Using wavelet transform and support vector regression for forecasting stock index
碩士 === 輔仁大學 === 應用統計學研究所 === 94 === Wavelet transform(WT) is a commonly adopted methodology in decomposing time series data. It has become more and more powerful due to its capability in unveiling the hidden characteristics buried in time series datasets. Due to the advantages of the generalization...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2006
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/00353093645696615293 |
| Summary: | 碩士 === 輔仁大學 === 應用統計學研究所 === 94 === Wavelet transform(WT) is a commonly adopted methodology in decomposing time series data. It has become more and more powerful due to its capability in unveiling the hidden characteristics buried in time series datasets. Due to the advantages of the generalization capability in obtaining the unique and global optimal solution, support vector regression (SVR) has been successfully applied in time series prediction, especially in the financial time series forecasting. The purpose of this paper is to use WT and SVR in financial time series forecasting. A two-stage modeling procedure using WT and SVR is proposed in financial time series forecasting in this study. The multi-resolution analysis(MRA)based on WT is firstly used to decompose the time series data into subseries. The obtained subseries are then used as the input variables of the designed SVR model. In order to verify the feasibility and outstanding forecasting capability of the proposed two-stage procedure, empirical studies are performed using the Nikkei 225 stock index and index futures. Analytic results demonstrated that the proposed approach outperforms the other commonly adopted methods.
|
|---|