Creating and validating risk models [36]

Go to: Summary | Previous | Next   
Bullet points include: Risk models are typically derived from statistical analysis of observed data Characterised by a (multivariate) probability distribution and as with any statistical analysis there is uncertainty, including, in this case, in the modelled distributional form Model risk is the risk that the model is inherently mis-specified Rather than the risk that the particular outcome just coming up (or that has just occurred!) is at one extreme of the modelled distribution This distinction may be difficult to explain to non-mathematicians If there is an explicit portfolio optimisation stage and risk model is mis-specified then we are likely to end up with the wrong portfolio

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile