Skip to content
Financial Context Wiki

Sharpe Ratio: Measuring Return per Unit of Volatility

For educational purposes only; not investment advice.

Direct answer

Section titled “Direct answer”

The Sharpe ratio estimates how much average return above a reference risk-free return was earned per unit of total return volatility:

Sharpe ratio = mean(Rp - Rf) / standard deviation(Rp - Rf)

Rp is the portfolio return and Rf is the risk-free return for the same period. A larger historical value means the observed excess return was high relative to its variability. It does not identify the source of return, measure every form of risk, or predict future performance.

How the calculation works

Section titled “How the calculation works”

Build a sequence of comparable total returns after the fees being evaluated, not one annual return divided by one annual volatility estimate. Convert the risk-free rate to the same currency, compounding convention, and daily or monthly period, then subtract it observation by observation. Calculate the arithmetic mean and sample standard deviation of that excess-return sequence.

If returns are independent and similarly distributed, a periodic ratio is commonly annualized as:

annualized Sharpe ≈ periodic Sharpe × √N

N is often 12 for monthly observations or about 252 for trading-day observations. This square-root rule is an assumption, not a unit conversion. Serial correlation, stale prices, overlapping returns, changing leverage, and volatility clustering can make it materially wrong. Different data frequencies can therefore produce different ratios for the same strategy.

The choice of risk-free return also matters. It should match the return currency and observation period; a convenient Treasury yield cannot simply be subtracted from daily returns without conversion. When comparing managers, use the same sample dates, return definition, risk-free series, frequency, and annualization method.

Worked example

Section titled “Worked example”

Suppose a portfolio produced 12 monthly returns. Its average monthly total return was 1.0%, the comparable monthly risk-free return was 0.3%, and the standard deviation of monthly excess returns was 2.5%:

monthly Sharpe = (1.0% - 0.3%) / 2.5% = 0.28

Under the independent-return approximation:

annualized Sharpe ≈ 0.28 × √12 ≈ 0.97

This is not the same as taking a 12% annual return, subtracting an annual risk-free rate, and dividing by monthly volatility. It is also a fragile estimate: 12 observations give substantial sampling uncertainty. Report the sample period and number of observations alongside the result.

Now compare Portfolio B over exactly the same months. If its mean monthly excess return is 0.5% and standard deviation is 1.5%, its annualized ratio is about 0.5 / 1.5 × √12 = 1.15. B had the higher historical risk-adjusted result even though its raw return was lower. That comparison says nothing by itself about liquidity, drawdown depth, tax, capacity, or future persistence.

Interpretation and risk checks

Section titled “Interpretation and risk checks”
  • Use total returns with consistent treatment of distributions, fees, cash flows, and currency. An investor return may differ from a fund’s time-weighted return.
  • Inspect the full distribution. Standard deviation penalizes upside and downside variation equally and does not describe skewness, fat tails, or maximum loss.
  • Treat a negative ratio carefully. Ranking negative Sharpe ratios can be unintuitive, and adding leverage does not repair a strategy whose expected excess return is negative.
  • Check for smoothed or infrequent valuations. Appraisal-based assets and thinly traded securities can show artificially low volatility.
  • Examine autocorrelation before applying √N. Lo’s adjustment demonstrates why serial dependence changes annualization.
  • Look beyond a high ratio produced by selling options or credit protection. Many small gains followed by rare large losses can appear attractive in a short sample.
  • Avoid selection bias. Backtest optimization, survivorship, multiple testing, and choosing a favorable start date can inflate the reported value.
  • Compare like with like and include uncertainty. A small difference between two estimated ratios may be noise rather than evidence of skill.

The ratio is most useful as one diagnostic beside cumulative return, drawdown, downside risk, beta and factor exposures, liquidity, turnover, leverage, and stress scenarios.

Common misconceptions

Section titled “Common misconceptions”
  • “A Sharpe ratio above a fixed number is always good.” There is no universal threshold independent of horizon, asset class, method, and data quality.
  • “It measures return per dollar of loss.” Its denominator is standard deviation, not loss, drawdown, or capital at risk.
  • “The risk-free rate can be omitted.” That may be a disclosed approximation when rates are negligible, but it changes the metric and can distort comparisons.
  • “Monthly and daily ratios should match after annualization.” They need not match when returns are dependent, prices are stale, or sampling hides intraperiod moves.
  • “A high backtest Sharpe proves a durable edge.” It may reflect overfitting, costs omitted from returns, a calm regime, leverage, or hidden tail exposure.
  • “Sharpe and Sortino ratios answer the same question.” Sortino uses a downside-deviation convention; its value depends on a chosen target and is not directly interchangeable.
Section titled “Related topics”

Authoritative sources

Section titled “Authoritative sources”