FRM Part II · FRM Exam Part II · Backtesting VaR
A risk manager notes that a 99% VaR model had 3 exceptions in 250 days, all in consecutive days during one stress week. What is the most appropriate interpretation using Christoffersen's approach?
The count is near the expected 2.5, but consecutive exceptions indicate dependence, probably because the model adapts slowly to volatility changes. The independence component of Christoffersen's test would likely reject, so the model may fail conditional coverage despite acceptable frequency.
- AFrequency is acceptable, but clustering suggests the model reacts too slowly to changing volatility, so independence may be rejectedCorrect
- BThe model is perfectly calibrated because 3 is above the expected 2.5
- CThe exceptions show the model is too conservative
- DClustering is irrelevant because only the count matters for validity
Explanation
Three exceptions is close to the expected 2.5, so frequency looks fine. Consecutive exceptions imply dependence, typically from volatility estimates that adjust slowly. The independence test is designed to detect this.
Did you get it right without looking?
One question tells you little. A timed set on Backtesting VaR shows your real accuracy, how long you take and where you lose marks.
More Backtesting VaR questions
- A risk manager backtests a 99% one-day VaR model over 250 days and wants to test both the correct number of exceptions and whether exception…
- A bank backtests its 95% one-day VaR over 500 days and records 35 exceptions. Using the normal approximation to the binomial, what is the z-…
- A bank's 95% VaR backtest shows exactly 5% exceptions over a long sample, so the Kupiec test is not rejected. However, the independence test…
- A bank's 99% one-day VaR model has 8 exceptions over 250 days in the yellow zone. Investigation shows 3 of them occurred because traders' po…
- A bank backtests its 99% one-day VaR over 500 days using hypothetical (clean) P&L, where positions are held fixed, and separately using actu…
- Over a backtest, a model produced 100 days after a non-exception day, of which 4 were exceptions, and 20 days after an exception day, of whi…