Estimating the Value-at-Risk : A Comparative Study of the Extreme Value Theory and Transformed Kernel Density Approach

This paper reports on the results of an extensive simulation study that compares three methods of estimating the value at risk of stock portfolios. Simulating returns from a wide range of data generating processes, we found that the peak-over-threshold (POT) method outperforms the transformed kernel density and the generalized extreme value block-maxima approaches. The POT approach is capable of producing accurate estimates of the tail quantiles, and for most of the cases considered, the observed violation rates are insignificantly different from the estimated quantiles