Hybrid Model for Anticipating Shanghai Composite Stock Price Index(SHCOMP) Underpinned by PSO-SVR and ARIMA

Abstract

Reliable anticipation of the SHCOMP is of great significance in investor behavior. However, forecasting the SHCOMP is quite difficult. To minimize the imprecision in stock price index estimation, this study first proposes an unprecedented hybrid forecasting model which adopts SVR machine forecast models and ARIMA models. Then, through ARIMA modeling, this hybrid model built on the SVR is ameliorated. With ARIMA, the deviation of PSO-SVR is offset and the projection of stock price index was got. It is found that the stock price index projected by this model is more precise than that by other single models. The study also verifies the accuracy and feasibility of the stock price index anticipation by the combined model and testifies the consistency of the predictive advantage of the model in nonlinear space using PSO-SVR and in linear space using ARIMA.