Out-of-sample prediction of profitability is a critical step in fundamental analysis and yet even sophisticated regression models do not generate predictions that significantly outperform random walk predictions. We employ random forests with classification trees, a method from machine learning,...