원문정보
초록
영어
In allusion to the prediction of the stock return rate, this paper has proposed nonlinear combined prediction method based on EMD (Empirical Mode Decomposition) and SVMs (Support Vector Machines). The method has divided stock return rate series into several components of different frequencies using EMD technology, getting three new sequences by grouping superposition of each component according to the frequency, which represents items of market volatility, major events, and trend respectively. Based on these three sequences, prediction is made by constructing different SVMs models to obtain the predicted value of each sequence. With SVMs, a combined model is built on the basis of predictive value of each sequence to obtain the final predicted value of stock return rate. Using CSI 300 Index, the validity of this method is verified, and the results show that the proposed model is better than the other models presented in this paper on forecasting CSI 300 Index.
목차
1. Introduction
2. Model Principle
2.1. Mode Decomposition
2.2. Decomposition Component Recombination
2.3. Support Vector Machine
3. EMD-SVM-SVMs
4. Data Experiment
4.1 Experimental Data and Evaluation Criteria
4.2. Experimental Results
5. Conclusion
Acknowledgements
References