
estimating the best distribution to fi t marginal data for, say, stock A or B returns, from
the issue of modeling the dependence between stock A and B returns, which can be
handled even if the return distributions of A and B are completely different. Sklar’s (1959)
fundamental theorem states that any joint distribution can be expressed in copula forms,
and vice versa, that each copula using marginal probability distributions as arguments
underlies a joint distribution. Of course, copula functions should satisfy certain proper-
ties. As an example, given a bivariate copula C(x, y) as a function of marginal probabilities
x and y of two different ...