An Empirical Research Of CSI300 Stock Index Futures

For a long time, the volatility is paid close attention in the financial research of many scholars, and the modeling and prediction about the volatility has been a core issue in financial econometrics, it has important applications in asset allocation, asset pricing, risk management etc.. The traditional volatility measure mainly by implied volatility and historical volatility index two, the former by the prices of derivatives into the pricing model to push down the calculation of volatility, mainly for the pricing of financial derivatives, rarely used to predict, while the latter is based on historical price to describe and predict the future volatility, is a common method used to return volatility of financial assets. However, GARCH model and SV model to estimate the historical volatility which is widely used by all have their own defects. In recent years, with the development of science and technology and the popularization of electronic transactions, in minutes and seconds for the acquisition of high frequency data unit becomes more and more easy, compared with the traditional low frequency data of these high-frequency data can more rapidly and effectively capture market information, market reaction of the real situation. On the basis of the Merton first proposed the concept of realized volatility has been gaining more attention.Implied volatility and historical volatility compared to the traditional, realized volatility has very obvious advantages, it does not depend on any measure model, there is no need for parameter estimation, it can be regarded as a "observable" variable, after years of development, has become the most commonly used the research on financial asset volatility tools rate. Through a lot of research at home and abroad of the realized volatility of capital markets that have realized volatility has obvious peak and fat tail, extreme right skewed and long memory(correlation coefficient is a hyperbolic rate) and other features, this is clearly not in line with the efficient market hypothesis conclusion.In this paper the heterogeneous market hypothesis is a good explanation of the realized volatility of these non random features. The hypothesis of the market traders is divided into short-term, medium-term, long-term three categories, in their risk appetite, difficulty to obtain information and rational degree exist some differences, but these differences led to differences in trading behavior. Corsi heterogeneous market hypothesis based on the HAR-RV model is proposed, which is a simple realized volatility with time series model, followed by Tian Jinfang and Zhang Xiaofei, and has carried on the improvement and development, introduced overnight volatility and asymmetric effect by HAR-L-M model.This paper selects the Shanghai and Shenzhen 300 stock index futures price data from January 2, 2014 to December 31, 2014 for five minutes, remove the non trading days and a month after the last trading day received a total of 13230 data samples, for 245 days. The mean square error of MSE under different frequency, determine the optimal frequency of CSI 300 stock index futures for 5 minutes. The statistical characteristics of the Shanghai and Shenzhen 300 stock index futures for 5 minutes of realized volatility is described, we find that it has the typical peak thick tail and right deviation, non normality, Hurst index obtained by R/S test, more than 0.5 proved to have a long memory, according to the previous literature research can also know its usual with the leverage effect, these characteristics indicate that the CSI 300 stock index futures with the heterogeneous market hypothesis, so we use the HAR-L-M model to the realized volatility modeling, and compared with the traditional HAR-RV, Found HAR- L- M model is superior to the predicted effect of HAR- RV model.