May 24, 2024
11:00
Louvain-la-Neuve
LIDAM D.251
Julien Hambuckers
(Université de Liège)
invited by Nathan Lassance
will give a presentation on :
LASSO-type penalization methods in distributional regression models, with application to hedge funds systemic risk analysis
Abstract :
For many applications in finance, it is of interest to provide full probabilistic forecasts, which are able to assign plausibilities to each predicted outcome. Therefore, attention is shifting from conditional mean models to probabilistic distributional models capturing location, scale, shape and other aspects of the response distribution. One of the most established models for distributional regression is the generalized additive model for location, scale and shape (GAMLSS). However, in high-dimensional data set-ups, classical fitting procedures for GAMLSS often become rather unstable and methods for variable selection are desirable.
In this talk, I will present first a regularization approach for high-dimensional data set-ups in the framework of GAMLSS, introduced in Groll et al. (2019). It is designed for linear covariate effects and is based on L1-type penalties. Beyond the conventional least absolute shrinkage and selection operator (LASSO) for metric covariates, we propose also group and fused LASSO-type penalization to deal with categorical predictors. Then, I illustrate the usefulness of this approach in the problem of measuring the systemic risk of hedge funds, where a regression structure in an extreme value regression model allows to escape small-sample issues and provide insights on its dynamics over time.