TL;DR. The Sharpe ratio measures how much excess return a portfolio earned per unit of risk it took. It is computed as (portfolio return − risk-free rate) ÷ portfolio standard deviation. Higher is better, but the number is highly sensitive to the time period, the risk-free rate you choose, and the assumption that returns are roughly bell-shaped. William F. Sharpe — who invented it in 1966 and refined it in 1994 — explicitly cautioned that the ratio is one diagnostic, not a verdict.
What the Sharpe Ratio Actually Measures
An investor weighing two strategies needs to know more than which one earned a bigger number. A fund that returned 15% by leveraging junk bonds is doing something different from a fund that returned 12% in a basket of investment-grade credit, even if the first looks better on a one-line return chart. The Sharpe ratio is the simplest, most-cited way to put both funds on a common footing: it asks how much they were paid for each unit of risk they accepted.
William F. Sharpe introduced the ratio in 1966 (then called the “reward-to-variability” ratio) and revisited it in a 1994 essay in the Journal of Portfolio Management, where he laid out the modern formulation that practitioners still use.[1] Two pieces of vocabulary matter:
- Excess return — the return of the portfolio above what you could have earned in a “riskless” asset, typically a short-dated Treasury bill. Investors should not be rewarded for returns they could have collected risk-free.
- Total risk — measured as the standard deviation of the portfolio’s returns. This counts both upside and downside variation around the mean.
The Formula
In its ex-post (historical) form, the Sharpe ratio is:
Sharpe = ( Rp − Rf ) ÷ σp
Where Rp is the portfolio’s realized return over the measurement window, Rf is the risk-free rate observed over the same window, and σp is the standard deviation of the portfolio’s excess returns. Sharpe’s 1994 paper uses the differential-return formulation explicitly: it is the historic average differential return per unit of historic variability.[1]
The ex-ante (forward-looking) version swaps the historical numbers for expectations: expected excess return divided by the standard deviation of that expected excess return. Sharpe’s original 1966 paper used the ex-ante form for selecting among mutual funds; today, ex-post is what shows up on factsheets.
Two technical conventions you should not skip:
- Use the same frequency throughout. If you compute returns monthly, both the numerator and denominator are monthly. To “annualize,” multiply the monthly Sharpe by √12 (or √252 for daily). Mixing daily volatility with annual returns is the most common silent error in factsheet Sharpe ratios.
- Use the risk-free rate over the same period, not today’s spot T-bill. If the 1-year window you are evaluating included three rate hikes, your Rf should be the average T-bill rate over that window, not the latest one.
A Worked Example: Two Funds With the Same Return
Suppose two equity funds each returned 11.0% over the past 12 months. The 3-month Treasury bill yielded an average of 3.59% over the same window (the latest H.15 print is 3.59% as of May 22, 2026).[2] Fund A’s monthly returns had a standard deviation that annualized to 11%; Fund B’s annualized to 18%. Same return, very different ride.
| Metric | Fund A (low vol) | Fund B (high vol) |
|---|---|---|
| Trailing 12-month return | 11.00% | 11.00% |
| Average 3-month T-bill rate (Rf) | 3.59% | 3.59% |
| Excess return (Rp − Rf) | 7.41% | 7.41% |
| Annualized standard deviation | 11.0% | 18.0% |
| Sharpe ratio | 0.67 | 0.41 |
Fund A delivered the same return with materially less volatility, so its Sharpe ratio is roughly 60% higher. If you had to pick one with no other information, the Sharpe ratio says Fund A used its risk budget more efficiently. That does not mean Fund A is the better fit for every portfolio — see the pitfalls section below — but it is a meaningful starting point.
How to Interpret the Number
The Sharpe ratio is unitless, which is convenient: a value of 1.0 means the portfolio earned one unit of excess return for each unit of standard deviation. A widely cited industry rule of thumb breaks it down like this:[3]
| Sharpe ratio band | Common interpretation |
|---|---|
| Less than 1.0 | Sub-par for the risk taken |
| 1.0 – 1.99 | Adequate to good |
| 2.0 – 2.99 | Very good |
| 3.0 or higher | Excellent (and worth scrutinizing for hidden risk) |
Treat the thresholds as rough scaffolding. A 0.50 long-term Sharpe is approximately what a buy-and-hold US large-cap stock investor has historically earned through full economic cycles, and a 1.0+ multi-year Sharpe on a diversified portfolio is genuinely difficult to sustain. Anything north of 3.0 sustained over many years almost always warrants a closer look at whether the strategy harbors tail risk that hasn’t yet shown up in the standard deviation.
