An extreme value distribution is a curve that does for abnormal values what the normal, Gaussian bell curve does for run-of-the-mill values. A theory to predict extreme events first appeared in 1928, gained its first real traction 20 years later, and Extreme Value Theory (EVT) has recently become more and more mainstream for insurance companies, financial services companies, and governments that have to decide how much to invest to protect against a calamity. EVT came to my attention in Robert Matthews, 25 Big Ideas: The Science That’s Changing Our World (MJF Books 2005) at 103.
The theory draws on data from the past. For a law department, or a group of them in the same industry, EVT might look back at the number and frequency of law suits that cost more than $3 million in a year, in present dollars, to defend. It appears to be a sophisticated mathematics that could project the likelihood of such a budget-crusher in the future.