Sharpe Ratio in Trading: Formula, Benchmarks and Limits
Learn how the Sharpe ratio works in trading, what counts as a good number, and when Sortino or Calmar tells you more, with a worked example.
The Sharpe ratio tells you how much return a strategy earned for each unit of volatility it put you through. You subtract the risk-free rate from the strategy's return, then divide by the standard deviation of its returns. Higher is better, but a single Sharpe number can flatter a dangerous strategy and punish a good one, so it works best next to the Sortino and Calmar ratios and a look at the actual trades.
This guide shows the math, a worked example, the benchmarks traders quote and why you should distrust them, and how to pick between Sharpe, Sortino and Calmar.
TL;DR
- Sharpe = (strategy return minus risk-free return) divided by the standard deviation of returns. Annualize it before you compare anything.
- Rules of thumb call 1.0 acceptable, 2.0 very good and 3.0 excellent, but these cutoffs are informal and depend on the data interval and sample length.
- Sortino counts only downside volatility. Calmar divides annual return by maximum drawdown. Each catches something Sharpe misses.
- A backtested Sharpe is a backward-looking estimate that overfitting inflates. Test it on data the strategy has never seen.
Table of Contents
- What the Sharpe ratio measures
- The formula, step by step
- What is a good Sharpe ratio?
- Sharpe vs Sortino
- Calmar ratio: return per unit of drawdown
- Calmar vs Sharpe in practice
- Where the Sharpe ratio misleads
- Our perspective: trust the trades, not the headline ratio
- A checklist before you trust any Sharpe number
- See these ratios on your own strategy
- FAQ
- Sources
What the Sharpe ratio measures
Economist William F. Sharpe developed the ratio in 1966 to compare an investment's return with its risk. The idea is simple. Two strategies can both make 15% a year, but if one gets there with violent swings and the other with a steady climb, you would not rate them equally. Sharpe puts a number on that difference.
In plain terms, it answers one question: how much excess return did I get for each unit of volatility I tolerated? Excess means return above what you could have earned with almost no risk, usually a Treasury bill rate in the US. Charles Schwab's explainer notes it is most often applied to mutual funds, ETFs and individual stocks, but it can be applied to any investment or portfolio, including a trading system.
If you want a plain-language companion read, this everyday-investor walkthrough of the Sharpe ratio frames it as a backward-looking check on historical risk-adjusted returns, which is exactly how a trader should treat it.
The formula, step by step
Sharpe ratio = (Rp − Rf) / σp
- Rp is the strategy's average return over a period.
- Rf is the risk-free return over the same period.
- σp is the standard deviation of the strategy's returns over that period.
For a trader, the practical steps look like this:
- Export the strategy's returns at a fixed interval: daily, weekly or monthly. Use the same interval for every strategy you compare.
- Subtract the risk-free rate for that interval from each return. For daily data, divide the annual T-bill yield by the number of trading days.
- Take the average of those excess returns.
- Take the standard deviation of the excess returns.
- Divide the average by the standard deviation.
- Annualize. With daily data, multiply by the square root of 252 trading days. With monthly data, multiply by the square root of 12.
Step 6 is where most mistakes happen. The ratio is sensitive to the measurement interval, and comparing a daily-based figure with a monthly one without putting both on an annual basis tells you nothing. Before you read any Sharpe figure in a video, a PDF or a report, ask which interval and which period it used.
A worked example
Take a hypothetical swing strategy tested over several years. These numbers are illustrative, not a real result.
| Input | Value |
|---|---|
| Annualized return | 18% |
| Risk-free rate | 4% |
| Annualized volatility (standard deviation) | 10% |
| Annualized downside deviation | 5% |
| Maximum drawdown | 12% |
- Sharpe: (18% − 4%) / 10% = 1.4
- Sortino: (18% − 4%) / 5% = 2.8
- Calmar: 18% / 12% = 1.5
Three different numbers describe the same equity curve. None is wrong. Each asks a different question, which is why the next sections matter.
What is a good Sharpe ratio?
Common guidance says a Sharpe ratio above 1.0 is acceptable, above 2.0 is very good, and above 3.0 is excellent. Those labels circulate widely, but as Schwab's explainer points out, they are conventions without official backing, and because the number moves with the measurement interval, fixed cutoffs are unreliable for comparing very different strategies.
A more useful way to read it:
| Sharpe (annualized) | Rough reading for a trading strategy |
|---|---|
| Below 0 | Lost money relative to cash. |
| 0 to 1 | Weak to modest. Often not worth the risk and effort. |
| 1 to 2 | Respectable if it holds on out-of-sample data. |
| 2 to 3 | Strong. Check for overfitting before you celebrate. |
| Above 3 | Rare. Treat it as a warning sign until proven otherwise. |
That last row surprises people. Retail traders often see a Sharpe of 3 or 4 in a backtest and assume they have found something special. More often it signals one of a few problems: too little data, parameters tuned until the curve looked smooth, a lookahead bug, or hidden tail risk. We cover each below.
