Portfolio optimization is a key issue in financial risk management. Recently, the Operations and Application Service Department proposed a generative evolutionary framework for portfolio selection aimed at addressing the generality limitations of traditional methods and the challenges in model design and training associated with learning-enabled approaches. The experimental results demonstrate that the framework can achieve an excellent Pareto solution set with outstanding diversity and convergence, providing decision support for portfolio optimization in real investment scenarios.

This research has been accepted by IEEE Congress on Evolutionary Computation 2024. The first author is Assistant Researcher Chen Li from the Operations and Application Service Department.

This research is supported by the Strategic Priority Research Program of Chinese Academy of Sciences, Grant No.XDB0500100.