RT Journal Article SR Electronic T1 A General Revised Historical Simulation Method for Portfolio Value-at-Risk JF The Journal of Alternative Investments FD Institutional Investor Journals SP 87 OP 103 DO 10.3905/jai.2005.591581 VO 8 IS 2 A1 Chu-Hsiung Lin A1 Chang-Cheng Chang Chien A1 Sunwu Winfred Chen YR 2005 UL https://pm-research.com/content/8/2/87.abstract AB In this article, portfolio value-at-risk (VaR) is estimated using a generalized error distribution (GED) in conjunction with historical simulation. Using 12 years of daily return data on five international stock indices (S&P 500, FTSE 100, DAX, TAIEX, and Hang Seng) and a hypothetical portfolio formed by these five indices, the authors test the performance of this modified approach based on the analysis of its estimation conservativeness, accuracy, and efficiency. They find that incorporating the GED model into historical simulation method results in a substantial improvement in capturing recent market volatility, and hence in the accuracy and efficiency of portfolio VaR. However, they also find that the regulated trading price limit may influence the performance of estimation accuracy to some extent. In addition, their empirical results show that the specified GED models with tail-thickness parameter perform much better than those specified as normal distributions. These results provide evidence that fat-tailed GED models are more suitable when returns are not normally distributed.