• Nitin Kumar
• Vaibhav Farenjiya Khatik
• Himanshu Dubey
• Hoshang Kumar Sahu

.This study presents a Q-learning-based stock trading agent optimized for financial markets, achieving a test profit of $45,522.74, a 23.42% annualized return, and a 98.51% win rate, with a maximum drawdown of -35.84%. By integrating stop-loss and take-profit mechanisms, the agent maintains a risk-reward ratio of 70.00, balancing high returns with robust risk management. The methodology employs tabular Q-learning with a state space incorporating price movements and technical indicators, validated over 500 training episodes. Compared to prior reinforcement learning (RL) models, the agent offers superior win rates and risk-reward ratios while remaining computationally efficient. Results highlight its adaptability to volatile markets, with implications for automated trading systems. Future enhancements include dynamic thresholds and additional indicators to improve generalizability. This work contributes a scalable RL framework for algorithmic trading in volatile financial environments.

Q-Learning, Reinforcement Learning, Algorithmic Trading, Risk Management, Financial Markets

"AI-based Reinforcement Learning Stock Trading Agent", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 4, page no.n149-n152, April-2025, Available :http://www.jetir.org/papers/JETIR2504D17.pdf

"AI-based Reinforcement Learning Stock Trading Agent", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 4, page no. ppn149-n152, April-2025, Available at : http://www.jetir.org/papers/JETIR2504D17.pdf