Applied Study Of Non-Parametric Heteroscedastic Model In Shanghai And Shenzhen Stock Market

Non-parametric model have good quality in describing nonlinear features of data;GARCH model plays an important role in quantitative researching of the stock market since its appear, NARCH model based on polynomial spline estimate and GARCH model category were used for volatility researching of Shanghai and Shenzhen 300 index in this paper, besides NARCH model joined with wavelet multi-resolution analysis theory was used for volatility researching of Shanghai and Shenzhen 300 index. The major contents of the thesis are as follows:(1) Statistical fitting analysis of the differenced data of Shanghai and Shenzhen 300 index is made by AR,GARCH and EGARCH model, and find that the differenced data has the peak fat-tail distribution and the heteroskedasticity characteristics. Especially Shanghai and Shenzhen stock markets have weak leverage effect by the fitting of EGARCH model. Besides, the fitting researching indicated that the AR(2)-GARCH(1,1) model had better effect than AR(2)-EGARCH(1,1) model in differenced data of Shanghai and Shenzhen 300 index.(2) The NARCH model based on polynomial spline estimate be used for researching forecasted effect of the Shanghai and Shenzhen 300 stock index volatility, the results show that the fitting and forecasting effects of the model is preferably by empirical study. The model can be used well in data analysis of stock market.(3) Firstly, the Shanghai and Shenzhen 300 index data be decomposed and reconstructed by wavelet. Secondly, the NARCH model based on polynomial spline estimate is established for the decomposed and reconstructed data. The model is used for fitting and forecasting the Shanghai and Shenzhen 300 index data. Finally, the fitting and forecasting results are contrasted with the fitting and forecasting results of the NARCH model based on polynomial spline estimate, the results show that the model joined wavelet decomposition and reconstruction improved the prediction accuracy.