Adaptive Q-Policy for Advanced Deep Reinforcement Learning Algorithm

  • Unique Paper ID: 180551
  • Volume: 12
  • Issue: 1
  • PageNo: 1380-1385
  • Abstract:
  • This study introduces the Adaptive Q-Policy for Trading (AQPT), an advanced Deep Reinforcement Learning (DRL) algorithm designed to optimize algorithmic trading strategies within highly stochastic and low-observability market environments. AQPT is built on traditional Q-learning frameworks. AQPT incorporates enhanced market indicators and a refined reward function that emphasizes risk-adjusted returns, specifically optimizing the Sharpe ratio to guide decision-making. This novel approach enables AQPT to adapt dynamically to market shifts, significantly improving profitability and risk management relative to baseline models. Through simulations and historical market data, AQPT was tested against traditional trading algorithms, demonstrating notable performance gains across key metrics. However, AQPT's limitations in high-frequency trading and sensitivity to extreme market volatility highlight areas for future enhancement. Ongoing research will explore the expansion of AQPT into diverse asset classes, including commodities and cryptocurrencies, and the integration of multi-agent DRL strategies to increase adaptability across varied market conditions.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 12
  • Issue: 1
  • PageNo: 1380-1385

Adaptive Q-Policy for Advanced Deep Reinforcement Learning Algorithm

Related Articles