Research of the possibility to forecast changes in financial state of a credit organization on the basis of public financial statements

The mathematical model for forecasting of license revocation of a credit organization in the 6-month period based on public financial statements is considered. The model represents an ensemble of combinatorial and logical methods and decision trees of different types. Its effectiveness estimated by ROC AUC (area under receiver operating characteristic curve) is 0.74. The model allows distinguishing groups of credit organizations with higher and lower license revocation risks. Also, the ranking of different financial statement indicators has been performed which marked the importance of liquid and highly liquid assets.