Abstract

The purpose of this study is to predict stock returns using accounting ratios and the neural network approach. In this research, the ability to predict stock returns through the use of accounting ratios has been investigated with two approaches of artificial neural networks and least squares regression. Independent variable in this study is the accounting ratio and the dependent variable is the stock return; therefore, during 8 years, accounting ratios for the cement and drug industries were collected. Research hypotheses include a basic hypothesis and two sub-hypotheses. The main hypothesis is to examine the ability of the neural network approach in predicting stock returns compared to the least squares regression at the level of the total of the companies active in the two industries. Meanwhile, sub-hypotheses examine this case at the level of the active companies in each industry.