Dynamic Characteristics Analysis And Intelligent Forecasting Models On Stock Market

Nowadays, non-linear theories such as fractal are applied in Financial Capital System. As the stock market is a complex non-linear system, it is more important to describe the tendency exactly. At the same time, people pay more attention to the forecast of stock price. For example, grey forecast is applied to forecast stock price. However, it is much important to select proper forecast model to obtain the trend of stock price in such turbulent period.In this paper, we integrate wavelet theory, fractal theory and forecasting technology together and apply them in Stock Market. The main work and results are following:1. Theories of R/S and Multifractal are given, the two methods are applied in six IT company to analyze the tendency of the stock price. Moreover, we analyze the multifractal feature of the stock price not only in ordinary period, but also in the financial crisis.2. Based on linear regression and support vector regression, we construct two models which are wavelet-linear regression model and wavelet-SVM model. Those two models both can vanish the noise disturbance and improve the forecasting accuracy.3. The combination based on IOWA operator is elaborated. We construct a new model which combines wavelet-linear regression model and wavelet-SVM model together. Compared to either wavelet-linear regression model or wavelet-SVM model, the new model can significantly improve the forecasting accuracy. The effective of the new model is verified.