Using Order Statistics to Estimate Confidence Intervals for Probabilistic Risk Measures

Abstract

Value at Risk (VaR) and related probabilistic risk measures have become ubiquitous in the realm of risk management. But any statistician knows that a point estimate is only part of what one needs to know, the other being a measure of confidence about that estimate. In this article, Dowd provides an easy way to estimate such confidence intervals using the theory of order statistics. The procedure is straightforward, nearly instantaneous to execute, and can be used with any parametric or non-parametric returns distribution.