Portfolio Selection

This research area investigates how multi-criteria and multi-objective decision-making frameworks, particularly Multi-Attribute Decision-Making (MADM) and Multi-Criteria Decision-Making (MCDM) methods, can be used to enhance portfolio selection. It addresses the inherent uncertainty of financial markets and the multifaceted criteria investors consider beyond risk and return. Incorporating diverse financial, economic, and sustainability dimensions into portfolio selection models enables more realistic decision processes that reflect investor preferences, market dynamics, and uncertainty.

This theme explores the use of alternative downside risk metrics, such as semi-variance and lower partial moments, and the integration of fundamental financial indicators or market multiples (e.g. book-to-market, earnings-to-market) into portfolio choice models. These enhancements aim to better capture investor risk perception, value investment opportunities based on company fundamentals, and improve portfolio resilience during market crises or downturns. The research demonstrates that traditional variance-based measures and return-only criteria may neglect important risk aspects and fundamental valuation insights critical to practical portfolio management.

Given the combinatorial and NP-hard nature of portfolio selection, recent research focuses on leveraging metaheuristics, hybrid approaches, and machine learning (ML) techniques like particle swarm optimization (PSO), genetic algorithms (GA), and neural networks to obtain high-quality solutions efficiently. These methodologies can handle complex constraints, transaction costs, multi-period rebalancing, and non-convex objective functions that challenge traditional optimization. By integrating predictive models such as LSTM for return forecasting and clustering for portfolio construction, these approaches offer both improved portfolio returns and computational tractability.