This note gives an easily verified necessary and sufficient condition for one probability forecaster to empirically outperform another one in terms of all strictly proper scoring rules.
Bitte sende mir diesen Fachartikel als PDF-Datei an meine E-Mail Adresse.