The Choice of the GARCH Family in Risk Measure And it's Application in China's Stock Market

摘要

GARCH family is a very important foundation in building VaR models. So in this article, we analysis and compare many GARCH models under different distributions, different model forms and different sample capabilities. We introduce a new method, called MQL function, to select the best model from large numbers of these models. At last, China stock market is our research object to validate and we obtain some conclusions such as: the influence which comes from model forms is little; AIC is not the best criterion in selecting models; MQL function is one of the approaches to select best VaR model.