The Research On High-frequency Data Model And Its Application In VaR

With the rising of volatility in the stock market, risk management has been the main research area of finance engineering and modern finance theory. VaR is a new method of financial risk management and, now it becomes the main risk management tool by many banks, companies and financial management organization.Scholars have studied volatility of stock market frequently with low frequency data such as year, month and day data. But it is rarely done with high frequency data. This paper tries to study some problems based on intraday data as follows: (1)what statistical information could be acquired from high frequency data in China stock market, (2)how to describe the volatility of high frequency data, (3)what difference might exist among the risk drawn from different modelsThis paper studies the statistical characters of ARCH and ACD models respectively. Furthermore, two frames dealing with regularly spaced data and irregularly spaced data are established and the risk character of China stock market is studied also. In order to learn if the calculated VaR can reflect the risk, the paper presents a back-test on a regularly spaced data sample and draws a good fit. According to these results, some suggestions about policies are show.ed. Moreover, characters of market liquidity are studied in ACD model, which is worth studying more deeply.This paper has some breakthrough in several aspects: (1) This paper has a systematical research of time series and risk measurement, and gives birth to signals which have actual value with both qualitative and quantitative analysis. (2) This paper has an empirical inference of high-frequency return and describes the microstructure characteristic. It is rather rare in domestic studies. (3) This paper presents two different frameworks namely GARCH and ACD models for regularly and irregularly sampled observations, which assures integrality as far as content concerned.