wilmott.com

I started to use this simulation software recently, which is why I have some doubts about how it works. My interest is to use it for Monte Carlo simulation and them to estimate VaR. What I need to check is how can I have a correlate data or how can I introduce the correlation on the simulation through crystal methods?
Hope some one has an answer. Thank you.

Hi Dannfer

To generate corrrelated normal variables in Crystal Ball, you need to have the convariance-variance matrix. Do a cholesky factorisation of the matrix.

Then, use Crystal Ball to generate iid normal variables, and the coorelated variables are then linear multiples (from the factorisation) of these iid variables.

Hope this helps.

Thank you EricT. I will use your suggestion to the estimation and let you know the results.

I know this is an old thread, but I wrote my Master's thesis using Decisioneering's Crystal Ball, and I am curious about anyone who uses it. Anyone?Even though I have self-programmed (okay, canibalized (stolen) from other folks) Monte Carlo routines in VB and C++ in lots of utilities, I still build lots of spreadsheets with CB for lots of (academic) things. It does have a learning curve....... but I like it, I like the reporting features, and the flexibility of the interaction once you spent some time with it.Does anyone else want to comment on this utility? I'd be pleased to know other fin users comments and applications, as most criticism I am familiar with is from enginners.