Application Research Of STAR Model Based On Non-parametric Smoothing

Linear model to solve practical problems has large limitation,so the nonlinear model is finding wider and wider application in all fields.the most commonly used nonlinear state transition model mainly includes the TAR model,STAR model and of MSR model,for a smooth transition regression model(STAR),common smoothing functions including logistic function and exponential function,with the gradually mature of the nonparametric smoothing technology,the technology is more and more widely applied in many fields,so this paper will choose a smooth transition autoregression model combined with nonparametric method.in a word,this article choose gaussian kernel function as a smooth function to build smooth transition autoregression(STAR)model,otherwise,the simulation results show that the model has better effect than LSTAR model and ESTAR model,this is the research significance and the innovation of this article.In empirical research,this paper choose a few stock price index datas,including the Shanghai composite index,Shanghai and shenzhen 300 index,the csi 100 index as representative,use these price index to do empirical research about STAR model.According to diagnostic test about the validity of STAR model in the empirical analysis,the residual error sequence of the model can pass the correlation test and heteroscedasticity test,that is to say,the STAR model based on nonparametric method is effective,otherwise,it plays a good role in portraying the nonlinear characteristics of the stock market.On the other hand,the root mean square error(RMSE)of the model is smaller,it shows that the prediction effect is better,according to the results of empirical analysis,the STAR model based on nonparametric smoothing has small RMSE value,which means that the prediction effect of the model is also very significant,so the model is effective on the whole,and use the model to study the nonlinear characteristics of the stock market is very effective.