This research paper investigates the application of Machine Learning algorithms for forecasting financial asset prices, addressing a core challenge in quantitative finance. It conducts a direct comparison between state-of-the-art ML implementations and the traditional Capital Asset Pricing Model using U.S. equities data. The study leverages high-performance computing infrastructures to train ML models on extensive datasets, including time series data for an entire stock universe and exogenous macroeconomic variables, aiming to enhance prediction accuracy.

The methodology involves implementing modern ML techniques and benchmarking them against CAPM, with a focus on out-of-sample performance. Key findings reveal that the ML models significantly outperform CAPM, demonstrating the potential of ML to handle large-scale financial data and improve forecasting outcomes. This work contributes to the field by validating the superiority of ML approaches in asset price prediction, offering insights for practitioners seeking to leverage data-driven methods in financial markets.