Skip to Main Content
Adopt wavelet decomposition and reconstruction technology to implement return time series denoising process and establish stock index futures margin forecasting model. Filter the original time series and extract hidden periodicity and nonlinearity of the volatility of stock index futures return time series by wavelet decomposition, then adopt GARCH model and Genetic Neural Network model for modeling and predicting the scale sequence and wavelet sequences respectively, afterwards employ wavelet reconstruction to synthesize the predicted result of different scales. Substitute the processed data for the original return time series in the four common margin setting model and then select prudent index and opportunity cost index to make comparision. The result shows that there are vary degrees improvement of the margin prudent level by employing the wavelet denoising data for the four models which proves the availability of the constructed model.