Prediction Of My Country's Stock Index Futures Return Volatility Based On MS Model

China's mainland began trading stock index futures from April 2010 ,whether it isused to invest, or hedging,one must face the question of forecasting volatilityscientifically, we consider Markov switching model ,which is mainly used to study thestate transition behavior of time series and is a nonlinear time series modelcharacteriz ing behavior of volatility.The model has been widely applied in foreignfinancial and securities markets and high forcasting precision. This article uses variousforms of state univariate Markov switching model to test and study regime swtiching oftreturn volatility of stock index futures market?and do the in-sample forecast.The results showed that the Chinese stock index futures market has the statetransition characteristics , the volatility of future could naturely be divided into tworegime-high volatility and low volatility and in the three state model regime would bedivided into the medium,high and low volatility, volatility, where low volatility and highvolatility has the lower expected duration relative to medime regime , averaging 1 to 2days.The market stay the medium state at most of the time that has average expectedduration of 6 to 10 days.The precision of three-state model is significantly larger thanthat of two-state model.Through in-sample slide forcast of 60 days 90 days ,we foundthat two-state model can accurately predict the success rate of Return of 57.69% ,whichmeans a higher prediction effect.MS-GARCH model of the permanent and temporary fluctuations in volatility areall divided into two states. Permanent low volatility regime has the maximum duration,temporary high volatility state has the minimum duration. That emporary fluctuationsare divided into two states means temporary shocks can either cause a huge impact onthe market or a small impact -the market could absorb such shocks.Simulation results of Autoregressive regime transition model (MS-AR) are moresmooth than the general model.In the MS-AR model, MS-AR(2) model is moreaccurate than that of MS-AR(1). By comparing them, we found second order lag factoris more important,state transition of futures' volatility mainly depends on the second lag term, so the second order lag factor often impact the stock market volatility greatly. Thispaper also made estimates of model with regime swtiching of the meanautoregression,verifying two and three-state model from one lag to three lag. and foundthat the performance of second-order model is the best model ,and performancegradually decreased with more order.At last we verified MS-TVTP model anddiscovered the stock index return volatility are not the factor impacting transitionprobability,the model accuracy is low. In-sample forcast direction with MS-AR (2) ofthree-state model has highest predictive accuracy, in which based on a sliding forcastwith 90 samples,the success rate is about 2 / 3,and two-state model has a better forecasteffect than that of three-state model.The volatility of stock index futures market inChina can be depicted with a general MS model, the general model to be able to verifythat the conversion of volatility is the innovation of this paper, while in the applicationMarkov switching model of stock index futures provides a guideline for forecastingvolatility.