Seminar| Institute of Mathematical Sciences
Time: Thursday, January 8th, 2026,11:00-12:00
Location: IMS, RS408
Speaker: Min Dai, The Hong Kong Polytechnic University
Abstract: This paper addresses the challenge of developing optimal investment strategies in the presence of transaction costs and uncertain market conditions—an issue of critical importance for portfolio managers and financial decision-makers. We reformulate the classical continuous-time portfolio selection problem as a Dynkin game, a strategic framework that captures the timing of buy and sell decisions under market frictions. To overcome the computational difficulties posed by the discontinuous nature of stopping decisions, we introducea randomized Dynkin game approach that incorporates entropy regularization to balance exploration and exploitation. Building on this formulation, we develop an interpretable reinforcement learning algorithm capable of learning near-optimal trading policies directly from market data without requiring explicit knowledge of model parameters. Our theoretical analysis establishes convergence guarantees and quantifies the trade-offs involved in the exploration-exploitation balance. Through extensive numerical experiments and empirical tests on simulated and real market data, we demonstrate that our method effectively approximates optimal trading boundaries and outperforms benchmark strategies, offering a practical tool for dynamic portfolio management in realistic trading environments. This work bridges advanced stochastic control theory and modern machine learning, providing actionable insights for managing transaction costs and adapting to evolving market dynamics. This is a joint work with Yuchao Dong and Zhichao Lu.