Body and tail: an automated tail-detecting procedure

In risk management, tail risks are of crucial importance. The quality a model, which is determined by data from an unknown distribution, depends critically on the subset used to model tail. Based suitably weighted mean square error, we present method that can separate required subset. selected determine parameters model. Notably, no parameter specifications have be made apply proposed procedure. Standard goodness fit tests allow us evaluate fitted We standard distributions usually considered in finance and insurance industries. addition, for MSCI World Index, use historical identify compute quantiles assessment.