Simul-RL Portfolio Framework: Black-Scholes-Merton and Reinforcement Learning for Asset Allocation

Abstract

Asset allocation method using reinforcement learning is being actively researched. However, the existing asset allocation methods do not consider the following viewpoints in solving the asset allocation problem. First, State design without considering portfolio management and financial market characteristics. Second, Model Overfitting. Third, Model training design without considering the statistical structure of financial time series data. To solve these problems, we propose a new Reinforcement Learning asset allocation method. First, financial market state and agent state. Second, Monte Carlo simulation data are used to increase training data complexity. Third, Monte Carlo simulation data are created considering various statistical structures of financial markets. We show experimentally that our method outperforms the benchmark at several test intervals.