Prediction of Stock Return Using Financial Ratios: A Decision Tree Approach

The purpose of this paper is to develop a model for prediction of present and prospect stock return using financial ratios. For this purpose, decision tree method was used. In this approach, a set of logical conditions in a hierarchical algorithmic model have been used for prediction or recognition of an event. Hereupon in this research, 70 percent of data were used to produce models in four popular decision tree algorithms (CHAID, ECHAID, QUEST and CRT) and the results of the tests were compared in 30 percent of residual of data with some of performance measures like accuracy, sensitivity and specificity. Information of 317 companies accepted in Tehran Stock Exchange was used in this study. The results indicated that ECHAID and CRT algorithms performed best in the prediction of present and CHAID algorithm in that of future. It was also shown that the models were better in that of present compared to that of future. The abilities of the models, however, were not significant in both cases. Accordingly, the hypothesis of the study was rejected.