Skip to content

FRM Part II · FRM Exam Part II · Backtesting VaR

A risk manager finds that a bank's 99% VaR had 4 exceptions in 250 days, all occurring within a single two-week period of market stress, and that exceptions were followed by further exceptions the next day. What is the most appropriate conclusion and model improvement?

Clustered exceptions indicate the independence property is violated, usually because the model reacts slowly to volatility shifts. The appropriate improvement is a more adaptive volatility estimate, such as a shorter window or exponential weighting, rather than merely accepting the acceptable total count or changing confidence level.

  1. AThe exceptions are independent, so no change is needed because the total count is acceptable
  2. BThe model likely responds too slowly to changes in volatility; consider a more adaptive volatility estimate such as a shorter window or exponential weightingCorrect
  3. CThe confidence level should be lowered to 95% so fewer exceptions occur
  4. DThe holding period should be lengthened to ten days to absorb the stress

Explanation

Clustering of exceptions signals a violation of independence, typically because the model is slow to react to rising volatility. Using a more adaptive volatility estimate, such as a shorter window or EWMA/GARCH weighting, addresses this. A total of 4 is within the expected range of about 2.5, but acceptable count does not cure clustering. Changing confidence level or horizon only changes the metric and does not fix the responsiveness.

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