Applied Research Based On GARCH-VaR Model In Our ETF Risk Measurement

ETF is a kind of financial investment tool, which has a self-evident risk. How to avoid and predict the ETF market risk becomes an urgent problem to be solved. Traditional risk measurement methods, such as standard deviation, beta coefficient, not only applicable scope is limited, but also the adaptative financial instruments are very small. Therefore, it can not fully reflect the market risk. This paper from the angle of quantitative analysis does a research on the ETF market risk. Select the daily price of 50ETF from the release date to September 6, 2011 as the sample data. By analying the time series, we knew that the 50ETF yields have the features of fat tails, asymmetry, volatility clustering through EViews 6.0. So, the paper chosen t distribution and GED assumption to describe the feature of fat tailes better, and selected the GRACH models which can eliminate the correlation to fit the data. We can directly get the conditional variance of earch model by using the proc function of EViews 6.0. At the same time, using the calculation and numerical integration functions of inverse cumulative distribution function on MATLAB soft, we can also calculate the quantile under the t distribution and GED respectively, and then took the quantile into the VaR formula, we got the 50ETF values at risk in different distribution. Finally, this paper tested the accuracy of calculated VaR value under historical simulation method, Monte Carlo simulation method and fat tails distribution on GRACH models. The results indicated that the calculated risk value by GRACH (1,1) model on GED can reflect market risk better.