Embrechts et al. - Modelling Extremal Events

What This Work Is Doing¶

Embrechts, Klueppelberg, and Mikosch build a rigorous probability and statistics toolkit for rare events: regular variation, subexponentiality, extreme-value limits, insurance risk, point processes, dependence, and financial applications. For this wiki, the book is best used as the first external bridge from Taleb-centered fat-tail intuition into standard extreme value theory.

We map the book to concept pages. We do not reproduce the book’s proofs or examples.

Notation Differences¶

Source emphasis	Shared notation	Meaning
Tail $\overline F$ and integrated tails	$\bar F$	Survival function.
Extreme-value index sometimes implicit in domains	$\xi$	Shape parameter for GEV/GPD limits.
Power-tail index often written through regular variation	$\alpha=1/\xi$ when $\xi>0$	Pareto-type tail exponent.
Threshold $u$ and excess $X-u$	$u$ and $F_u$	Peaks-over-threshold notation.

The canonical symbol table is Notation.

Concept Map¶

Book topic	Current wiki page	Notes
Regular variation and slowly varying functions	Regular Variation	The shared language for Pareto-type tails.
Subexponentiality	Subexponentiality	The one-big-jump class for heavy-tailed sums.
Karamata-style moment consequences	Karamata’s Theorem, Pareto Moment Existence	The bridge from tail exponent to moment finiteness.
Block maxima	Generalized Extreme-Value Distribution	The GEV limit family for maxima.
Peaks over threshold	Pickands-Balkema-de Haan Theorem	The asymptotic justification for threshold excess models.
Threshold exceedances	Generalized Pareto Distribution	The GPD family for excesses.
Threshold diagnostics	Mean Excess Function, Tail Threshold Selection	Used before and after GPD modeling.
Tail-index estimation	Hill Estimator, Extreme-Value Index	Estimation needs threshold sensitivity, not a single magic number.
Empirical finance tails	S&P 500 Tail Diagnostics	A small applied bridge, with iid caveats.

Reading Route¶

Start with the regular-variation material before the statistical modeling chapters. In this wiki’s terms, the route is:

This order keeps notation stable: first the tail class, then the moment consequences, then the threshold approximation, then diagnostics and empirical estimation.

Topics awaiting dedicated notes¶

Tail empirical process and point-process view.
Declustering for dependent exceedances.
Return levels and expected shortfall under GPD fits.

Scope and limitations¶

We paraphrase the external source and link to concept pages rather than copying book text.
The current applied bridge is intentionally narrow: univariate, right-tail, peaks-over-threshold examples only.
Dependence and multivariate extremes remain planned, not silently absorbed into the iid examples.

References¶

Embrechts et al. (1997), Modelling Extremal Events.
Resnick (2007), Heavy-Tail Phenomena.

References¶

Embrechts, P., Klüppelberg, C., & Mikosch, T. (1997). Modelling Extremal Events. Springer. 10.1007/978-3-642-33483-2
Resnick, S. I. (2007). Heavy-Tail Phenomena: Probabilistic and Statistical Modeling. Springer. 10.1007/978-0-387-45024-7