Risk Management is a multi-step process, where the modelers must face several challenging issues.
These problems span from the theoretical definition of what properties a proper risk measure should satisfy, to the back-test of the forecast performances of the risk measures. Finally, one has to implement in a sound theoretical framework a risk decomposition algorithm, i.e. to split in a linear way (by portfolio, product, risk factor, etc) a typical non-linear function (the risk measure). Data play a relevant role in these topics, because of their size, computational efficiency, poor quality and "missing" frequency.
In the seminar the above problems will be formally defined, along with the state of the art and the inspection of real world data.
Michele Bonollo - IASON (Director) and Politecnico di MIlano (research fellow)