Skewness

What is Skewness?

Skewness is the third statistical moment of a return distribution, after the mean (the first moment) and the variance that produces volatility (the second). Where volatility tells you how wide the spread of returns is, skewness tells you whether that spread is symmetric or lopsided. A symmetric distribution has skewness near zero: gains and losses of a given size are roughly equally likely. A negatively skewed distribution has a long left tail, meaning the rare extreme moves are losses, while the typical month is a modest gain. A positively skewed distribution has a long right tail, where the rare extreme moves are gains and the typical month is a small loss.

The sign is the part that matters most to an investor. Negative skewness is the uncomfortable kind. It describes strategies that look calm and consistent most of the time but occasionally suffer a sharp loss that erases many months of steady gains. Selling insurance, shorting volatility, and many carry trades all share this shape: small, reliable income punctuated by infrequent severe drawdowns. Positive skewness is the opposite and is generally easier to live with, even when it feels less pleasant day to day, because the rare surprises work in your favor. Trend-following and long-option strategies tend to be positively skewed: many small losses while waiting, redeemed by occasional large wins.

Equity index returns are mildly negatively skewed over most horizons, which is part of why crashes feel more violent than rallies; the market tends to take the stairs up and the elevator down. Reading skewness on a backtest tells you which kind of surprise the strategy is prone to, information that neither the average return nor the volatility can convey, because both of those treat an upside and a downside move of equal size identically.

Formula

The formula standardizes each return by subtracting the mean and dividing by the standard deviation, cubes the result, and averages. Cubing is what makes skewness directional: a cube preserves sign, so large negative deviations contribute large negative terms and large positive deviations contribute large positive terms. Because the deviations are standardized first, skewness is unitless and comparable across strategies with different volatilities. The version shown is the population (biased) estimator, which divides by N. There is also a sample-adjusted estimator that multiplies by a correction factor of N² / [(N−1)(N−2)] to reduce small-sample bias, the form many spreadsheet functions use by default; the two converge as the number of observations grows.

SledgeKey computes skewness on the strategy's monthly returns over the test window using the standardized third-moment definition above. The reported value is a single dimensionless number: zero means symmetric, a clearly negative value flags a left-tailed strategy whose surprises are losses, and a clearly positive value flags a right-tailed strategy whose surprises are gains. As a rough reading, values between about negative 0.5 and positive 0.5 are close to symmetric, while values beyond roughly negative 1 signal a pronounced and worth-noting left tail.

Why Skewness Matters in Backtesting

Skewness informs whether a strategy's headline numbers are telling you the whole story. Two strategies can post the same average return and the same volatility, and therefore the same Sharpe ratio, while having opposite skew. One quietly accumulates and then gives a chunk back in a crash; the other grinds out small losses and then catches a windfall. Sharpe cannot distinguish them because it only uses the first two moments. Skewness is the metric that exposes the difference, and for most investors a negatively skewed path is harder to hold through, both psychologically and financially, because the bad surprises arrive when liquidity and nerve are scarcest.

The failure mode of ignoring skewness is being seduced by a smooth track record that hides a fragile shape. Strategies that systematically harvest small premiums, the short-volatility family being the clearest case, can show beautiful Sharpe ratios for years precisely because their losses are rare. The 1998 collapse of Long-Term Capital Management and the 2018 unwind of short-volatility products are textbook illustrations: returns that looked steady and low-risk by standard measures were sitting on a deeply negative skew that materialized all at once. A backtest that reports strong returns with low volatility but sharply negative skewness is waving exactly that flag.

Skewness is most useful read alongside the metrics that quantify the tail it points to. Skewness tells you which direction the surprises lean; Conditional VaR and Maximum Drawdown tell you how deep the downside surprises actually go; kurtosis tells you how frequent the extremes are. Together they describe the shape of the return distribution far more completely than mean and volatility alone, and they turn a single Sharpe number into a fuller risk profile.

How SledgeKey Implements Skewness

Skewness appears on the backtest results page as a single dimensionless number within the risk metrics, computed from the same monthly returns the rest of the results use. There is no unit and no time horizon attached, because skewness is a shape statistic rather than a return or a rate: it simply reports whether the monthly returns lean left, lean right, or sit roughly symmetric around their average. A negative reading signals that the strategy's rare large moves were losses; a positive reading signals they were gains.

The benchmark's skewness is computed the same way over the identical window, so the comparison is the useful reading. A strategy more negatively skewed than the benchmark carries a more dangerous asymmetry than the index, even if its average return looks attractive; a strategy with skewness closer to zero or positive has a friendlier distribution of surprises. Because skewness depends on cubed deviations, it is dominated by the most extreme months and is therefore noisy over short windows, where a single outlier can swing the number substantially. Read it as a directional signal about the tail rather than a precise figure, and lean on longer windows before drawing firm conclusions.

Common Pitfalls

The first pitfall is assuming positive skewness is always good and negative always bad. The sign tells you which way the surprises lean, not whether the strategy is worth holding. A positively skewed strategy can still lose money on average if its frequent small losses outweigh its rare wins, and a mildly negatively skewed strategy with a strong average return can be perfectly sound. Skewness is a description of distribution shape, not a verdict; read it together with the return and the tail metrics rather than as a standalone buy or avoid signal.

The second pitfall is trusting a skewness estimate from a short sample. Because the calculation cubes deviations, it is extraordinarily sensitive to the single most extreme observation. A backtest of only two or three years can show a strongly negative skewness that is really just one bad month, and the number can flip sign if the window shifts slightly. Skewness from a short window is a hint, not a measurement; the longer the record and the more market regimes it spans, the more the figure can be trusted.

The third pitfall is confusing skewness with kurtosis or with the magnitude of risk. Skewness is purely about direction of asymmetry; it says nothing about how wide the distribution is or how fat its tails are. A strategy can have near-zero skewness and still be dangerous if it has high kurtosis, meaning symmetric but frequent extremes in both directions. Skewness answers "which side are the surprises on," not "how big" or "how often"; those questions belong to volatility, the tail metrics, and kurtosis respectively.

Related Concepts