Subtle & customizable audible cues to indicate when charts turnover, plus world clocks for any market — by & for professional traders
Open the app
Subtle & customizable audible cues to indicate when charts turnover, plus world clocks for any market — by & for professional traders
Open the appGuidance last reviewed 2026-08-19
There is a whole family of ratios that claim to tell you whether a return was worth the risk taken to get it. Sharpe is the famous one. Sortino, Calmar, MAR, Sterling, Ulcer, Omega and Treynor are the rest.
They all have the same shape. Return on top, some definition of risk underneath. Every argument between them is an argument about the denominator — about what the word risk should mean.
This page covers what each one chooses, where each one breaks, and the awkward finding that they usually reach the same conclusion regardless.
| Ratio | Risk is measured as | Best at | Where it breaks |
|---|---|---|---|
| Sharpe | Standard deviation of returns | Being universally understood | Punishes big winning months |
| Sortino | Downside deviation only | Strategies with upside skew | Fewer observations, noisier |
| Calmar | Maximum drawdown, 36 months | Matching what you actually feel | One bad day sets the whole figure |
| MAR | Maximum drawdown, since inception | Long records | Gets kinder as time passes |
| Sterling | Average of the largest drawdowns | Resisting a single outlier | No single agreed definition |
| Ulcer Index | Depth and duration underwater | How long the pain lasted | Barely supported anywhere |
| Omega | The entire distribution | Fat tails and skew | Needs a threshold you must choose |
| Treynor | Beta against a benchmark | Judging one piece of a portfolio | Meaningless without a sensible benchmark |
You would expect these to disagree. Sharpe assumes returns are normally distributed and trading returns clearly are not; Sortino and Omega were built specifically to fix that.
In practice they rank the same strategies in almost the same order. Studies comparing performance measures across hedge funds — a population with famously non-normal returns — have found rank correlations above 0.97 between Sharpe and its more sophisticated replacements.
⭐ The practical consequence: choosing a different ratio rarely changes your conclusion. What changes your conclusion is understanding the one you use. Collecting nine ratios and understanding none is the common failure, and it feels like rigor while producing nothing.
The exception is the case each alternative was designed for — and those cases are real, which is why the rest of this page exists.
Sharpe = (return − risk-free rate) ÷ standard deviation of returns
The original, from 1966. Excess return per unit of volatility.
The famous objection: it punishes upside. Standard deviation counts a spectacular month as risk in exactly the way it counts a terrible one. A strategy with occasional huge winners is penalized for the thing that makes it good. That is the observation Sortino was built on.
⚠️ The more serious objection is that it is gameable. Any strategy that collects small premiums reliably and loses badly but rarely — selling options, carry trades, anything insurance-shaped — produces an excellent Sharpe right up until the event it was exposed to happens. A high Sharpe on a short record is exactly what a hidden tail risk looks like.
⚠️ It also moves with interest rates. The same returns produce a different Sharpe when the risk-free rate changes, so figures from different eras are not directly comparable.
Sortino = (return − target) ÷ downside deviation
Identical to Sharpe except the denominator counts only returns below a target — usually zero, sometimes the risk-free rate. Upside volatility stops being a penalty.
When it genuinely matters: strategies with positive skew — many small losses and occasional large gains, which is what trend following and most breakout trading look like. Sharpe understates those. Sortino does not.
⚠️ The catch is sample size. By discarding every positive period you compute the denominator from perhaps a third of your data. A Sortino ratio is a noisier estimate than a Sharpe from the same record, and its apparent precision is partly borrowed.
Calmar = annualized return ÷ maximum drawdown
Calmar conventionally uses the last 36 months. MAR is the same calculation over the entire record since inception.
⭐ This is the family member closest to lived experience. Nobody experiences standard deviation. Everybody experiences the largest drop from a peak. Calmar answers "what did I earn for the worst thing I sat through", which is the question people are actually asking.
⚠️ And it rests entirely on one number. Maximum drawdown is a single observation — the worst one — so Calmar is set by a single stretch of the record and ignores everything about the rest.
⚠️ MAR flatters long records. Because the maximum drawdown stops growing while returns keep accumulating, an old fund's MAR improves simply by continuing to exist. That is not skill; it is arithmetic.
Sterling = annualized return ÷ average of the largest drawdowns
The fix for Calmar's dependence on one bad stretch: average several of the worst drawdowns instead of taking the single worst.
