Nonlinear Analysis And WNN Based Prediction On Stock Market

With the rapid development of Chinese stock market, prognosticating and analysis are wildly used in domestic investment area. Thanks to increased approach in Chaos and Fractal theory, it shows powerful life force in researching behavior of stock price with rules of none-linear system.Modeling and prognosticating of time series in economics have became a hot spot in now days. The chaos signals are not random signals, but pesudo-random signals produced by low-dimension none-linear dynamical system. Beacause of senstivite depend on initial condition in chaos, we cannot prognosticate the behavior of long evolvement in stock chaos system.We adapted the FMH theory, which opposite to traditional EMH theory, absorbed entropy idea in information theory, reconstructed the stock price series, based calendar, and formed the stock price functions in accordance to process drived by information flow consequently. We also showed the demonstration of the function by GARCH models, and drew the conclusion that the stock price series can descirbe the character of stock market more easily and accurately. Based the work we made upon, short period behavior and long-range correlation of Shanghai and Shenzhen Stock market have been researched and explained with the R/S analysis of individual stocks and index. In the end, applied with wavelet neural network, short area prediction of stock prices have been made, and show its better effect comparatively to traditional method.At last, we prospect the potential improving ways or methods of this argumentation.