Credit risk modelling

More advanced credit risk management techniques require sophisticated statistical methods, including the development of application and behavioural prediction models. These are widely used not only in underwriting processes, where they significantly reduce the riskiness of the loan portfolio but can also significantly reduce the cost of regulatory capital or the amount of provisions under IFRS 9 through properly implemented PD, LGD, CCF and EAD models.

Our team is experienced in using classical techniques such as logistic or linear regression and survival analysis, but also more modern artificial intelligence approaches such as machine-learning, neural networks, or psychometric client scoring.

We also help clients with implementing and auditing underwriting, antifraud, and collection processes. Specifically, this includes optimising the use of external registers and big data information, risk-based pricing, and working with early warning systems. We also work with revisions of the effectiveness of recovery strategies.