“…Clearly, since the two moments are sampled from the same distribution they are correlated, and thus the formalism introduced in Cramer (1962) depends on the covariance of the two moments. The approach taken in Dubi and Kolin (2016) is a bit different, and geometric by nature. Consider YðTÞ ¼ YðM 1 ðTÞ; M 2 ðTÞÞ as a general differential function of the first two moments.…”