Portfolio Optimization with Cardinality Constraints via Random-Key Optimization (RKO)

This work investigates the application of the Random-Key Optimization (RKO) framework to the Cardinality-Constrained Portfolio Optimization Problem (CCPO), incorporating a Selection-Allocation decoder to decouple the decisions of asset inclusion and capital distribution. As a computational benchmark, we employ the commercial solver Gurobi, focusing on performance comparison in terms of solution time. The experiments were conducted on instances from the OR-Library with varying dimensions. The results indicate that the proposed RKO-based approach is capable of obtaining high-quality solutions in shorter computational time, consistently outperforming Gurobi in some of the analyzed instances. These findings highlight the potential of the proposed strategy as an efficient alternative for large-scale portfolio optimization problems.