Model optimization aims at calibrating the best values of model design and operating policy variables – values that will lead to maxmimize the level of model performance. Combined with detailed and precise simulation methods, CT RISK will place its quantification efforts on identifying model sensitivity to errors that could result in economic or reputational losses. The efforts will exhaustively cover on the following:

• Strengthening data governance to ensure the quality, integrity, traceability and consistence of the data that feeds the models;

• Periodic model back-testing to compare expected outputs with actual outputs and deduce the degree of model accuracy;

• Academic support or market benchmark, if applicable, for the methodological decisions adopted;

• Use of alternative models to contract results;

• Development of complementary analyses that test the validity of the model results;

• Regular model follow-up like monitoring the model’s predictive power and an early warning system for deterioration;

• Pilot testing before initial disposition and after every substantial change of the model.