The extreme value distribution describes the behavior of the maximum or minimum values in a dataset. There are several types, but the most common is the Gumbel distribution, which models the distribution of the maximum of a sample of independent variables. It is defined by a location parameter (μ) and a scale parameter (β) and is typically right-skewed.
Extreme value distributions are widely used in risk analysis, hydrology, insurance, and structural engineering—anywhere rare, high-impact events matter. For example, engineers might use an extreme value distribution to model the maximum annual flood level of a river when designing a dam or levee.
The graph below shows a Gumbel distribution modeling maximum annual flood levels. The peak of the curve represents the most likely maximum flood height in a given year. The right-skewed shape reflects how unusually high flood levels are rare but still possible. The long tail to the right means that extremely severe floods, while uncommon, are more likely than they would be under a normal distribution. This makes the extreme value distribution especially useful in risk analysis, where the focus is on understanding and preparing for rare, high-impact events.
