Sharpe, Sortino and Calmar ratios explained

Two funds can post the same return and still not be equivalent at all. If one gets there with mild swings and the other with 40% drops along the way, the second one asked a lot more for the same result. Risk-adjusted metrics, Sharpe, Sortino, Calmar and Information Ratio, exist to put a number on that difference: how much return you got for each unit of risk you carried.

Why return alone isn't enough

CAGR tells you how it went, not what it cost. A portfolio that grows from $100,000 to $200,000 in ten years on an almost flat path, and one that reaches the same figure after a 50% crash halfway through, have the same final return. Whoever lived through the second path ran a much higher risk of selling at a loss, long before the average return had a chance to materialize. Risk-adjusted metrics normalize return for the risk taken, so two instruments become comparable even when their paths looked nothing alike.

Sharpe: return above the risk-free rate, divided by volatility

The Sharpe Ratio takes the portfolio's return, subtracts the risk-free rate, typically a short-term government bond, and divides the result by total annualized volatility. As a formula: (mean annual return − Rf) / annual volatility. The site calculates both on monthly returns over the period, as on ETF factsheets.

On this site the reading convention is: a value below zero is poor, between 0 and 1 acceptable, between 1 and 2 good, above 2 excellent. William Sharpe proposed it in 1966, calling it the “reward-to-variability ratio”. It's still the most widely used risk-adjusted performance indicator in the world, partly because it only needs return and volatility, data that's almost always available.

Its weakness sits in the denominator. Total volatility counts a 5% jump in a good month the same way it counts a 5% drop in a bad one. For an investor the first swing is a gift, not a risk, but Sharpe treats them as identical.

Sortino: only downward swings count

The Sortino Ratio grew out of that objection. Frank Sortino proposed replacing total volatility with just the deviation of returns below a chosen threshold, usually the same risk-free rate used in Sharpe. The formula has the same shape, (mean annual return − Rf) / downside deviation, only the denominator changes.

A fund that swings a lot on the upside and little on the downside will show a higher Sortino than its Sharpe, because the “good” volatility stops counting against it. The reading scale stays the same, below zero poor, above 2 excellent, but the number tells a different story: how much return you got for each unit of real loss risk, not generic swings. It's more informative than Sharpe precisely for funds with asymmetric returns, the ones that climb often and drop rarely but sharply.

Calmar: return against the worst drop

Sharpe and Sortino look at how returns are dispersed over time. Calmar looks at a single number, maximum drawdown, and answers a different question: how much return did you produce for each percentage point of maximum loss endured?

It's calculated by dividing CAGR by the absolute value of the maximum drawdown. A portfolio with a 6% CAGR and a 30% maximum crash has a Calmar of 0.20; one with the same CAGR but a drawdown contained at 15% reaches 0.40, double, despite returning exactly the same. On this site the reading convention is: below 0.2 is poor, between 0.2 and 0.5 acceptable, between 0.5 and 1 good, above 1 excellent.

The name comes from Terry Young, a California-based manager who published it in 1991 in Futures magazine: Calmar is an acronym for CALifornia Managed Accounts Reports, his firm's newsletter. It's particularly sensitive to the length of the period analyzed, because a longer horizon has a statistically higher chance of running into a large drawdown: comparing the Calmar of two funds only makes sense if it's calculated over the same period.

Information Ratio: is the outperformance systematic or just luck?

The first three metrics don't need a point of comparison. Information Ratio, on the other hand, only exists relative to a benchmark: it measures excess return over the benchmark, divided by tracking error, meaning how consistent that excess return was over time or concentrated in a few lucky months.

On this site the reading convention is: an IR below 0.25 is poor, between 0.25 and 0.5 good, above 0.5 excellent; sustaining it above 0.5 over time is considered rare in active management. It's the most useful metric for a specific question: does this manager beat the benchmark through skill, or through one lucky bet?

A numbers comparison

Take two funds, both with a 7% mean annual return and a risk-free rate of 2%. The Sharpe of both is (7% − 2%) / 8% = 0.63, because total volatility is 8% for each. Fund A swings symmetrically. If monthly returns are normal, with mean 7%/12 and volatility 8% divided by the square root of 12, annual downside deviation (the gap below the risk-free rate, then multiplied by the square root of 12, as in the site's calculation) is about 4.9%. Fund A's Sortino is (7% − 2%) / 4.9% ≈ 1.0. Fund B has downside deviation at 4%, by construction of the example: its Sortino is 1.25. On Sharpe the two funds match. On Sortino, Fund B sits about a fifth higher.

This is a constructed example meant to isolate the difference between the two metrics, not a comparison of real products. The numbers illustrate the mechanism, not a suggestion that a fund with low downside deviation is always the better choice.

Which one to check first

There's no fixed hierarchy, it depends on what you're evaluating:

a generic comparison between two diversified instruments, Sharpe is the most common starting point
a fund with asymmetric returns, many small gains and rare large drops, Sortino tells you more
surviving a major crash matters more than return dispersion, Calmar is the right metric
you're evaluating an active fund against its benchmark, Information Ratio tells you whether the outperformance is repeatable

None of these metrics should be read in isolation. A high Sharpe achieved during years that were exceptionally favorable for markets doesn't guarantee the same result over the next ten: they're snapshots of a historical period, not promises.

In BacktestFolio

The Risk-Adjusted Returns section, inside ETF/Fund Analysis, calculates Sharpe, Sortino, Calmar, Information Ratio and other benchmark-relative metrics for every instrument you compare, over the same period. Each row of the table also shows the reading scale, so you don't need to memorize it. Selecting a benchmark also unlocks Information Ratio, Alpha and the Capture Ratios; without a benchmark only the absolute metrics stay visible.

The section is included in Plus and Pro; the Analysis Demo shows it on sample instruments. On the Analysis page you load two or more instruments and set the risk-free rate in the initial parameters.

Frequently asked questions

Does a negative Sharpe mean the fund lost money?

No, it means the return was below the risk-free rate over the period analyzed, even if the fund stayed positive. With a risk-free rate at 3% and a fund that returned 2%, Sharpe is negative even though there was no loss at all.

Why can the same fund have a high Sharpe and a low Calmar?

Because they measure different things. A high Sharpe indicates steady returns with little return dispersion; a low Calmar indicates that, despite this, the fund went through at least one episode of deep decline. They're not in contradiction, they're looking at two different faces of risk.