Chaotic Dynamics Analysis and Forecast of Stock Time Series

In this paper, the time series which formed by the daily closing price of the Shanghai stock composite index and the daily opening price of Huaxia Bank have been studied. The log-linear detrending (LLD) method is used to treat the data, then based on phase space reconstruction, it has been proved that the studied time series have the chaotic behavior by drawing phase diagram, calculating the characteristic parameters of time series like correlation dimension and the largest Lyapunov exponent. Finally, the Back Propagation (BP) neural network is adopted to forecast the further data of time series, and the satisfying forecast result is obtained.