Portfolio selection using hybrid algorithm of data envelopment analysis based on Kohonen neural network and Cuckoo algorithm

• Published in: Journal of information & optimization sciences Vol. 37; no. 4; pp. 549 - 567
• Main Authors: Faezy Razi, Farshad, Shadloo, Naeimeh
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 03.07.2016
• Subjects: Cuckoo algorithm; Data envelopment analysis (DEA); Kohonen neural network; Multiple-criteria decision-making; Portfolio optimization
• ISSN: 0252-2667; 2169-0103
• DOI: 10.1080/02522667.2015.1103483

In stock markets, one of the main issues in which investors are involved is how to select stocks. If the investors select stocks rationally, he can achieve efficiencies greater than the market average. It is of great importance to find techniques that help investors in selecting stocks in particular at stock exchange. Investors are always looking for a set of stocks in the financial markets which is more profitable and less risky. The objective of this study is to presents a new model for portfolio selection. For portfolio selection, this model uses a hybrid algorithm of Data Envelopment Analysis (DEA) based Kohonen neural network and cuckoo search algorithm. The data are clustered using Kohonen neural network and then the efficiency is calculated separately for each cluster using DEA model based on the nature of the output. Pareto-optimal combination of risk and efficiency is analyzed via cuckoo algorithm.