This article deals with the Value at Risk concept as it is used in practice. We show that,like the Gaussian distribution, elliptical distributions lend themselves to simple practicalcomputations. All necessary computations are detailed for the symmetric hyperbolicdistributions. A test on real stock market and exchange rate data shows the new distributionsfit the data better and outperform equivalent estimators used in RiskMetrics.
(volume 52, number 5)