To improve the detection of the economic ”danger zones” from which severe banking crises emanate, this paper introduces classification tree ensembles to the banking crisis forecasting literature. I show that their out-of-sample performance in forecasting binary banking crisis indicators...