RogerDearnaley comments on The Theory Behind Loss Curves