The Time Series Analysis of Shanghai Composite Index

Abstract

This paper uses ARMA-GARCH model to investigate the Shanghai Composite Index. The result shows that the time series mean is an ARMA(4, 2) structure, while the ARMA model cannot explain the heteroscedasticity, which indicates that a GARCH model is needed. Different tests demonstrate that GARCH(3, 3) is the optimal option, for the residual series is white noise. Finally, we use ARMA(4,2)-GARCH(3, 3) to make some prediction, finding that the model only works within a certain period.