Group discussion on Choice in EVT methods
Ivette Gomes & Frederico Caeiro – Universidade de Lisboa (Portugal)

Is threshold selection still a nasty problem in EVT?
Does reduced-bias estimation help the threshold choice?
Which methodology to choose to deal with extremes in finance? Non-parametric, semi-parametric or parametric?

Group discussion summary:

Threshold selection is still a nasty problem in EVT, due to the strong dependence on the procedure under consideration, but reduced-bias estimation of parameters of extreme events can help the threshold choice.

Before choosing a threshold, it is important to check assumptions, like the second-order condition. A null second-order shape parameter needs to be adequately addressed.

It is possibly sensible to go beyond the tail, modeling all available data through mixture models.

Dependence should also be properly addressed. The extremal index estimation is indeed relevant to the choice of the threshold.

Ivette Gomes was a Full Professor at the Department of Statistics and Operations Research, Faculty of Sciences, University of Lisbon (1988-2011), being now a principal researcher at the Centre for Statistics and Applications, University of Lisbon (CEAUL). She has a PhD in Statistics (University of Sheffield, UK, 1978) and a Habilitation Degree in Applied Mathematics (UL, 1982). One of her main areas of research is Statistics of Extremes. She was a founding member of the Portuguese Statistical Society and member of several scientific Associations. She has been involved in the organization of several international conferences, including the 56th Session of ISI, 2007. Among other editorial duties, she has been chief editor of Revstat, since 2003, and associate editor of Extremes since 2007. To know more…

Frederico Caeiro is an Auxiliary Professor at the Mathematics Department of the Faculty of Science and Technology – Nova University of Lisbon and a member of the Mathematics and Application Research Center (Portugal). He has an MSc degree in Probability and Statistics (2001) and a PhD degree in Statistics (2006) from Faculty of Science – Lisbon University. His current research interests include Statistics of Extremes, Extreme Value Theory, Nonparametric Statistics and Computational Statistics Methods. To know more…