Abstract: Machine learning models for fraud detection. The method includes receiving a schedule of weights, the schedule comprising a plurality of entries, each entry comprising a transaction value weight, a transaction volume weight, and a range of intervention rates; testing the at least one machine learning model using a holdout data set, resulting in a ranked transactions data set; and evaluating the performance of the at least one machine learning model by computing the weighted harmonic mean of the ranked transactions data set using the schedule of weights.