AI and the World's Hard Problems · Economics & social sciences · arXiv · 2026-10-26
Continuous-Time Mean-Variance Portfolio Selection Reinforcement Learning Problems for a Markov Regime-Switching System with No-Shorting Constraints
This is the first study to apply reinforcement learning to continuous-time mean-variance portfolio selection with no-shorting constraints in a Markov regime-switching market. The authors impose the constraint only on the mean of a Gaussian exploratory policy and show that, with a suitable exploration variance, its optimal mean matches the optimal policy of the original constrained problem. They then develop a model-free algorithm that learns the portfolio policy from trajectory data, and numerical experiments show it recovers the optimal no-shorting portfolio.