Teaser

Extremes shape our world. They are the floods that break dams, the heatwaves that strain cities, and the financial shocks that unsettle global markets. Capturing these rare but powerful events is one of the great challenges of modern statistics. This dissertation explores a new way forward: the sliding block maxima method. Where classical approaches often waste information or falter under real-world complexity, this method offers sharper insights and more reliable estimates.

In this work, I prove fundamental theorems that extend the mathematical foundations of the method, develop new tools to make it practical—including a universal bootstrap scheme—and even solve an open problem in probability theory to show why sliding blocks outperform their classical counterpart. Large-scale simulations and a real-world case study on precipitation extremes confirm the power of these ideas.

Link to the PhD thesis.