Research On Chinese CSI300 Stock Index Futures Realized Volatility

Stock index future is an important part of financial market. It can not be avoid when we deal with risk hedging and complex investment strategies. Volatility is one of the most important parameter of stock index future. It's reaching for volatility is valuable in both theory and practice.This essay based on many foreign and domestic literature reviews. The Realized Volatility Model HAR-RV introduced under heterogeneous market hypothesis, and expand on the base model. First, isolated high-frequency fluctuations jumping by using LM test data. Whereby realized volatility component and a continuous component jump, and build HAR-RV-J model. Second, introducing Markov regime switching into the model by joint estimation of a set of coefficients between different state transition. In the empirical part, by introducing 78806 trading Data of CSI 300 stock index futures from June 2010 to June 2015, into HAR-RV model?HAR-RV-J model and MR-HAR-RV model wish Markov Switching. In the mean time, reaching the relationship between status smoothed probability, index futures returns and volatility. In the final part, this essay forecasts all models one by one, and build multiple loss function to calculate bias between forecast volatility and real volatility. To get the prediction effect of models.After these empirical research. The conclusion as below:1. CSI 300 stock index futures past realized volatility can better explain the next day realized volatility. In different view, the next day realized volatility has stronger relationship with last week realized volatility, but it is less with last day and last month realized volatility.2. By introducing the jumping test decompose volatility into discrete components and jumping components,the fitting of the model has slightly improved. It is significant in explain discrete components and jumping components by day realized volatility and month realized volatility. Also Positive and negative jumping explain is the same important as above.3. After introducing Markov regime switching, day and month jumping components of realized volatility can explain part of the future volatility. However, last week jumping components make more sense to the future volatility.It means the CSI 300 stock index futures realized volatility memory is in weekly level. disturbance obey T distribution perturbation model has a slight advantage by compare wish Normal distribution.4. All the HAR-RV model can forecast the results. But the prediction is a little bit late from actual sequence. It is worse in dramatic changes. However all the forecasting becomes after introducing Markov regime switching. In all the model, the disturbance obey T distribution Markov Switching realized volatility model with the best predictive effect.