关键词: ARIMA模型, 信息粒化, SVR模型, 股价指数, 投资组合优化

Abstract: For investors in the securities market, accurate prediction of future development direction and trend of asset price based on reasonable assumptions is related to investment success or failure. This paper constructs a combination prediction model based on ARIMA and information granulation SVR to predict the trend of stock market index price and income change. The empirical results show that the index prediction model based on the combination of ARIMA and information granulation SVR has a great improvement in prediction accuracy and validity compared with the traditional time series model. It can predict the price fluctuation range of risk assets , such that it can be predicted reliably in a certain time period, but the range of time series fluctuation can not be accurately predicted. In the future, it is necessary to further study other prediction models and prediction techniques in order to effectively improve the accuracy and practical guidance of stock index trend prediction.

Key words: ARIMA model, information granulation, SVR model, stock price index, portfolio optimization