Skewness, Kurtosis, and Fat Tail Risk

Mean and variance tell you the center and the noise. Skewness and kurtosis tell you the danger.

Two return series can share the same mean and standard deviation yet behave completely differently in a crash. The difference lives in the shape of the distribution — measured by skewness and kurtosis.

Skewness: which side is the long tail?

Skewness measures asymmetry:

Skew = (1 ÷ n) · Σ( (xi − μ) ÷ σ )³

• Negative skew → long left tail. Large losses happen more often than large gains. This is typical of equity indices.
• Positive skew → long right tail. Occasional huge winners, frequent small losers. Seen in long-option or trend-following strategies.

A market with skew of −1.0 looks "normal" most days but hides a tendency to gap down. Sizing that ignores skew will be blindsided.

Kurtosis: how fat are the tails?

Kurtosis measures tail weight:

Kurtosis = (1 ÷ n) · Σ( (xi − μ) ÷ σ )⁴  − 3

The "−3" makes the normal distribution equal zero (excess kurtosis).

• Excess kurtosis > 0 → leptokurtic: fatter tails and a sharper peak than normal
• Excess kurtosis = 0 → normal
• Excess kurtosis < 0 → platykurtic: thinner tails (rare in markets)

Real index returns routinely show excess kurtosis of 5–20+. This is the quantitative signature of fat tails: more quiet days and more crashes than the bell curve predicts.

Why this matters for traders

• Stop placement: negative-skew markets reward tight stops that you may never hit — until they do
• Option pricing: out-of-the-money puts are systematically underpriced by Black-Scholes precisely because of negative skew and excess kurtosis
• Position sizing: a fat-tailed market makes kelly-style sizing lethal — occasional 8σ losses wipe leveraged accounts

Reading the numbers

Reading	Interpretation
Skew near 0, kurtosis near 0	Bell-curve model roughly safe
Skew < −0.5	Expect asymmetry, favor protection
Excess kurtosis > 3	Fat tails — size down, model tails separately

How to use shape

1. Compute skew and excess kurtosis on your trade returns and underlying returns
2. Negative skew + high kurtosis = the most dangerous combination for longs
3. Replace normal-based risk limits with historical or t-distribution limits when kurtosis is high
4. Stress-test stops and exits against the worst observed tail, not the 3σ theoretical one

Summary

Skewness reveals which side the rare events fall on. Kurtosis reveals how rare those events really are. Both are invisible to mean and variance — and both are exactly what determines whether your strategy survives its first crash.