A few other points on "good":
- Is 0.5 good? For a standalone active strategy, it is mediocre. A buy-and-hold stock index fund can deliver a similar ratio with no effort, so a strategy has to beat that to justify the work.
- Sample length matters. A Sharpe estimated from six months of trades has a wide margin of error. A reliable estimate needs data long enough for every aspect of the strategy's returns to show up.
- Asset class matters. A trend-following futures system and a market-neutral equity book have different return shapes, so one benchmark does not fit both.
Sharpe vs Sortino
Sharpe treats every deviation from the average as risk, including the good ones. If your strategy has a big winning month, Sharpe's denominator grows and your ratio falls. That is odd, since nobody complains about upside surprises.
The Sortino ratio fixes this. It keeps the numerator, but divides by downside deviation, which only counts returns below a target you choose, often zero or the risk-free rate. Investopedia's comparison describes it as a variation of Sharpe that penalizes only returns falling below that minimum acceptable return.
Sortino ratio = (Rp − target) / downside deviation
The CME Group's education paper on the Sortino ratio is a good primary read if you want to see why the downside-only denominator was proposed.
When each one is the better tool
| Situation | Better fit |
|---|---|
| Smooth, low-volatility return stream | Sharpe |
| Trend-following, breakout or other strategies with a few large winners | Sortino |
| Comparing against a benchmark or fund factsheet that reports Sharpe | Sharpe, for consistency |
| Judging whether you could stay in the strategy through its worst stretch | Calmar or max drawdown |
The research summary we work from suggests Sharpe is typically used for lower-volatility portfolios, while Sortino is preferred for skewed strategies such as trend-following or option-selling. Trend systems are a clear case: they lose small and often, then win big. Sharpe penalizes the big wins as volatility. Sortino does not.
Two cautions on Sortino. First, it depends on your chosen target, so two people can report different Sortino ratios for the same strategy. Second, with few losing periods the downside deviation is estimated from very little data and can swing wildly. A short backtest with only a handful of losing trades can produce a spectacular, meaningless Sortino.
So what is a good Sortino? It runs higher than Sharpe for the same strategy by construction, since downside deviation is usually smaller than total deviation. A Sortino of 2 is a reasonable bar, and 3 or more deserves the same skepticism as a Sharpe above 3.
Calmar ratio: return per unit of drawdown
Sharpe and Sortino both use return volatility. Calmar uses the worst peak-to-trough loss instead.
Calmar ratio = annualized return / maximum drawdown
In our example, 18% a year with a 12% maximum drawdown gives 1.5. Some versions subtract the risk-free rate from the numerator, and many practitioners compute it over a trailing 36-month window. Whichever version you use, state it, because the numbers are not interchangeable.
Why traders like it: drawdown is what you actually feel. A 30% drop is a real event that makes people abandon strategies, cut size at the worst time or get margin calls. Volatility is an abstraction by comparison.
What counts as a good Calmar? There is no official scale. As a working rule, many traders want at least 1, meaning a year of average return would recover the worst drawdown, and treat figures well above 2 or 3 with caution unless the sample is long. The weakness is the same as with any single extreme: the maximum drawdown is one event in your history. A short backtest that happened to miss a rough market regime will show a lovely Calmar that the next bad year destroys.
Calmar vs Sharpe in practice
Put them side by side and the difference is clear:
| Ratio | Risk measure | Strength | Blind spot |
|---|---|---|---|
| Sharpe | Standard deviation of all returns | Universal, easy to compare | Penalizes upside, hides tail risk |
| Sortino | Downside deviation below a target | Rewards asymmetric strategies | Sensitive to target and sparse losses |
| Calmar | Maximum drawdown | Matches how losses feel | Depends on one single event |
A strategy can look excellent on one and poor on another. When they disagree, that is useful information. A high Sharpe with a mediocre Calmar suggests smooth returns that still contain one ugly drop. A high Calmar with a low Sharpe suggests choppy returns that never fell far. Our own worked example of Sharpe and Sortino on a backtest walks through how to read these side by side.
Where the Sharpe ratio misleads
It looks backward
A Sharpe ratio is computed from past returns. As Schwab's analysts note, it relies on historical returns and volatility that may not reflect future conditions. It audits the past. It does not forecast.
