Forecasting stock prices using sentiment information in annual reports - A neural network and support vector regression approach

Abstract

Stock price forecasting has been mostly realized using quantitative information. However, recent studies have demonstrated that sentiment information hidden in corporate annual reports can be successfully used to predict short-run stock price returns. Soft computing methods, like neural networks and support vector regression, have shown promising results in the forecasting of stock price due to their ability to model complex non-linear systems. In this paper, we apply several neural networks and ?-support vector regression models to predict the yearly change in the stock price of U.S. firms. We demonstrate that neural networks and ?-support vector regression perform better than linear regression models especially when using the sentiment information. The change in the sentiment of annual reports seems to be an important determinant of long-run stock price change. Concretely, the negative and uncertainty categories of terms were the key factors of the stock price return. Profitability and technical analysis ratios have significant effect on the long-run return, too.