Improved Stock Market Prediction by Combining Support Vector Machine and Empirical Mode Decomposition

Now equity returns are predictable has been called """"new fact in finance"""". in this paper, a two-stage neural network architecture constructed by combining Support Vector Machine (SVM) and Empirical Mode Decomposition (EMD) is proposed for stock market prediction. in the first stage, EMD is used to partition the whole input space into several disjoint regions. in the second stage, multiple SVMs that best fit each partitioned region are constructed by finding the most appropriate kernel function and the optimal learning parameters of SVMs, and finally through the combination of different region predictions to get the forecasting of the financial time series. We use China Stock Market Index (Shanghai Composite Index) in the experiment and find out that the proposed method achieves significantly higher prediction performance in comparison with a single SVM model.