Abstract: Traditional Historical Simulation and Filtered Historical Simulation are the most used risk measurements among international commercial banks. Another filtered historical simulation approach——BHW is presented in this paper, which is constructed by Hull and White (1998) and bootstrap methods. Then the spot trading of gold price is taken as samples, and the accuracy of the BHW method and other popular methods, such as traditional historical simulation methods, BRW methods, Hull and White(1998) methods and parametrical GARCH methods are backtested. Since Dumitrescu et al. (2012) pointed out that there's none backtesting method have absolute advantage on others, and suggest that take more backtesting methods as possible, six different methods are taken. The conclusions include:(1) Compared with the other 4 popular risk measurement methods, BHW methods shows relative better accuracy; (2) In the small sample case (125 days), the advantage of BHW are significantly better than other methods, when the sample size become larger (250 days), HW, BHW and GARCH models all show relative better accuracy, and the HW approach is slightly better than other methods; (3) the accuracy of different historical simualtion methods are influenced by rolling sample size differently.

Key words: Historical Simulation, BHW, Value at Risk, gold market, backtesting