Financial Time Series Forecasting Using Wavelet Frame and Support Vector Regression
Stochastic Processes Seminar
Joint Seminar Series on Stochastic Processes and Financial Mathematics
by Vincent Dai
Abstract: Wavelet transform is a commonly used tool for signal process, it can process the signal in chaos and display clearly the characteristics hidden in the predict variable, and has been getting more and more application recent years. Support vector regression (SVR) is an artificial intelligent forecasting tool based on statistical learning theory and structural risk minimization principle. It has been attracting more and more attention for its characteristics of getting global optimum and regarding structure risk etc. This paper brings forward a two stage model combining with wavelet transform and support vector regression to predict stock prices. First stage, use the wavelet frame to decompose the predict variable to be several subseries with different scales. Second stage, construct SVR forecasting model with these subseries as input variables. The two stage model improves the forecasting accuracy of the SVR model because the information hidden in the forecasting variables will stand out through pre-processing by wavelet frame. The empirical results show that the proposed model outperforms the SVR model and random walk model, the cumulating returns from the stratagem proposed by this two-stage model is also better than from other models.
For More Information: Professor Daniel Dufresne: email@example.com