Sharpe, Sortino and Calmar ratios: three ways to measure return per unit of risk

Risk adjusted ratios put the return and the risk on one scale. Each divides some measure of reward by some measure of pain. They differ in what they count as pain: all volatility, only the bad volatility, or the worst fall. The risk metrics module covers volatility and drawdown on their own; this guide puts the three ratios side by side.

Sharpe ratio = (Rp − Rf) ÷ σp

• Rp is the portfolio's return over the period, usually annualised.
• Rf is the risk free rate for the same period, typically a short term government or money market rate. A factsheet footnote names the rate it used.
• σp is the standard deviation of the portfolio's returns, annualised. For monthly returns, the usual convention multiplies the monthly standard deviation by the square root of 12.

The numerator is the extra return earned for taking risk at all. The denominator is how much the returns swung around their own average. A higher Sharpe means more excess return per unit of swing.

The Sharpe ratio is the most common of the three because the inputs are easy to compute and it fits portfolio theory neatly. Its weakness is in the denominator: standard deviation counts a month of +8% as just as risky as a month of −8%. For an investor, those two months do not feel the same.

Sortino ratio = (Rp − T) ÷ DD

• T is a target or minimum acceptable return. Many calculators set it to the risk free rate; others use zero. Whichever is used, it has to be stated.
• DD is the downside deviation: the square root of the average of the squared shortfalls below the target, where any period at or above the target counts as zero.

So a portfolio that swings a lot on the way up but rarely falls below the target gets a much better Sortino than Sharpe. A portfolio whose swings are mostly downward gets little benefit.

The squared shortfalls add up to 34. Divided by all six months, that is 5.67, and its square root is about 2.38%: the monthly downside deviation. The ordinary standard deviation of the same six months is about 4.2%, because it also counts the +6% and +4% months as swings. The upside months raise the Sharpe denominator but leave the Sortino denominator alone.

Calmar ratio = compound annual return ÷ maximum drawdown

The maximum drawdown is the largest fall from any peak to a later low, written as a positive number. Young's original version used the last 36 months of monthly data; many tools now apply the same formula to the whole record, which changes the result, so the window has to be stated.

The Calmar ratio has the most direct reading of the three. A Calmar of 0.5 means the portfolio's worst fall was twice its yearly return: two years of typical gains lost in one episode. It is also the ratio closest to how investors experience risk, because the deepest fall is the moment most people are tempted to sell.

Read it ratio by ratio:

• Sharpe prefers B. A earned two points more, but swung half as much again.
• Sortino prefers A. Much of A's extra volatility came from strong up months, which Sortino does not count. Its downside deviation is only two points higher than B's.
• Calmar prefers B by a wide margin. A's worst fall was 28%, almost twice B's, for two extra points of return.

None of these is the right answer on its own. An investor who cares most about the worst moment would weigh Calmar; one who accepts deep falls but dislikes frequent small losses might weigh Sortino. What the example shows is that a single ratio, quoted alone, hides a choice about which kind of risk matters.

• It penalises good surprises. Standard deviation counts upside swings as risk, so a strategy with occasional large gains can look worse than it was.
• It assumes the swings are well behaved. Returns with rare, very large losses can show a high Sharpe for years before the loss arrives. Strategies that sell options or hold hard to price assets are the usual examples.
• Smoothed prices flatter it. An asset valued infrequently shows less volatility than it really has, which raises the ratio.
• The inputs move it. The risk free rate chosen, the period and whether returns are daily, weekly or monthly all change the figure. Two Sharpe ratios are comparable only when all three match.

• The target is a choice. A Sortino with a 0% target and one with a 6% target are different numbers for the same record.
• It rests on fewer data points. Only the periods below the target feed the denominator, so a short record with few bad months gives a noisy estimate.
• It ignores the order of losses. Five small losses spread over five years and five in a row produce the same downside deviation, though the second is a drawdown and the first is not.

• It rests on one event. A single crash sets the denominator. Everything else that happened in the record is ignored.
• It depends on the window. A longer record has more chances to include a deep fall, so Calmar ratios over different lengths are not comparable. A record that happens to start after a crash looks better than it should.
• It ignores recovery time. A fall that recovers in three months and one that takes three years give the same Calmar if they are equally deep.

Every RupeeCase strategy page shows the model CAGR, the deepest fall and the Sharpe ratio side by side, with the dates they cover. Allcap Multi Asset holds 52 positions across equity, debt and gold, rebalanced every 2 weeks: model CAGR 40.47%, deepest fall −21.32%, Sharpe 1.81, as of 27 Apr 2026. All are model, back tested figures after costs, not client returns. The Sharpe and volatility calculator works out the ratio from your own numbers, and the note Sharpe, Sortino, Calmar: three leaderboards ranks the RupeeCase lineup on each.