The Value At Risk Measure Of GEM Based On High-frequency Financial Data

Since the 21 st century, the global economy has entered a brand-new situation, high-tech innovation industry has become the core part of the economic development.The rapid rise of the high-tech industry innovation and industrialization is indispensable to economic comprehensive development momentum. In recent years, with the birth of a series of financial innovation tools and more open global financial market, financial assets risk become more and more significant and complex. Therefore, China urgently needs to quantitatively measure financial risk in financial markets. There are many ways to quantitative measure of the financial market risk and the VaR method because of its simple, practical, comprehensive features, thus being widely used in many financial institutions and regulators around the world to measure institutions facing the market risk, in order to better control and the market risk faced by the regulatory authorities. Because of this, this article will be introduced to the VaR method in the risk management system of financial market in Chinese GEM. It’s of great theoretical and practical significance.Considering of our own development needs and combin with the global economic situation. On October 30, 2009 officially launched the growth enterprise market in shenzhen stock exchange marks, after ten years of temper of the growth enterprise market finally from ideal to reality, this is a milestone in the development of Chinese securities market.The launch the growth enterprise market for our capital market and the social economic sustainable healthy development has the vital significance.First of all, the growth enterprise market for high-tech innovative enterprises in China provides a convenient financing platform, for our country to create huge high-tech company laid the foundation;Second, the growth enterprise market also provides investors with a new investment channels, in order to meet the needs of risk preference;Third, the growth enterprise market provides a very good supplement of the main market and small plate market, make a deep step of the goal of the establishment of multi-level capital market in our country.For the growth enterprise market risk rather than the main board market, there is no found in different measure accurately on the confidence level of the value at risk. In this paper, choosing the GEM composite index 5 minutes intraday data sequence during high frequency, the statistical characteristic analysis of the sequence data in detail, on the basis of according to its unique characteristics, establish proper model, and the data sequence is divided into estimated sample and forecast sample data sequence for rolling forecasts, finally, the prediction results compared with the actual value, by reviewing test out the most suitable for the quantitative measure of Chinese GEM market risk model.The empirical results show that: the 5 minutes series of RV and RRV of GEM showed significance of the characteristics of the peak, thick tail and skewness. And the series have significant volatility-clustering, no unit root, but have a long memory. There are the "W" type or "U" type days earnings significantly and the positive effect of Monday, negative effct of Thurday. ARFIMA model, ARFIMAX model and HAR model can depicting the long memory series of RV and RRV very well. Model calculation based on the volatility of the growth enterprise market risk value and carries on the review of test, the results show that based on days of high frequency data requested under low digit level performance is good, under the high score digits tend to overestimate the market risk, and the desires of the family of GARCH model based on data in the day on the contrary, in the high double-digit levels significantly outperform the digit level low, the low double-digit levels tend to underestimate the market risk. Considering the influence of the market quality information to investors are asymmetric volatility model built for both in the high double-digit levels or low double-digit levels are better than that of without considering asymmetric situation, shows that due to the limitation of short position, good news and bad news has the asymmetry influence. POT model based on the theory of the extremum in overcoming the data sequence is based on the assumption of the population distribution, only the tail to carry on the generalized Pareto distribution fitting, considering the correlation of sequence data lead to extreme value of the string, in addition to the string method after processing, the model can estimate the VaR of the gem market in China very well.