Bootstrap for Extreme Generalized Order Statistics

In this paper we study some bootstrap properties of normalized extreme generalized order statistics under the assumption of a nondegenerate limit. The inconsistency, weak consistency and strong consistency of bootstrapping normalized extreme generalized order statistics for an appropriate choice of re-sample size are investigated when the normalizing constants are either known or unknown. A simulation study is given as an illustrative numerical example to explain how the suggested procedure is a very efficient tool for modeling extremes.