Presentation DEA - MLP Neural NetworkModel in Selecting the Optimal Portfolio: Reviewing the Information Content of Accounting Criteria, Value-Based Criteria and BSC Criteria

Logical investment decisions require attention to different factors and different criteria at the same time. This goal can be achieved using various methods and algorithms. The purpose of this study is to develop an optimal stock portfolio model using a combination of data envelopment analysis methods, anomaly clustering algorithm and MLP neural networks.The statistical population of the research is the accepted companies in Tehran Stock Exchange during the period of 1386 to 1396. To create an optimal stock portfolio, all available criteria were grouped to reach the optimal stock portfolio.Then, the results were compared in different approaches based on the Sharp ratio. The results of the research indicate that using the combination of data envelopment analysis, anomaly clustering, MLP neural networks and accounting metrics in the provision of an optimal portfolio of stocks led to Increasing Sharp's ratio compared to other approaches (Risk and Efficiency, Value-Based, and Balanced Scorecard). In general, the simultaneous use of hybrid optimization techniques and comprehensive criteria derived from accounting reports can provide a more efficient basket of portfolios and more desirability for the investors.