Despite the widespread use of value at risk (VAR) to evaluate the risk of p
ortfolios, it has several shortcomings. Most noticeably, it does nor addres
s the risk of extreme events that occur in the tails of distributions, and
as such does not capture the financial distress implications of traders' ac
tions. We address this issue in this paper by formulating absolute and rela
tive risk measures that focus on unsecured loss by computing the risks asso
ciated with events that fall in the tails of a portfolio's value distributi
on. This measure is then normalized to obtain a relative risk measure that
facilities direct interportfolio risk comparisons. These new measures are t
hen used to rank trader or management performance. This approach is capable
of capturing the risk of nonlinear portfolios and is thus an improvement o
ver VAR. As a result it will enhance the ability of management and regulato
rs to set accurate capital requirements.