I'm going to perform a backtest on some VaR estimates (a huge sample) for a personal project. I'm wondering if the tests which are commonly used to evaluate VaR (Christoffersen, Kupiec) are in some way biased when $n\rightarrow +\infty$. In other words, I need the LR ratio of the test to be a finite number but, as in my case, if $n\rightarrow +\infty$ the test seems to give a value of 0, despite the proporption of violations is 'significantly' greater than the $\alpha$ of the VaR estimates. Does anybody have any suggestion? Thanks


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.