Research on Volatility Prediction of Stock Returns Based on ARIMA Model

Abstract

This article illustrates the prediction of stock return volatilities for enhancing confidence and minimizing the risk of investors in making investments. The most suitable ARIMA model for making a forecast is selected with the help of KPSS and portmanteau test. To avoid the unpredictable influences from macroeconomic factors and to prove the validity of the volatility prediction in the real world, a training dataset and a test dataset are selected from a specific time period with minimum variations in these factors. The forecast data tells that in a period of few variations in macroeconomic factors, the selected ARIMA model could make precise predictions for the volatilities. In the period which the volatilities are strongly affected by factors such as inflation and money supply, the model could still be helpful in reducing the risk and enhancing the accuracy of predictions for the investors if they understand the way volatilities change with these factors.