⚠️ There is no single agreed definition. Some versions average the largest three annual drawdowns, some add an arbitrary 10% to the denominator, some use every drawdown beyond a threshold. Two platforms can report different Sterling ratios for the same record and both be right by their own definition — the same problem the K-ratio has, and the same rule applies: never compare a Sterling ratio to one produced somewhere else.
Ulcer Index = the root-mean-square of percentage drawdown, measured at every point
The odd one out, and the most under-used. It is not a ratio at all but a risk measure, designed to be a denominator.
It is the only one that counts DURATION. Every other measure here treats a 20% drawdown recovered in three weeks and a 20% drawdown that took eleven months as the same event. The Ulcer Index does not: it samples the drawdown continuously, so time spent underwater accumulates.
⭐ That makes it the closest thing to a measure of what a strategy costs to hold, which is usually what decides whether someone is still trading it at the end.
⚠️ Support for it is thin. Most platforms do not report it, so you may have to compute it yourself.
Omega = probability-weighted gains above a threshold ÷ probability-weighted losses below it
Instead of summarizing the return distribution with one or two numbers, Omega uses the whole thing. No assumption of normality, and skew and fat tails are captured rather than averaged away.
⚠️ It requires you to pick the threshold, and the answer depends on it. That is honest — the threshold is your definition of an acceptable return — but it means an Omega ratio without its threshold stated is not interpretable.
⚠️ It also needs a lot of data to estimate a distribution rather than describe one.
Treynor = (return − risk-free rate) ÷ beta
Uses sensitivity to a benchmark instead of total volatility.
Built for a component, not a whole. Treynor asks what a holding contributed relative to market exposure, and it assumes everything else has been diversified away. For a single trading account it is usually the wrong tool — and 🔴 beta against a benchmark you did not choose deliberately is a number, not a measurement.
| Ratio | Fails when |
|---|---|
| Sharpe | Returns are skewed, tails are hidden, or the record is short |
| Sortino | The sample is small — the denominator uses a fraction of it |
| Calmar | The worst drawdown was unrepresentative, good or bad |
| MAR | The record is long — it improves with age alone |
| Sterling | Comparing across platforms, because definitions differ |
| Ulcer | You need someone else to have computed it |
| Omega | The threshold is unstated, or data is thin |
| Treynor | The benchmark is arbitrary, or the account is not diversified |
Every ratio here is computed on a series of PERIODIC RETURNS — daily, weekly or monthly — not on a list of trades.
That matters more than it sounds. If you hold a trade log, you do not yet have the input any of these need. You have to turn trades into an equity curve, then sample it at fixed intervals, and the interval you choose changes the answer. The same account produces different Sharpe ratios computed daily versus monthly, and the convention for annualizing between them assumes returns are independent — which for a trader running one idea across correlated positions, they are not.
⚠️ So a Sharpe ratio quoted for a personal trading account deserves a follow-up question: computed on what, and how often? Two platforms can hand you very different numbers from identical trades without either being wrong.
⭐ This is exactly the gap the K-ratio was designed for — it works on the equity curve directly and asks whether it rose steadily or in lucky lumps, rather than requiring you to invent a returns series first.
| Term | Meaning |
|---|---|
| Risk-free rate | The return available with no risk — conventionally short-dated government debt |
| Standard deviation | How widely results scatter around their average |
| Downside deviation | The same, counting only results below a target |
| Skew | Whether the distribution leans — a long tail on one side |
| Fat tails | Extreme results happening more often than a normal distribution predicts |
| Beta | How much something moves for a given move in its benchmark |
| Drawdown | A fall from a previous peak in account value |
| Underwater | The stretch between a peak and recovering it |
| Annualized | Scaled to a yearly equivalent for comparison |
These ratios judge an account over time. They say nothing about individual trades — whether your exits worked, whether your stops were sized right, whether one trade carried the record.
For that, see Trading Performance Metrics Explained, which covers win rate, profit factor, expectancy, R-multiples, drawdown and MAE/MFE — and What is the K-Ratio? for the shape of the curve itself.
Conventions differ between platforms — thresholds described as rules of thumb are conventions rather than findings, and it is worth checking how your own tools calculate a figure before comparing it with anyone else's.
Chart Ding is a market clock with customizable alarms for traders — candle-close alerts, world market sessions, and holiday warnings. Open the app.
Nothing on this website is trading or investment advice. Chart Ding is a market clock and alarm tool — it does not recommend trades, evaluate strategies, or take account of anyone's circumstances. Nothing here should be construed as a recommendation or relied on as the basis for a trading decision. Consult a licensed professional before placing real money at risk. Trading involves risk of loss.