Using a comprehensive sample of 2585 bankruptcies from 1990 to 2019, we benchmark the performance of various machine learning models in predicting the financial distress of publicly traded US firms and we find that gradient-boosted trees outperform the other models in one-year-ahead forecasts. Permutation tests show that excess stock returns, idiosyncratic risk and relative size are the most important variables for predictions. Textual features derived from corporate filings do not materially improve performance. In a credit competition model that accounts for the asymmetric cost of default misclassification, the random survival forest is able to capture large dollar profits.