Implementing new machine learning (ML) algorithms for credit default prediction is associated with better predictive performance; however, it also generates new model risks, particularly concerning the supervisory validation process. Recent industry surveys often mention that uncertainty about how...