Volatility Forecast Based on Intelligent Egarch Error Correction Model

Original scientific paper As the stock market volatility is highly nonlinear, coupling and time varying, it is difficult to predict by the traditional forecasting methods. For explaining the existing problems of the current volatility forecasting method, we use the model based on the weighted least squares support vector regression (WLSSVR) method to predict the stock index volatility in this paper. After the prediction, there is the error sequence that is a random time series. Therefore, this paper proposes the use of EGRACH model to construct an error forecast model based on the returns of stock predicted error time series. Then, we use these results to correct the volatility of stock. Finally, we use the volatility of Shanghai Composite Index as the application object. The experimental results show that the prediction accuracy of this method has improved significantly with regard to other forecasting methods.