Internal Ratings-Based modelling sounds like a regulatory acronym. In practice it is the job spec for half the junior risk roles in European banking.
Internal Ratings-Based (IRB) modelling — the regulatory framework under which European banks calculate credit-risk capital — is the day job for a significant fraction of junior risk analysts. The MCF risk-modelling sequence is built around what those analysts actually do.
The structure of IRB work
There are three model types: PD (probability of default), LGD (loss given default), and EAD (exposure at default). Each goes through a development cycle — data preparation, model fitting, validation, calibration, regulatory submission, ongoing monitoring. Junior analysts spend most of their time on data preparation, validation, and monitoring — not on the modelling itself.
What we cover
The MCF sequence walks through one full PD model end-to-end — sampling, definition of default, segmentation, modelling (logistic regression as baseline, gradient boosting as challenger), validation (calibration, discrimination, stability), and the documentation pack expected by regulators. Students leave knowing what an IRB validator looks for and how to produce a defensible model.
The sequence is taught using anonymised bank data provided by our advisory board partners — the data is real, the names are not.