Message-ID: <6730762.1075856551738.JavaMail.evans@thyme>
Date: Wed, 30 Aug 2000 02:57:00 -0700 (PDT)
From: vince.kaminski@enron.com
To: chris@lacima.co.uk
Subject: Re: EPRM article
Cc: vince.kaminski@enron.com
Mime-Version: 1.0
Content-Type: text/plain; charset=ANSI_X3.4-1968
Content-Transfer-Encoding: 7bit
Bcc: vince.kaminski@enron.com
X-From: Vince J Kaminski
X-To: "Chris Strickland" <chris@lacima.co.uk>
X-cc: Vince J Kaminski
X-bcc: 
X-Folder: \Vincent_Kaminski_Jun2001_4\Notes Folders\'sent mail
X-Origin: Kaminski-V
X-FileName: vkamins.nsf

Chris,

This is very well written and will serve the reader well. A few comments.

1. When I think about the taxonomy of the VaR models, I typically use the 
classification based on 3 categories:

 a. variance/covariance  method
                b. historical simulation
                c. Monte Carlo simulation.

Delta approach is the way of representing a position in a nonlinear 
instrument.

2. I would define Monte Carlo as an approach based on statistical simulation 
of behavior of
all the market prices/rates, etc., and revaluation of the entire portfolio.
The revaluation may be based on an approximation (using Taylor's expansion) 
that may
 involve delta, delta/gamma, delta/gamma/omega or may be exact (based on the 
same
model that produces the mark-to-market portfolio valuations).

The main benefit of using the  MC simulation in the energy markets is the 
ability to capture
the gapping behavior of the energy markets in a straightforward way. I would 
emphasize that
there are attempts to incorporate jumps in the V/C model (I shall send you 
the references 
from home).

3. I would mention that historical simulation may break down in the markets 
that are evolving
quickly (new instruments for which we have no comparable prices,
behavior of prices may change as markets mature or de-mature).

4. For bigger portfolios, virtually all methods require some level of 
aggregation into
 atomic, elemental instruments to reduce the dimensionality of the problem.
This process may be a source of a big error.

5. The computational burden of MC can be reduced through clever preprocessing
of a portfolio that introduces no error. Many swaps with the same underlying
can be aggregated into one positions ( they are  portfolios of forwards and 
they are linear
instruments).

Please, feel free to use any comment (or none).

Vince







"Chris Strickland" <chris@lacima.co.uk> on 08/28/2000 02:58:56 PM
Please respond to "Chris Strickland" <chris@lacima.co.uk>
To: <vince.j.kaminski@enron.com>
cc:  
Subject: EPRM article



Dear Vince,
?
D you think you might be able to look at this in  the next day or so? Robin 
is after something by the end of this  week.
?
Best regards.
?
Chris.
?
 - EPRM_01_VaR.doc

