Sharpe vs Sortino vs Calmar: Which Risk-Adjusted Ratio Answers Which Question?

Sharpe, Sortino and Calmar are three ways of asking the same question: how much return did this portfolio earn for the risk it took. They differ only in how they define risk. Sharpe uses total volatility, Sortino uses downside volatility alone, and Calmar uses the single worst peak-to-trough loss. That one substitution in the denominator is the whole story, and it is why the three ratios can rank the same set of portfolios differently.

Raw return on its own is never the whole answer, because two portfolios can arrive at the same end point by very different roads. Risk-adjusted ratios exist to describe the road. The trap is treating them as interchangeable scores rather than as three different questions.

What each ratio measures

All three share a similar shape: a return figure on top, a risk figure underneath. Here is each one in words.

Sharpe ratio. Take the portfolio’s average return over a period, subtract the risk-free rate (the return you could have earned without taking market risk), and divide by the standard deviation of the portfolio’s returns. Standard deviation measures how widely returns scatter around their average, in both directions. So Sharpe answers: how much excess return did I earn for each unit of overall variability?

Sortino ratio. Keep the same numerator, an excess return over a risk-free rate or over some minimum acceptable return, but change the denominator to downside deviation. Downside deviation is calculated the same way as standard deviation, except that periods above the threshold are treated as contributing no risk. Only the shortfalls count. So Sortino answers: how much excess return did I earn for each unit of bad variability?

Calmar ratio. Take an annualised return, usually a compound annual growth rate over a defined window such as three years, and divide it by the maximum drawdown over that same window, expressed as a positive number. Maximum drawdown is the largest fall from a peak in portfolio value to the lowest point before a new peak is made. So Calmar answers: how much annual return did I earn for each unit of worst-case loss I had to endure?

Ratio	Risk in the denominator	The question it answers
Sharpe	Standard deviation of all returns	Return per unit of total variability
Sortino	Downside deviation only	Return per unit of loss-side variability
Calmar	Maximum drawdown	Return per unit of worst peak-to-trough pain

How to read them

Read them as a set, not as a leaderboard. The most useful information is usually where they disagree. Suppose a portfolio grinds out small steady gains for years and then falls heavily once. Its standard deviation may still be modest, so Sharpe looks respectable, while the deep single fall crushes Calmar. That gap is the finding. It says the portfolio’s risk is concentrated in rare events rather than spread across ordinary months.

Now flip it. A portfolio with a handful of unusually large positive months will show a high standard deviation, which drags Sharpe down even though nothing bad happened. Sortino removes that penalty by ignoring the upside swings. If Sortino is far above Sharpe, the return distribution is lopsided to the upside. If Sortino is only slightly above Sharpe, the swings are fairly symmetric.

Compare like with like, or do not compare at all. These ratios are only comparable when the inputs match. That means the same measurement period, the same return frequency (daily, monthly or annual), the same annualisation convention, and the same risk-free rate assumption. A Sharpe computed on monthly data and annualised is not the same number as one computed on daily data, and a Sharpe measured over a calm two-year window is not comparable to one measured across a full cycle. If a report does not state these conventions, the ratio is decoration.

Do not compare across scales. Sharpe and Sortino are broadly on a similar scale because both denominators are volatility measures. Calmar is not. It is a return divided by a drawdown, so its typical range is different, and a Calmar of 1.0 does not mean the same thing as a Sharpe of 1.0. Compare Calmar to other Calmars only.

Mind the sample length. Sortino needs enough down periods to estimate downside deviation reliably. In a short, mostly rising sample there may be very few negative observations, and the ratio becomes unstable and flattering. Calmar has a related problem in reverse: it depends on one event, the worst drawdown in the window. If the window happens to exclude the last serious market fall, Calmar can look far better than the strategy deserves.

What they do not tell you

This is the part worth reading twice.

They do not tell you whether the past will repeat. Every one of these ratios is computed on history. A high figure describes what happened, not what will happen. Market regimes change, and a strategy that scored well in one environment can behave quite differently in another.

They do not know about costs unless you feed them in. If the return series is gross of brokerage, securities transaction tax, stamp duty, GST and the slippage between the modelled price and the traded price, then every ratio built on it is overstated. A high-turnover approach is affected far more than a low-turnover one. Check what the return stream includes before you trust the ratio.

They assume the return series is honest about risk. Illiquid or infrequently priced holdings produce smoothed return series, which understate volatility and drawdown, which in turn inflate all three ratios. The same distortion arises when returns are marked less often than the underlying assets actually move.

They say nothing about the shape of the tail. Standard deviation treats a return distribution as if extreme events were rare in a well-behaved way. Real markets deliver fatter tails than that. Sortino is closer to the intuitive idea of risk, but it still summarises the entire loss side into one number, so it cannot tell you whether the losses came as many small dents or one crater.

They do not describe how long the pain lasted. Calmar tells you the depth of the worst fall, not the time spent underwater or the time taken to recover. Two portfolios with identical maximum drawdowns can feel completely different if one recovered in four months and the other took four years.

They do not explain the source of the return. No ratio here separates skill from luck, or return earned from broad market exposure from return earned by something the manager did. A benchmark-relative measure is needed for that, and even then the choice of benchmark drives the answer.

They can be improved without improving anything. Because these are computed numbers, they respond to choices: the window, the frequency, the risk-free rate, whether outliers are trimmed. Any of those can be selected after the fact to make a figure look better, which is why the stated convention matters as much as the value.

Used properly, the three ratios are a quick triangulation. Sharpe describes ordinary variability, Sortino describes the losing side of it, and Calmar describes the worst single stretch. If all three agree, you have a consistent picture. If they disagree, you have found the question that actually needs answering.

Related reading

• Portfolio metrics explained: the hub that maps every risk, return and turnover metric to the question it answers.
• What is the Sharpe ratio: excess return per unit of total volatility, and why it penalises upside swings too.
• What is the Sortino ratio: the downside-only version, and when it is the fairer measure.
• What is the Calmar ratio: return measured against the worst peak-to-trough fall.
• Risk-adjusted returns explained: the umbrella idea behind all three ratios.

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.