NuclearPhynance Archive

Dear all,

In my job, I need to get FX VaR of FX Forward and international asset. I want to simulate all currency path for get VaR, like get Monte Caro VaR for Stock. For stocks, I can use choleasky method to simulate correlation random variance. But, for FX Rate, I don't know whether it follow the same method to do that ? Anybody can guide me some direction ?

yes, Cholesky works.

If you are just trying to calculate a simple one-day VAR then the most straightforward approach would be to apply Cholesky to a historical covariance matrix (I'll leave you to work out the details). Over longer periods, if you are simulating paths there is a drift adjustment to be careful about (in the risk neutral measure, all assets should have the same expected return in the valuation currency). However possibly a more important consideration in calculating a realistic VAR is that, in times of stress, "bad" correlations tend to strengthen; you may have heard people criticising the use of the Gaussian copula for this reason.



There are a number of different approaches used to calculate VAR (T-VAR etc) - it is important to bear in mind that practicality and regulatory concerns will have a large input into the decision making process.