Tail risk hides in the denominator
Researchers have shown that Sharpe-type measures can be manipulated with option-like strategies. Selling out-of-the-money options, for example, collects small steady premiums and produces a very smooth curve until the rare crash. Standard deviation barely registers that risk before it happens. A very high Sharpe on such a strategy is a reason to check the worst-case losses, not a reason to relax.
Smoothed returns inflate it
When assets trade rarely, reported prices lag reality. A study of hedge funds from the SEC's site found that returns of hedge funds and other alternatives are often highly serially correlated, which understates volatility and flatters the ratio. For active traders, the lesson is to be wary of any curve that looks too smooth for its market.
Overfitting inflates backtested Sharpe
This is the big one for anyone testing strategies. Academic work on strategies built from several signals finds that combining signals and weighting them to maximize in-sample results creates selection bias, so backtested Sharpe ratios run well above what you can expect live. Every parameter you tweak after seeing results is another chance to fit noise.
Practical defenses:
- Keep a hold-out period the strategy never saw during design.
- Change one rule at a time, and keep a record of every variation you tried.
- Prefer simple rules with a plausible reason to work.
- Test across different market regimes, not just one trending stretch. Our piece on how much historical data a backtest needs shows how a handful of extra trials can undermine a ten-year Sharpe.
- Read our guide to spotting and stopping overfitting before you trust a clean curve.
Our perspective: trust the trades, not the headline ratio
We built Backtestify because we kept seeing the same pattern. A creator shows a strategy, quotes a win rate or a great-looking equity curve, and viewers trade it with real money. Almost nobody checks it against real historical data first.
Our view is that no single ratio should decide whether you trade something. Sharpe is a good summary, but it compresses hundreds of trades into one number, and a number can hide how the result was produced. We would rather you see win rate, profit factor and maximum drawdown together, then open the individual trades and ask whether the story makes sense.
We also think results should be published whether they flatter a strategy or not. That is why we publish reports on both winning and losing setups, and why we compare an original strategy against an improved version on recent data rather than only on the period where it was tuned. If an improvement raises Sharpe on old data but not on recent data, it was probably curve fitting, and you want to know that before risking capital.
If you want to see what this looks like, browse our published strategy library, or read how we simulate a strategy so you can judge the method yourself.
— WAJDI
A checklist before you trust any Sharpe number
- Interval and annualization. Is it daily, weekly or monthly data, and has it been annualized?
- Risk-free rate. Which rate was used? In a low-rate period it barely matters. When Treasury yields are high, it changes the answer noticeably.
- Sample size. How many trades and how many years? Does the period include a downturn?
- Costs. Are commissions, spreads and slippage included? Frequent traders can lose most of their edge here.
- Out-of-sample proof. Was any of the data held back from the design?
- Drawdown and tail. What is the worst loss, and could you have lived through it?
- Companion ratios. Do Sortino and Calmar tell the same story?
If you want a structured process for the test itself, our step-by-step guide to backtesting a trading strategy covers setup and common errors, and our regulator-style approach to backtesting covers what to document.
See these ratios on your own strategy
Backtestify lets you describe a strategy in plain English, test it on real historical market data, and read win rate, profit factor and maximum drawdown for every trade. The Free plan gets you started, and Pro is $29 per month or $190 per year, with unlimited backtesting, improvement and forecasting. You can also look at the Free plan first.
FAQ
What is a good Sharpe ratio for trading?
Above 1 is commonly treated as acceptable and above 2 as very good, but these labels are informal conventions. A figure above 3 in a backtest should make you look for overfitting, short samples or hidden tail risk. Judge the number against the same market, interval and period for each strategy you compare.
Is the Sharpe ratio or the Sortino ratio better?
Neither wins in every case. Sharpe is the standard for comparison, while Sortino suits strategies with large occasional winners because it ignores upside volatility. Many traders check both, and add Calmar to see how deep the worst loss ran.
How do you calculate the Calmar ratio?
Divide the strategy's annualized return by its maximum drawdown, using the absolute value of the drawdown. An 18% annual return with a 12% maximum drawdown gives a Calmar ratio of 1.5. Some versions subtract the risk-free rate first, so state which one you use.
Can a Sharpe ratio predict future returns?
No. It summarizes how a strategy behaved in the past, so it describes risk-adjusted efficiency, not future results. Markets change, and a strategy tuned to history often performs worse live.
Sources
- Charles Schwab: How to Calculate the Sharpe Ratio
- Investopedia: Sharpe Ratio: Definition, Formula, and Examples
- Investopedia: The Difference Between the Sharpe Ratio and the Sortino Ratio
- Wikipedia: Sharpe ratio
- CME Group: A Sharper Ratio (Sortino)
- NBER: Sharpe ratio manipulation with option-like strategies
- SEC: Hedge fund return smoothing study
- NBER: Backtesting Strategies Based on Multiple Signals