Sharpe ratio

The Sharpe ratio, developed by Nobel laureate William F. Sharpe, measures the performance of an investment compared to a risk-free asset, after adjusting for its risk. It is calculated using the formula:

where $R_p$ is the return of the portfolio, $R_f$ is the risk-free rate (e.g., the return on U.S. Treasury bills), and ${\sigma }_p$ is the standard deviation of the portfolio’s excess return, representing the total risk. The benchmark in a Sharpe ratio is the risk-free rate, specifically — substituting some other target return gives you a different statistic that should not be called a Sharpe ratio. The minimum-acceptable-return generalization belongs to the Sortino ratio (section “Sortino ratio”), which is what the next subsection is for.

Interpretation A higher Sharpe ratio indicates a more desirable risk-adjusted return. For example, a Sharpe ratio of 1.5 is better than a Sharpe ratio of 1.0. It is particularly useful when comparing investments with differing levels of volatility — and the whole point of the ratio is that it can reverse the naive ranking. Take a 4% risk-free rate. Fund A returns 10% with 20% volatility: Sharpe $=(10-4)∕20=0.30$. Fund B returns 7% with 6% volatility: Sharpe $=(7-4)∕6=0.50$. B “lost” by three points of return and won by any risk-adjusted reading — lever B to A’s volatility and it would have returned about 14%.

A negative Sharpe ratio means the portfolio returned less than the risk-free rate — not that it lagged a market benchmark, which is a different and much weaker statement. A fund can beat its index handily and still post a negative Sharpe in a year when Treasury bills outyielded the whole asset class. Generally, investors prefer a higher positive Sharpe ratio, which signifies either higher returns or lower volatility. However, increasing the Sharpe ratio from negative values can be achieved by either increasing returns, which is beneficial, or increasing volatility, which is typically undesirable. Therefore, the Sharpe ratio may not align well with typical investor utility functions when it is negative.

The calculation of the Sharpe ratio is straightforward as it only requires a series of observed returns, without the need for additional information about the source of profitability. Nonetheless, this simplicity makes it susceptible to manipulation, particularly through the smoothing or discretionary pricing of illiquid assets. To detect such manipulations, statistical measures like the bias ratio and first-order autocorrelation are often employed.

The Sharpe ratio accounts for both systematic and idiosyncratic risks, with the relevance of each depending on the specific portfolio context. The returns used in the calculation can be of any frequency—daily, weekly, monthly, or annually—and are typically annualized. However, a critical limitation of the Sharpe ratio is its assumption of normally distributed returns. In reality, asset returns often exhibit abnormalities such as kurtosis, skewness, and non-normal distribution, which can undermine the effectiveness of the standard deviation in the Sharpe ratio calculation.

The selection-bias problem, which is worse than the distributional one. A Sharpe ratio computed on the strategy you chose after examining many is not the Sharpe ratio of that strategy — it is the maximum of a sample of noisy estimates, and the maximum of $N$ draws is biased upward by an amount that grows with $N$. Bailey and López de Prado formalize this as the deflated Sharpe ratio,⁵⁷ which discounts the observed figure for the number of trials run, the length of the track record, and the skewness and kurtosis of the returns. The practical consequence when someone shows you a backtest: ask how many variants were tested before this one was selected. If the answer is "a few hundred," a Sharpe of 1.5 on five years of monthly data is consistent with a true Sharpe of zero. This is the single most useful question you can ask about any quantitative track record, and almost nobody asks it.