Portfolio Optimization of Indian Stock Market using Reinforcement Learning and Explainable AI

This paper gives a hybrid AI system to optimize risk-adjusted portfolio for the Indian stock market with Deep Reinforcement Learning (DRL) methods and eXplainable AI (XAI). The structure relies on an individual NSE dataset with 1900+ stock symbols and DRL models, such as Proximal Policy Optimization (PPO), Advantage Actor-Critic (A2C), Soft Actor-Critic (SAC) and Deep Deterministic Policy Gradient (DDPG). It has a dynamic reallocation policy and constraints to reduce interest exposure to high-performing stocks and over-concentration. Compared to the prior performance, risk-adjusted performance is evaluated, using Sharpe Ratio, Sortino Ratio, Annualized Volatility and Maximum Drawdown in the model. Evaluation metrics indicate PPO was superior in comparison to the other models, with a Sharpe Ratio of 1.3972 and Cumulative Return of 34.04%. The framework demonstrates how DRL and XAI are effective in building transparent and robust in-vestment portfolios, specific to emerging markets such as India are concerned.