Skip to content
Prev 2940 / 15274 Next

Cumulative Multivariate Normal Distribution

Just be aware that the output of pmvnorm is not deterministic.
Genz's method for trivariate (given in Haug) is deterministic.

David L. Reiner, PhD
Head Quant
Rho Trading Securities, LLC


-----Original Message-----
From: r-sig-finance-bounces at stat.math.ethz.ch
[mailto:r-sig-finance-bounces at stat.math.ethz.ch] On Behalf Of Krishna
Kumar
Sent: Friday, September 19, 2008 6:01 PM
To: Chiquoine, Ben
Cc: r-sig-finance
Subject: Re: [R-SIG-Finance] Cumulative Multivariate Normal Distribution

Hi Ben,

Have you looked at pmvnorm  in package mvtnorm.? This is Genz's method, 
I am working with Jan-Dash on a new approximation
for cumulative normal and I have some semi-tested code of that, we are 
currently testing it for Worst-of and other rainbow type options.
http://arxiv.org/pdf/cs.CE/0611061

The alternative to price these would be to use monte-carlo to do the 
integration and just price them using that, see for example
 the attached R code which generates multi-variate paths. You can then 
use these paths to then compute the discounted payoff function.

Hope this helps,

Best
Krishna

ps:  the drift is simply the risk-neutral drift however for currencies 
you have to factor in the quanto adjustment.
Also this piece of code was written about 4 years ago so pardon the 
coding standards.
,2),1,100,2)
Chiquoine, Ben wrote:
for