Forecasting risk and returns: CAPM model with belief functions

Abstract

This paper presents a CAPM model with a belief function approach for forecasting the Integrated Oil and Gas Company (CHK) stock and the S&P500 index. The approach composed of two steps. First, we estimate the systematic risk or the beta coefficient in the CAPM model using the maximum likelihood method. Second, to improve the forecasting performance, we incorporate the likelihood-based belief functionmethod. Likelihood-based belief functions are calculated from the historical data. The data set contains of 209 weekly returns during the period of 2010–2013. The finding shows evidence on systematic risk which is associated by the belief function derived from the distribution likelihood function given the market return. Finally, we use the method to predict the return of a particular stock.