Time Series Model And Outliers Detection Of Shanghai Composite Index

Financial time series is a blend of the information of investors and listed companies.It contains the regularity and characteristics of economic development.Further study of financial sequence of events can analyze accurately and it is beneficial to the perfection and development of capital market theory.This paper is to predict the short-term movements of the Shanghai Composite Index as the research target and logarithmic yield as the research object.According to the nature of the sequence,it is fitted of nonlinear time series modelResearch shows that the existence of abnormal points in the GARCH model has important influence on parameter estimation and heteroskedasticity test.Some scholars have proposed some methods to mine anomalies in the GARCH model,but they are rigidly attached to the additional points.In the financial data,many outliers have the impact of follow-up,and they should be seen as innovation outliers.Therefore,it is very important to propose a method to effectively excavate two kinds of outliers.In this paper,combined with Chen and Liu(1993)proposed mining ARMA model outliers of the method,it puts forward to GARCH model digging the additional points and innovative points at the same time.It is used for the logarithmic yield series of the Shanghai Composite Index and analyzes the influence of the existence of the outliers on the parameter estimation and the heteroskedasticity test.The empirical results show that the innovation points in the financial data is widespread and cannot be ignored.Contrast with the actual results,our mining method is effective.Finally,the GARCH fitting model of the sequence after the anomaly correction and the prediction result are obtained.Then this paper clarifies the explanation and analysis of its economic significance...