Skewness measures the asymmetry of a return distribution.

Skewness measures the asymmetry of a return distribution. In the Sonar Sciences fundamentals material, the distribution of returns is described through moments including skewness, which captures whether outcomes are balanced around the mean or whether one tail is longer or heavier than the other. A symmetric distribution has skewness near zero. Positive skewness means the right tail is more pronounced, so the distribution contains occasional unusually large positive outcomes. Negative skewness means the left tail is more pronounced, so the distribution contains occasional unusually large negative outcomes.

This matters because two strategies can have similar averages and volatility while having very different payoff shapes. A positively skewed strategy can produce many modest losses or small gains punctuated by occasional large gains. A negatively skewed strategy can produce many modest gains or small losses punctuated by occasional large losses. Skewness therefore helps separate strategies that earn through frequent small outcomes from strategies whose results depend on rare tail events.

The mechanism is visible in histograms of returns. In the fundamentals material, return distributions are presented as shapes rather than only summary averages. When the mass of observations is concentrated near the center but one side stretches farther, the histogram is skewed. If the right side extends with a small number of large positive returns, the skew is positive. If the left side extends with a small number of large negative returns, the skew is negative. The same intuition carries to a time series of strategy returns: the sign of skewness reflects which side contains the more extreme outliers.

For quantitative strategies, this changes the interpretation of wins and losses. A strategy with negative skewness can appear stable for long periods because it realizes many ordinary outcomes and only occasionally experiences a severe drawdown or large loss. A strategy with positive skewness can look less smooth in routine periods while depending on infrequent outsized gains. Skewness is therefore not just a descriptive statistic. It is a compact way to summarize how the frequency and magnitude of outcomes differ across the gain and loss sides of the distribution.

Skewness also matters when evaluating whether an apparent edge is robust or a product of backtest selection. The Sonar Sciences backtest overfitting audit is built around the idea that repeated trialing can produce attractive in sample characteristics that do not survive out of sample. In that setting, an unusual skew profile can arise from selecting a specification that happened to capture a few extreme observations in the historical sample. If a strategy’s apparent appeal depends heavily on a small number of tail events, the asymmetry in returns may be fragile rather than structural.

The Sonar Sciences glossary entry on the Deflated Sharpe Ratio reinforces this point by emphasizing that performance evaluation must account for non normal return features such as skewness and kurtosis when many strategies or parameter choices are tested. Under these conditions, conventional summary metrics can overstate significance. Audits for backtest overfitting therefore look beyond headline statistics and ask whether the distributional shape, including skewness, remains credible after accounting for multiple testing and the possibility that a few extreme returns are driving the result.

In practice, skewness should be read alongside mean, volatility, drawdown, and other distributional diagnostics. It does not say whether a strategy is good or bad on its own. It says whether the return pattern is tilted toward rare large gains or rare large losses, and whether the apparent behavior may depend on tail observations that deserve extra scrutiny in research and validation.

Covered in depth in the Strategy research fundamentals pillar hub.