Could it be that actual profit is considerably higher than expected profit because gamblers on average stop more when they have been exceptionally unlucky? (Whereas the house always continues to play also when exceptionally unlucky)