Portfolio Multi-Objective Optimization and Its Performance Evaluation by Quantile-based Risk Measures

This paper suggested a new procedure for stock portfolio multi-objective optimization using the combination of three module contain of SPEA, ANFIS and CAPM. In order to create a more accurate forecasting model, ANFIS is applied to forecast future trend values of the Tehran stock exchange TSE indices like TSE TEPIX and TSE TEDIX using technical indicators. The SPEA is applied to select some of the features from technical indicators that these selected important features improve the performance of the prediction model. This study examines and evaluate the presented model in Tehran Stock Exchange. The sample search includes panel data for 50 top companies of Tehran Stock Exchange over a ten year period from 2007 to 2017. The robust strategies on market data and discusses performance of presented model empirically under real constraints. Finally, the portfolio that created by offered model have been evaluated by Quantile-based risk measures. From the experimental tests, we clearly observe that SPEA/ANFIS/CAPM prediction model significantly outperformed the other portfolio models

Keywords:

Multi-objective Optimization , Portfolio Management , Strength Pareto Evolutionary Algorithm , Adaptive Neuro-Fuzzy Inference System , Adaptive Neuro-Fuzzy Inference System , Capital Asset Pricing Model , Quantile-based Risk Measures