Nematrian
/

Stress Testing and Back-testing for UCITS [25]

Go to: Summary | Previous | Next   
Bullet points include: By adopting a ‘meta’ probability distribution characterising plausibility of scenarios, we can in principle objectively select stress test scenarios. And by, say, using the concept of ‘relative entropy’ we can even identify which scenario (with a given level of implausibility) generates the worst outcome for our portfolio (out of all possible scenarios). See e.g. Breuer (2009) “If worst comes to worst: Systematic stress tests with discrete and other non-Normal distributions”, presentation to Quant Congress 2009. But these sorts of approaches sit uncomfortably with the lesser focus on likelihood that is typically implicit in stress testing. And may not be so helpful in identifying appropriate mitigating actions

NAVIGATION LINKS
Contents | Prev | Next | Library

Desktop view | Switch to Mobile