Visualizing the Risk-Return Tradeoff
The picture above is the single most useful mental model for the Sharpe ratio. Plot a portfolio on a chart of excess return versus standard deviation; the line from the origin to that point has slope equal to the Sharpe ratio. Two portfolios with the same return but different volatility do not sit on the same line. The steeper the line, the more efficient the portfolio is — and in mean-variance theory, you would prefer to combine the highest-Sharpe risky portfolio with cash or borrowing to reach your desired risk level.
Common Mistakes and When the Number Misleads
Sharpe himself enumerated several places his ratio can mislead.[1] The big ones, paraphrased:
- Time-period dependence. The ratio computed over a 1-year window is not the same number you would get over 5 years. Bull markets inflate Sharpe ratios across the board; volatility spikes deflate them. Always note the window.
- It says nothing about correlation. A fund with a Sharpe of 0.8 that is uncorrelated with your existing portfolio can be a better add than a fund with a Sharpe of 1.2 that doubles your existing exposure. Sharpe explicitly warns that when you already hold risky assets, picking the highest Sharpe is not always optimal.
- Past Sharpe is not a forecast. Sharpe wrote that the reliability of using historical ratios as surrogates for the future is “subject to serious question.” Manager skill rotates; risk premia drift.
- It assumes returns are well-behaved. Standard deviation is the right risk measure if returns are roughly symmetric and bell-shaped. Strategies that sell tail risk — selling deep out-of-the-money options, short-vol carry trades, certain credit strategies — produce a long string of small wins punctuated by infrequent large losses. Their realized Sharpe ratios look spectacular right up until they don’t.
- It penalizes upside volatility. Because standard deviation counts moves in both directions, a fund that has a few unusually good months is dinged the same as one that has a few unusually bad months. This is why the Sortino ratio — which only counts downside deviation — exists.
One more practical trap: with very low or negative excess returns, the Sharpe ratio loses interpretive force. If Rp − Rf is negative, a higher standard deviation actually makes the (negative) ratio look less bad. In that regime, do not compare Sharpe ratios across funds.
How the Risk-Free Rate Changes the Picture
This is why comparing a 2021-vintage Sharpe ratio to a 2026 one is misleading on its own. When Rf was near zero, every fund’s Sharpe ratio got a tailwind: the bar to clear was the floor. With T-bills near 3.6%, the bar is materially higher, and a fund that looked attractive in the zero-rate era may be barely beating cash on a risk-adjusted basis today.
The Sharpe Ratio’s Cousins
Several risk-adjusted measures share the same shape — excess return on top, some flavor of risk on the bottom — but use different denominators to fix specific shortcomings:
- Sortino ratio. Replaces standard deviation with the standard deviation of only negative returns. Useful when upside volatility is a feature, not a bug (most investors don’t mind their fund spiking up).
- Treynor ratio. Replaces standard deviation with the portfolio’s beta. Treats only market risk as the cost; appropriate when the portfolio is one piece of a larger diversified book.
- Information ratio. Numerator is the return above a benchmark (not the risk-free rate); denominator is the standard deviation of that tracking error. The natural metric for an active manager whose mandate is to beat the S&P 500, not cash.
- Calmar / MAR ratio. Replaces standard deviation with maximum drawdown. Popular with trend-following CTAs whose investors care more about the worst peak-to-trough loss than month-to-month wobble.
None of these replaces the Sharpe ratio outright; they complement it. A serious due-diligence packet will show several risk-adjusted metrics side by side so the limitations of each are partly offset by the others.
What to Learn Next
If the Sharpe ratio made sense, the natural next stops are: (1) the Capital Asset Pricing Model, which gives the theoretical foundation for thinking about excess return per unit of systematic risk; (2) Markowitz mean-variance optimization, which formalizes how to build the highest-Sharpe portfolio from a menu of assets; and (3) the empirical literature on factor returns (size, value, momentum, quality), which explains why some long-only portfolios earn higher Sharpe ratios than the market over long horizons. Each one is a refinement on the basic idea that risk-adjusted return, not raw return, is the thing worth maximizing.
Sources
- William F. Sharpe, “The Sharpe Ratio,” Journal of Portfolio Management, Fall 1994 (Stanford mirror of the original paper).
- Federal Reserve Board H.15 Selected Interest Rates — daily 3-month Treasury bill rate, used here for the current Rf reference.
- Corporate Finance Institute, “Sharpe Ratio Definition and Formula” — source for the industry interpretation thresholds.
- Aswath Damodaran, NYU Stern, “Historical Returns on Stocks, Bonds and Bills: 1928–2024” — canonical long-run reference for stock, bond, and bill returns used to anchor “what a typical Sharpe looks like.”
Disclosure: This article was produced with AI assistance and reviewed before publication. It is for informational purposes only and is not investment advice.