Wavelet Filtering and Support Vector Machines for Index Timing

Summary

This report summary describes an index-timing model that applies wavelet analysis to filter price series before using a support vector machine to predict direction. It compares daily and weekly data and different model settings for the Shanghai Composite and CSI 300. The summary reports stronger timing results for the Shanghai Composite than for the CSI 300, and says wavelet filtering improved prediction and strategy results across the tested cases.

The reported historical results include substantial returns after an assumed per-trade cost, but the summary does not provide enough methodology to assess validation design, signal timing, or robustness. It also cautions that frequent trading raises costs and market impact, particularly for larger capital. Prediction accuracy alone did not translate into equally strong timing returns for the CSI 300, illustrating that classification performance and tradable profitability can diverge. These findings are historical and do not establish future performance.

Key ideas

• The model filters index data with wavelet analysis before applying a support vector machine to predict direction.
• The summary reports improved predictions and strategy results after filtering across the tested indexes and sampling frequencies.
• Reported timing performance was stronger for the Shanghai Composite than for the CSI 300.
• The CSI 300 results show that directional accuracy does not necessarily produce strong trading returns.
• High turnover can increase transaction costs and market impact, especially for larger positions.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.