What is a Fat Tail?
A fat tail describes a probability distribution in which extreme values occur more frequently than a normal (bell curve) distribution would predict. In a normal distribution, events beyond 3 standard deviations should happen about 0.3% of the time. In financial markets, however, daily moves exceeding 3 standard deviations occur roughly 1 to 2% of the time, approximately 5 times more often than the normal model suggests. This excess kurtosis is a defining feature of asset return distributions.
Impact on Risk Management
Fat tails mean that risk models based on normal distributions systematically underestimate the likelihood of large losses. A portfolio manager using Gaussian VaR might calculate a 1% chance of losing more than $2 million, when the true probability under a fat-tailed distribution could be 3% or higher. Student's t-distribution with 4 to 6 degrees of freedom, the Generalized Pareto Distribution, and stable distributions are alternatives that better capture tail behavior. Stress testing with historical worst-case scenarios provides a practical complement to statistical models.
Key Considerations
Fat tails are not a flaw in markets but a fundamental characteristic of complex systems with feedback loops, herding behavior, and leverage. Investors who ignore fat tails tend to overallocate to risky assets and use excessive leverage, leaving them vulnerable to rare but devastating drawdowns. Position sizing should account for the possibility of losses 2 to 3 times larger than what normal-distribution models suggest, and leverage ratios should include a margin of safety for tail events.