Comparing VaR based on different covariance matrices

Five ways on comparing covariance matrix are applied to the Shanghai 50 Indexes Stock Exchange, which are sample covariance matrix, single exponential matrix, constant correlation matrix, scalar matrix, two-parameter covariance matrix. We adopt principal components method and Markowitz portfolio method to measure stock market risk using VaR, getting the effect of measuring market risk. The result indicates that sample covariance matrix and two-parameter covariance matrix could measure market risk more effectively. The VaR of Markwoitz portfolio method has a great bias from actual loss.