# Extreme Value Theory

*Risk & Portfolio — Finicade finance glossary*

Extreme value theory models the tail of a distribution directly rather than fitting the whole thing and hoping the tail follows. It uses only observations beyond a high threshold, which is the honest response to the fact that a normal fit is driven by ordinary days and says nothing useful about catastrophic ones. The trade-off is data scarcity: by construction there are few extremes, so tail parameter estimates carry wide error bars.

**Also known as:** EVT, peaks over threshold, tail modelling

**Related terms:** [Fat Tails](https://finicade.com/glossary/fat-tails), [Expected Shortfall (CVaR)](https://finicade.com/glossary/expected-shortfall), [Tail Risk](https://finicade.com/glossary/tail-risk), [Value at Risk (VaR)](https://finicade.com/glossary/value-at-risk), [Kurtosis](https://finicade.com/glossary/kurtosis)

Source: https://finicade.com/glossary/extreme-value-theory