<t>Sorry Vladimir, found some mistakes... Actually the underlying is a basket of two assets ( S1+S2 ), but not S1*S2. In BS, I mean the geometric Brownian Motion, corr (dS1/S1, dS2/S2) = corr (dW1, dW2) = rho dt , which is what I meant by "return correlation". But this does not make it right for Hes...