Inspired by the One World Probability Project and supported by the Bernoulli Society, the One World Extremes Seminar is an initiative to keep researchers with an interest in Extreme Value Theory (EVT) virtually connected in novel ways. It features both theoretical advances and important applications of EVT.
Graphical models for infinite measures with applications to extremes
Abstract: Statistical modelling of complex dependencies in extreme events requires meaningful sparsity structures in multivariate extremes. In this context two perspectives on conditional independence and graphical models have recently emerged: One that focuses on threshold exceedances and multivariate pareto distributions, and another that focuses on max-linear models and directed acyclic graphs. What connects these notions is the exponent measure that lies at the heart of each approach. In this work we develop a notion of conditional independence defined directly on the exponent measure (and even more generally on measures that explode at the origin) that builds a bridge between these approaches. We characterize this conditional independence in various ways through kernels and factorization of a modified density, including a Hammersley-Clifford type theorem for undirected graphical models. As opposed to the classical conditional independence, our notion is intimately connected to the support of the measure. Structural max-linear models turn out to form a Bayesian network with respect to our new form of conditional independence. Our general theory unifies and extends recent approaches to graphical modeling in the fields of extreme value analysis and Lévy processes. Our results for the corresponding undirected and directed graphical models lay the foundation for new statistical methodology in these areas.
This is joint work with Sebastian Engelke and Jevgenijs Ivanovs.
Please check your local time!
Log-in link: TBD
A Sparse Max–Fréchet Framework for Multivariate Clustered Extremes with an Application to Cryptocurrency Spillovers
Abstract: Multivariate extreme value analysis faces major challenges in high dimensions, particularly when both temporal persistence and cross-sectional dependence must be modeled. Classical approaches such as the multivariate maxima of moving maxima (M4) process are computationally intractable and produce unrealistic dependence patterns, while dynamic univariate models like the autoregressive conditional Fréchet (AcF) are limited by independence assumptions in innovations. We propose a unified econometric framework that addresses these limitations. First, we introduce a sparse variant of the M4 model (vSM4R), which provides a parsimonious factor structure for multivariate extremes. Second, we extend the AcF model to an m-dependent setting (mAcF), capturing clustered tail behavior across time. Third, we integrate these into a dynamic multivariate Fréchet framework (mAcF–vSM4R). We derive probabilistic properties and develop a composite likelihood estimator, establishing its consistency and asymptotic normality. Monte Carlo evidence supports strong finite-sample performance. An empirical illustration with cryptocurrency returns, a real-world application, shows the model’s ability to capture time-varying tail dependence and spillovers, providing empirical validation of our framework, while the main contribution is methodological.
Please click here or see the calendar below.
The calendar below has been created in UTC (Coordinated Universal Time).
The timezone displayed depends on the settings of your device. When synchronized with your google calendar, it will be your local time.
Dan Cooley (Colorado State University)
Anna Kiriliouk (Université catholique de Louvain)
Jordan Richards (University of Edinburgh)
Tiandong Wang (Fudan University)
Kirstin Strokorb (Bath University) - past organiser, now technical advice
Gilles Stupfler (ENSAI) - newsletter editor
2023 - 2025
Thomas Opitz (INRAE, Avignon)
Kate Saunders (Monash University)
Emma Simpson (University College London)
Stilian Stoev (University of Michigan)
2020 - 2023
Raphaël Huser (KAUST)
Natalia Nolde (UBC Vancouver)
Marco Oesting (University of Stuttgart)
Kirstin Strokorb (Cardiff University)
Gilles Stupfler (ENSAI)
Yizao Wang (University of Cincinnati)