Comparing two approaches for modelling the loss given default of credit cards: Run-off triangles vs regression

Battle Count: 50

Highly relevant for credit risk modeling and banking regulation (IFRS 9/Basel), but low relevance for algorithmic trading or market microstructure strategies.

Moderate complexity. The regression approach uses standard GLMs, but the ROT approach involves complex matrix operations, chain-ladder calculations, and handling of sparse data structures. Variable selection and dichotomisation add further complexity.

The study uses proprietary data from a large South African bank, which is not publicly available. While the methodology is detailed, exact reproduction requires access to the specific dataset.