Sharad

As expected we didn't get time to discuss in detail what to do in Houston.  
We have discussed the subject informally, but I'm laying it out here for 
clarity, and for Vince/Stinson's information.  Here are my thoughts:

1. Exotica.  This is the top priority, as you'd rightly identified yourself.
 a. Learn how to build xlls.
 b. Catalogue functions in Houston Exotica not yet available in London.
 c. Catalogue differences between inputs to London and Houston Exotica 
functions.
 d. Produce London xll that allows current function calls to be used where 
possible, and incorporates new functions.
 e. Update documentation as appropriate.
 If you do nothing else in your two weeks, I'll be very happy with this.  
Main contacts are Paulo and Zimin.

2. Understand American Monte Carlo (AMC).  The aim is to get a feel for AMC's 
characteristics.  Before we embark on the use of AMC for real option 
valuation, I'd like to understand how it behaves for financial options.  I 
suggest something like the following:
 a. Produce AMC code for single GBM for American option.  Compare Greeks 
against those from appropriate tree methods.
 b. Two GBM spread option model, to compare against 1-d numerical integration 
method.  Although the "American" bit of AMC isn't going to be used here, it 
will be interesting to think about the bucketing issue in price/payoff space 
and, again, the Greeks.
 c. Some form of mean reverting model, preferably two factor.
 d. HJM for pure financial option valuation.
 If you get somewhere on a-d, and AMC behaves sensibly on pure financial 
models, then I'll be very keen to roll it out for real option valuations.

Vince/Stinson - since I'm expecting the worst from IT i.t.o. email support, 
would you be good enough to print this email out and hand it to Sharad?

Many thanks,
Steve