QADQN: Quantum Attention Deep Q-Network for Financial Market Prediction

Siddhant Dutta,Nouhaila Innan,Alberto Marchisio,Sadok Ben Yahia,Muhammad Shafique

2024-08-06

Abstract:Financial market prediction and optimal trading strategy development remain challenging due to market complexity and volatility. Our research in quantum finance and reinforcement learning for decision-making demonstrates the approach of quantum-classical hybrid algorithms to tackling real-world financial challenges. In this respect, we corroborate the concept with rigorous backtesting and validate the framework's performance under realistic market conditions, by including fixed transaction cost per trade. This paper introduces a Quantum Attention Deep Q-Network (QADQN) approach to address these challenges through quantum-enhanced reinforcement learning. Our QADQN architecture uses a variational quantum circuit inside a traditional deep Q-learning framework to take advantage of possible quantum advantages in decision-making. We gauge the QADQN agent's performance on historical data from major market indices, including the S&P 500. We evaluate the agent's learning process by examining its reward accumulation and the effectiveness of its experience replay mechanism. Our empirical results demonstrate the QADQN's superior performance, achieving better risk-adjusted returns with Sortino ratios of 1.28 and 1.19 for non-overlapping and overlapping test periods respectively, indicating effective downside risk management.

Quantum Physics,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

The paper aims to address the challenges in forecasting financial markets and developing optimal trading strategies. Specifically, due to the complexity and volatility of financial markets, traditional forecasting models often struggle to effectively utilize large datasets and have limitations in capturing the complex dynamics of market behavior. To tackle these issues, the research team proposes a Quantum Attention Deep Q-Network (QADQN) method that leverages the advantages of quantum computing, using enhanced reinforcement learning to address the forecasting challenges in financial markets.

The QADQN framework embeds variational quantum circuits into the traditional deep Q-learning framework to exploit the potential advantages of quantum computing in the decision-making process. The paper also rigorously backtests the framework under real market conditions, considering fixed costs for each trade. Empirical results show that the QADQN agent outperforms other benchmark methods on historical data, demonstrating excellent risk-adjusted returns, particularly during non-overlapping test periods, and shows effective downside risk management capabilities.