This article conducts a comparative analysis of machine learning methods for empirical asset pricing, focusing on measuring asset risk premiums. The authors demonstrate significant economic gains for investors using machine learning forecasts, with some methods doubling the performance of leading regression-based strategies. They identify trees and neural networks as the most effective methods, attributing their success to the ability to capture nonlinear predictor interactions. All methods agree on a set of dominant predictive signals, including variations on momentum, liquidity, and volatility. The study highlights the potential of machine learning in improving the measurement of risk premiums and provides a benchmark for future research in this area. The authors also discuss the limitations of traditional empirical methods and the advantages of machine learning in handling high-dimensional data and complex functional forms.This article conducts a comparative analysis of machine learning methods for empirical asset pricing, focusing on measuring asset risk premiums. The authors demonstrate significant economic gains for investors using machine learning forecasts, with some methods doubling the performance of leading regression-based strategies. They identify trees and neural networks as the most effective methods, attributing their success to the ability to capture nonlinear predictor interactions. All methods agree on a set of dominant predictive signals, including variations on momentum, liquidity, and volatility. The study highlights the potential of machine learning in improving the measurement of risk premiums and provides a benchmark for future research in this area. The authors also discuss the limitations of traditional empirical methods and the advantages of machine learning in handling high-dimensional data and complex functional forms.