Data driven partition-of-unity copulas with applications to risk management

TL;DR

This paper introduces a data-driven method for constructing partition-of-unity copulas, including Bernstein, negative binomial, and Poisson types, with applications to modeling highly asymmetric data in risk management.

Contribution

It provides a new constructive approach to fit partition-of-unity copulas to complex, asymmetric datasets, expanding their practical applicability.

Findings

Effective fitting of copulas to asymmetric data

Demonstrated applications in risk management scenarios

Enhanced flexibility of partition-of-unity copulas

Abstract

We present a constructive and self-contained approach to data driven general partition-of-unity copulas that were recently introduced in the literature. In particular, we consider Bernstein-, negative binomial and Poisson copulas and present a solution to the problem of fitting such copulas to highly asymmetric data.