medium · FRM Part 2 Market Risk

To formally support the colleague's concern, which test should be used, and what is the key statistical property that makes it distinct from the Kupiec test in this specific situation?

1. The Christoffersen independence test, which conditions on whether an exception is followed by another exception and can reject a model that passes Kupiec's because it targets the clustering of failures rather than their total count
2. The Christoffersen conditional coverage test, which is simply the sum of two separate Kupiec POF statistics and therefore cannot reject once the POF coverage component already fails to reject unconditional coverage on its own
3. The Basel traffic-light test, which classifies the model into green, yellow, or red zones based purely on the cumulative exception count observed over the full backtesting window, and would place 4 exceptions squarely in the green zone
4. The Lopez magnitude-loss function test, which penalizes the model according to the squared size of each exceedance and thus captures loss severity rather than the temporal clustering pattern that Kupiec's exception-count test entirely ignores

Sign up free to see the explanation and track your rank →