|
Predicting Stock Return Using Accounting Variables with Linear Regression Approach and Artificial Neural Network
|
| Seyyed Hossein Sajadi, Sohrab Osta, Roohollah Gheitasi |
|
|
|
Abstract: (1284 Views) |
This study Predicting the stock return by two models of linear regression and Artificial neural networks using accounting variables in Tehran Stock Exchange. Therefore, we developed a hypothesis using 140 companies` data during 1389-1384. In this study, we`ve used Pooled regression model and Ordinary Least Squares (OLS) method to investigate the linear relationship and usefulness of linear relation. We used the multilayer perceptron (MLP) based neural network by the algorithm of Back-propagation neural networks to investigate the nonlinear relationships and usefulness of these relations. We used the Adjusted R-squared, RMSE) root-mean-square error), Mean absolute error and NMSE) normal mean sum of squares of errors) to evaluate the results .The results confirm these two models in Predicting stock return, but the neural network model is the better. In other words, using accounting variables neural network could reduce forecasting errors in comparison with linear regression |
|
| Keywords: artificial neural networks, linear regression, stock return, MLP |
|
|
Full-Text [PDF 2879 kb]
(30 Downloads)
|
Type of Study: Research |
Subject:
General
Received: 2011/10/9 | Accepted: 2011/12/14 | Published: 2011/12/22
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
