Research On Stock Index Futures Volatility Prediction Based On Markov Switching GARCH-MIDAS Model

The stock market is an important part of the financial market,which can play the role of resource allocation,capital accumulation and reflecting the economic situation.It plays an indispensable role in the field of international finance by establishing channels for transactions and financing between capital suppliers and demanders who lack investment opportunities,thus improving the efficiency and effectiveness of the economy.The volatility of the stock market is one of the focuses that investors pay the most attention to.The volatility of the stock market is used to measure the value at risk of the stock.The lower the volatility is,the smaller the risk is and the more valuable the investment is.Moreover,the volatility of the stock market affects the direction of the economy,and then affects the development of the world finance.Therefore,the research on stock market volatility prediction has very important significance.In research of market volatility model,a large number of different types of models available,most scholars adopt GARCH model and its extended model when studying the volatility prediction of stock market,however,it has some shortcomings in some aspects.GARCH model does not take into account the problem of structural transformation,which is a common feature in the financial market.as a result,MS-GARCH model has been recognized and applied by scholars.This kind of model can well solve the problem of volatility state change caused by structural transformation and effectively improve the prediction accuracy.Secondly,GARCH models cannot make use of data with different frequencies which is often used in volatility prediction such as monthly data,daily data and even 5-minute data.Therefore,some scholars put forward GARCH-MIDAS model,which can directly fit data of different frequencies without changing frequencies,and solve the problem of poor prediction effect caused by different frequencies.In this paper,on the basis of previous studies,we bring Markov state transition process into GARCH-MIDAS model,and MS-GARCH model and GARCH-MIDAS model are combined to construct a class of MS-GARCH-MIDAS model.This kind of model includes three kinds of models.The Short-GARCH-MIDAS model which bring Markov state transition process into short-term,the Long-GARCH-MIDAS model which bring Markov state transition process into long-term,and the Full-GARCH-MIDAS model which bring full parameters into the Markov state transition process into the full parameters.the complete parameter estimation method of the model also is given.the empirical results also show that the improved model has a certain improvement in volatility forecasting ability compared with the GARCH model,and the Full-GARCH-MIDAS model's improvement is the most obvious.This paper applies the new model MS-GARCH-MIDAS model to the volatility prediction of S&P500 stock index futures.to discusses whether the new model can improve the volatility prediction ability of the stock market.The empirical study found that the in-sample parameter estimation results could well fit the S&P500 stock index futures data,and the out-of-sample prediction results passed the out-of-sample prediction MCS test too.This paper also divides stock market volatility into two parts,long-term and short-term,to describe the changes of volatility in a more detailed way.It can be seen that MS-GARCH-MIDAS model effectively improve the accuracy of stock market volatility prediction.