Multi-objective Portfolio Optimization via Deep Reinforcement Learning based on Generalized Hyperbolic Distributio

基于广义双曲分布的深度强化学习多目标投资组合优化

100,000

This research project explores the integration of Deep Reinforcement Learning (DRL) with portfolio management, leveraging advancements in machine learning to address optimization challenges in financial decision-making. By incorporating deep neural networks and reinforcement learning models, the project aims to revolutionize portfolio optimization by overcoming the limitations of traditional mean-variance optimization methods. The methodology encompasses the use of diverse risk measurements, alternative distributions for asset returns, and higher moments of suitable distributions to propose universal solutions for portfolio optimization problems. The project also examines the application of DRL in solving multi-objective optimization challenges, utilizing various policy networks and reward functions. Additionally, the project evaluates the feasibility of adopting DRL in portfolio construction, emphasizing the potential for technological innovations to drive significant disruptions across financial markets. Overall, this research project offers new perspectives on combining AI with portfolio management, contributing to enhanced decision-making, risk management, and trading strategies in the finance industry.

Outcome: Identification of universal solutions for portfolio optimization using the DRL model. This endeavor is poised to benefit investors, scholars, and advance AI’s role in finance