TL;DR. The Sharpe ratio measures how much extra return a portfolio earned per unit of total risk. You compute it by subtracting a risk-free rate from the portfolio return and dividing by the standard deviation of returns. Higher is better, but only when returns are roughly normal and the sample is long enough — otherwise the number lies to you.
What the Sharpe Ratio Actually Measures
The Sharpe ratio answers a single question: for every unit of volatility a portfolio took on, how much return did investors get above a truly safe asset? Two portfolios that both earned 10% look identical on a return line — until you notice one did it with the placid drawdowns of a short-duration bond fund and the other with the whiplash of a leveraged tech basket. The Sharpe ratio makes that difference legible.
William F. Sharpe introduced the measure in his 1966 paper “Mutual Fund Performance,” originally calling it the reward-to-variability ratio. He revised it in a 1994 Journal of Portfolio Management article that made the benchmark explicit: the numerator is not just “return minus T-bills,” but the return on the strategy minus the return on whatever benchmark portfolio the investor would otherwise have held.
The Formula
The ex-post (backward-looking) Sharpe ratio is:
S = ( Rp − Rf ) / σp
Where Rp is the portfolio’s realized return, Rf is the return on a risk-free asset over the same period (typically the 3-month U.S. Treasury bill), and σp is the standard deviation of the portfolio’s excess return.
The ex-ante (forward-looking) version replaces realized figures with expected ones. In practice, most Sharpe numbers you read — in fund fact sheets, Morningstar reports, quant research — are ex-post: they describe the past. That distinction matters because a fund’s past Sharpe is not a reliable forecast of its future Sharpe, as Sharpe himself warned.
Two conventions matter for reading a reported Sharpe:
- Annualization. Monthly Sharpe ratios are usually annualized by multiplying by √12; daily by √252. This assumes returns are independent across periods, which is often wrong for illiquid or auto-correlated strategies.
- Risk-free rate choice. As of the week ending August 10, 2026, the 3-month T-bill secondary-market rate stood at 3.74%, per the Federal Reserve H.15 release. That is the standard proxy for Rf in U.S. dollar work.
A Worked Example
Suppose you are choosing between two portfolios over the past 12 months. Assume the risk-free rate is 3.74%.
| Portfolio | Return | Volatility (σ) | Excess Return | Sharpe Ratio |
|---|---|---|---|---|
| A — Concentrated growth | 18.0% | 22.0% | 14.26% | 0.65 |
| B — Balanced 60/40 | 9.5% | 8.0% | 5.76% | 0.72 |
| C — Short-duration credit | 6.0% | 2.5% | 2.26% | 0.90 |
Portfolio A earned nearly twice the return of B, but B has the higher Sharpe. C, the sleepy short-duration credit book, quietly wins. That is the lesson: return alone flatters the loudest strategy; Sharpe rewards the quiet compounder — assuming its volatility number honestly captures its risk.
How Sharpe Behaves as Volatility Changes
The curve above is a concept diagram, not a real time series. It shows the mechanical point that dominates every Sharpe intuition: for a fixed excess return, doubling volatility halves the ratio. That linear-in-the-denominator behavior is why leverage does not, by itself, improve Sharpe. Levering a strategy 2x scales both numerator and denominator, leaving Sharpe unchanged (before financing costs and non-linear risks).
What Long-Run Sharpe Ratios Actually Look Like
Investor pitch decks love to advertise Sharpe ratios above 1.0. The historical record for real, long-lived, publicly measurable investment programs is more sober.
| Program / Asset | Period | Annualized Sharpe |
|---|---|---|
| U.S. equity market | 1976–2017 | 0.49 |
| Berkshire Hathaway | 1976–2017 | 0.79 |
| Cash (3-mo T-bills) | By construction | 0.00 |
Two things stand out. First, even one of the greatest long-run investment records in history — Berkshire from the mid-1970s to 2017 — produced a Sharpe of about 0.79. Second, that is only about 0.30 higher than the passive U.S. equity market over the same span. When a pitch shows a “Sharpe of 3” over 18 months, ask what the sample size is and what the tails look like.
Historical S&P 500 Excess-Return Snapshot
Five Common Mistakes
1. Comparing Sharpe across incompatible time periods
Sharpe is not independent of the window over which it is measured. Sharpe himself wrote that “the Sharpe Ratio is not independent of the time period over which it is measured.” A three-year Sharpe from a bull market is not comparable to a ten-year Sharpe that spans a drawdown.
2. Ignoring auto-correlation in monthly returns
Illiquid or private-market strategies often smooth their reported returns because assets are marked to model, not to market. Smoothed returns understate true volatility. That inflates the denominator’s implied “risk” and pushes the reported Sharpe artificially high. Private credit funds and gate-heavy hedge funds are notorious for this.
3. Assuming normal returns
Standard deviation captures only the second moment of a return distribution. It says nothing about skew or kurtosis. Strategies that sell tail risk — short-vol books, put-writing, insurance-like credit trades — look great on Sharpe until the tail arrives. Long-Term Capital Management posted a stellar Sharpe until August 1998.
4. Interpreting negative Sharpe as if it were positive
When excess return is negative, dividing by a larger standard deviation makes Sharpe less negative, which superficially looks better. It is not. Comparing negative Sharpes across strategies is close to meaningless without also comparing raw returns and volatility side by side.
5. Confusing the benchmark
Sharpe’s 1994 revision explicitly generalized the numerator to any benchmark portfolio, not just T-bills. If you are evaluating a small-cap value manager against a small-cap value benchmark, the “risk-free” leg should be the passive index return, not cash. Using the wrong benchmark can flip conclusions.
When the Sharpe Ratio Misleads
Sharpe writes that its ratio “does not take correlations into account. When a choice may affect important correlations with other assets in an investor’s portfolio, such information should be used to supplement comparisons based on Sharpe Ratios.” That is the deep limitation: a low-Sharpe asset that is negatively correlated with your existing book can improve the Sharpe of the whole portfolio far more than a high-Sharpe asset that duplicates what you already own. Sharpe is a stand-alone score, not a portfolio-construction verdict.
Two rules of thumb protect against the biggest traps: require at least a decade of data before taking a Sharpe seriously, and always look at maximum drawdown next to the Sharpe number. If a fund reports Sharpe 1.8 and a 55% drawdown, the volatility number is not telling you the whole story.
Related Metrics Worth Knowing
Once you understand Sharpe, three cousins are worth learning:
- Sortino ratio. Same numerator, but the denominator uses only downside deviation, penalizing losses without punishing upside volatility. See our Sortino explainer.
- Treynor ratio. Divides excess return by beta rather than total volatility. Useful when the portfolio is one component of a diversified book and only systematic risk matters.
- Information ratio. Excess return over a specific benchmark divided by the standard deviation of that excess return (tracking error). Standard for judging active managers against a passive index.
Sources
- William F. Sharpe, “The Sharpe Ratio,” Journal of Portfolio Management, Fall 1994, 21(1): 49–58 (author’s hosted version, Stanford).
- William F. Sharpe, “Mutual Fund Performance,” Journal of Business, 1966, 39(S1): 119–138 (original paper).
- Federal Reserve, H.15 Selected Interest Rates, week ending August 10, 2026 (3-month T-bill secondary market rate: 3.74%).
- Wikipedia contributors, “Sharpe Ratio” (summary of Frazzini, Kabiller & Pedersen, “Buffett’s Alpha,” Financial Analysts Journal, 2018 — Berkshire Sharpe 0.79 vs. U.S. market 0.49, 1976–2017).
Disclosure: This article is for informational purposes only and is not investment advice.