A constrained multi-period robust portfolio model with behavioral factors and an interval semi-absolute deviation

• Published in: Journal of computational and applied mathematics 2020-08, Vol.374, p.112742, Article 112742
• Main Authors: Moghadam, Mostafa Abdolahi, Ebrahimi, Seyed Babak, Rahmani, Donya
• Format: Article
• Language: English
• Subjects: Interval numbers; Multi-period portfolio; Prospect theory; Robust optimization; Semi-absolute deviation
• ISSN: 0377-0427; 1879-1778
• DOI: 10.1016/j.cam.2020.112742

In this paper, a multi-period portfolio selection model is presented considering the behavioral factors of investors. The behavioral factors are presented in the form of prospect theory with three main parameters as reference dependence, risk aversion, and diminishing sensitivity; these parameters define the behavioral features of investors that can influence their investing. The future return of assets is regarded according to the interval based on uncertainty. Besides, a criterion of interval semi-absolute deviation is presented to control the risk of investing. The influence of uncertainty exists in the optimal parameters; the final result is controlled by a robust optimization approach. In order to make this model applicable and adapting it to the conditions of the real world, we considered the cardinality constraints, entropy and the possibility of risk-free borrowing and lending with different rates. First, the model is solved by GAMS in small size to check the feasibility, then, due to the NP-hard feature of the model, we used three meta-heuristic algorithms, which are more appropriate for solving large problems, to solve the proposed model. To survey the model’s performance, we used real data from a case study. The results show that the model reflects the behavioral factors of investors in their portfolio and it also maintains the robustness level in accordance with the degree of conservatism. Besides, the sensitivity analysis is performed to show the model’s efficiency. Finally, we compared the results of different meta-heuristic algorithms and benchmark model’s performance. •A new portfolio model based on prospect theory utility function.•A new robust model for portfolio selection based on Bertsimas and Sim’s approach.•Presenting a new interval semi-absolute deviation for risk measure.•A case study in DOW30 and BSE Bharat22 index securities is utilized.