Leveraging Cloud Computing for Stock Market Forecasting with Reinforcement Learning

Stock market volatility poses significant challenges for models aiming to consistently outperform benchmark indices, as market-specific characteristics can undermine model performance across different markets. This variability necessitates extensive retraining, which is both time-consuming and computationally expensive with traditional sequential methods. To address these issues, we employed cloud-based concurrent execution to optimize the training of five reinforcement learning algorithms for stock market forecasting. Our approach led to a 19.40% reduction in training time compared to sequential methods. Notably, during the COVID-19 pandemic, the reinforcement learning algorithms demonstrated superior performance, with the Proximal Policy Optimization (PPO) algorithm achieving a 60% increase in portfolio returns compared to traditional benchmarks. This enhanced efficiency and performance underline the benefits of leveraging concurrent execution and cloud-based resources for advanced financial modeling.