The bootstrap methodology in statistics of extremes -- choice of optimal sample fraction (Q1409828)

The Hill estimator \(\widehat\gamma_n=n^{-1} \sum_{i=1}^K(\ln X_{n-i:n}-\ln X_{n-k:n})\) is considered for an i.i.d. sample \(X_1\),\dots, \(X_n\). Bootstrap methodologies of the optimal choice of sample proportion \(k/n\) are discussed. The techniques of bootstrapped MSE and asymptotic MSE optimization are described. As an alternative the use of auxiliary statistics is discussed. Different approaches are compared via simulation.