, moved from one end of that range to the other depending only on whether a bear market fell inside the window.

Calmar ratio of the S&P 500 ETF over seven 36-month windows

Calmar ratio of the S&P 500 ETF over seven 36-month windows. 36 months to Feb 2009: −0.30; 36 months to Dec 2011: +0.77; 36 months to Dec 2020: +0.72; 36 months to Dec 2022: +0.32; 36 months to Sep 2024: +0.49; 36 months to Sep 2025: +2.98; 36 months to Sep 2026: +3.00. Our computation; the inputs and the method are in the text.
Our computation; the inputs and the method are in the text.Month-end closes, dividend-adjusted, the windows of the table. Over all 368 windows ending February 1996 to September 2026, the median was 0.99.

The two windows ending in December 2022 and September 2024 share the same drawdown, from December 2021 to September 2022. Once that fall left the 36-month window, the ratio jumped from 0.49 to 2.98 in a year.

Calmar, Sterling and Sharpe

  • Sterling ratio. The ratio the Calmar ratio varies on. In PerformanceAnalytics, it adds an excess risk measure to the maximum drawdown, 10% by default.
  • MAR ratio. Compound annual return divided by the maximum drawdown since inception, not over 36 months; Wikipedia traces it to the newsletter Managed Account Reports and notes that the two are sometimes taken for the same ratio.
  • Sharpe ratio. It measures risk with the volatility of returns instead of the worst fall.

Drawdown is also the input of the risk of ruin, the chance that losses take an account below the point where it can recover. The trading performance page covers the other metrics used to judge a track record. More concepts are in the trading concepts section.

Every pattern's results on one sheet

The only pattern cheat sheet with tested stats: 152 patterns, what the textbook says and what really happened on 6 markets. Free PDF, sent to your inbox.

The PDF by email, plus our newsletter: new tests and results that change. Unsubscribe anytime. Privacy.

Frequently asked questions

What is the Calmar ratio?

A performance measure that divides the annualized return of an investment by its maximum drawdown over the same period. Three years of monthly data is the traditional window, but tools let you choose another.

Is a higher Calmar ratio better?

Yes: a higher ratio means more annual return for each point of the worst fall. Compare ratios only on the same window and data frequency, since on the S&P 500 ETF the same three years give 3.00 on month-end closes and 1.21 on daily closes.

What is a good Calmar ratio?

There is no official threshold. For reference, the S&P 500 ETF's 36-month ratio on month-end closes had a median of 0.99 over the 368 windows ending between February 1996 and September 2026, and was above 1 in 49.2% of them.

What is a good maximum drawdown?

It depends on the market and on what you can stand. The S&P 500 ETF's maximum drawdown over 36 months on month-end closes ranged from 7.58% (ending September 2026) to 50.78% (ending February 2009) in the windows on this page.

What is the difference between the Calmar ratio and the Sharpe ratio?

The Calmar ratio measures risk with the deepest fall; the Sharpe ratio uses the volatility of returns. One large loss weighs on the Calmar ratio in full, while its effect on volatility is diluted among all the other returns.

Who created the Calmar ratio?

Wikipedia credits Terry W. Young, owner of a firm called California Managed Accounts, and cites a 1991 article of his in Futures magazine; it reads the name as short for California Managed Accounts Reports. We have not read that article.

Sources

  1. Morningstar, Custom Calculation Data Points (October 2016)
  2. PerformanceAnalytics for R, CalmarRatio documentation (source file)
, Founder of PatternsWizard

Passionate about finance, quantitative trading and data analysis. Amaury built the PatternsWizard test bench, which detects each pattern on full price history and measures what happened next against random entries.

PatternsWizard tests chart and candlestick patterns on full market history, against random entries. How we test patterns.

179 patterns testedRandom-walk self-test passed

Related guides

All trading concepts articles