How to Do VaR Backtesting

Value at Risk (VaR) is the maximum loss not exceeded with a given confidence level alpha over a given period of time tau. Formally, VaR is given by the largest number r such that the return x smaller than r is no larger than (1-alpha), obviously, VaR is thus simply a quantile of the return distribution. After computing VaR values, the next step is to judge the performance of the models applied.
Read More... [Source: Reference-and-Education Articles from EzineArticles.com - Posted by FreeAutoBlogger]