Risk Estimation and Backtesting

Abstract

The financial market always fluctuates. The most popular risk measures is Value of Risk. We introduce the four methods for VaR estimation.They are historical simulation method, filtered historical simulation, analytical solution and Monte Carlo method. In order to verify the different VaR estimation approaches, we utilize backtesting on chosen time series, which are Kupiec's unconditional coverage test and Christoffersen's conditional coverage test. In order to show the calculation steps of VaR estimation and backtesting procedures clearly, we present simplified examples in the beginning of each empirical study. Then the VaR estimation is calculated with different probability levels based on different observed periods. In the end, we find out the most accuracy model on chosen time series.