I've seen some do this by performing Cholesky factorization on the correlation matrix and some by performing it in the covariance matrix. Which way is correct?
You could do it either way. Starting with a vector of uncorrelated standard normals, using the covariance matrix is more direct since you have to factor in the individual volatilities anyway.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.