[02348] Multi-day Value-at-Risk estimation by GARCH and Extreme Value Theory
Session Time & Room : 3C (Aug.23, 13:20-15:00) @E505
Type : Contributed Talk
Abstract : The conventional VaR models have been unable to predict huge losses by market prices because these underestimate the probability of extreme price fluctuations. To overcome this problem, McNeil and Frey introduced a two-step approach combining the GARCH model and EVT. In this study, we investigate the estimation of multi-day VaR based on a bootstrapping simulation approach with GARCH-EVT, as well as perform back-testing in order to evaluate its ability to provide appropriate multi-day VaR estimation.