Earnings Per Share Forecast: the Combination of Artificial Neural Networks and Particle Swarm Optimization Algorithm

Subject Areas : Journal of Investment Knowledge

Dariush Forougi ¹ , Heidar Foroughnejad ² , Manochehr Mirzaei ³

1 - Associate Professor, University of Isfahan, Isfahan, Iran.
2 - Ph.D Student of Financial Management, Sama Technical and Vocational Training College, Islamic Azad University, Andisheh Branch, Andisheh, Iran. (corresponding author)
3 - M.A Department of Accounting, University of Isfahan, Isfahan, Iran.

Received: 2013-02-08 Accepted : 2013-06-15 Published : 2013-06-15

Keywords: Earning Per Share, Particle Swarm Optimization Al, Fundamental Accounting Variabl, Artificial Neural Networks,

Abstract :

Expectations about earning have significant effects on managers and investors’ decisions. Today, one of the measures that are takenin to consideration as an indicator ofcompanies’profitability is the concept of earningpershare.Also earningper share has major effectson stock price of companies. Hence, forecastingearning per shareisof great importance forbothinvestorsandmanagers. The aimof thisstudy is to modelearning pershareforecast of listed companies in Tehran Stock Exchange(TSE) by using the combination ofartificial neural networksand particle swarm optimizationalgorithmbased onunivariate andmultivariate models. To do this,the data of114 companies among the existing listed onesinTehran Stock Exchange was usedduring1380-1389(2001-2010).The results showed that univariate model with 78.5% accuracy and multivariate models with 91.7% accuracy, forecast earning per share.