Portfolio Selection on the Jakarta Islamic Index using the UTASTAR Method and Compromise Programming

Abstract

This paper proposes a data-driven multicriteria decision analysis framework for portfolio selection by integrating the Utility Additive STAR (UTASTAR) method with compromise programming. The proposed framework seeks a Pareto-efficient compromise portfolio by simultaneously maximizing expected return, minimizing portfolio risk, and maximizing the global utility estimated by the UTASTAR model. Continuing the conventional UTASTAR approach, which relies on preference information elicited directly from decision-makers, the proposed framework constructs data-driven reference information from historical market data. The Entropy Weight Method is first employed to derive pseudo-utility values from multiple financial criteria, which are subsequently incorporated as supplementary reference information in a modified UTASTAR formulation while preserving ordinal preference consistency through ranking constraints. The estimated utility function is then integrated with expected return and portfolio risk within the Compromise Programming model to determine the portfolio allocation. The Euclidean distance metric is adopted because it provides a balanced trade-off among the conflicting objectives. Experimental results on stocks listed in the Jakarta Islamic Index 70 demonstrate that the proposed framework successfully constructs a compromise portfolio while preserving ordinal preference relations with low utility reconstruction errors. Comparisons with the classical Markowitz mean--variance model and the Equal-Weighted portfolio indicate that the proposed framework provides a competitive multicriteria alternative by incorporating utility information derived from multiple financial indicators. Furthermore, sensitivity analysis using various norms shows that different compromise metrics produce different return--risk--utility trade-offs while maintaining relatively stable portfolio allocations.