Message-ID: <3584545.1075857080228.JavaMail.evans@thyme>
Date: Thu, 1 Jun 2000 03:21:00 -0700 (PDT)
From: john.arnold@enron.com
To: vince.kaminski@enron.com
Subject: Re: VaR
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-From: John Arnold
X-To: Vince J Kaminski
X-cc: 
X-bcc: 
X-Folder: \Vincent_Kaminski_Jun2001_9\Notes Folders\Var
X-Origin: Kaminski-V
X-FileName: vkamins.nsf

Let's meet at 4:00.




Vince J Kaminski
06/01/2000 09:19 AM
To: John Arnold/HOU/ECT@ECT
cc: Vince J Kaminski/HOU/ECT@ECT, Tanya Tamarchenko/HOU/ECT@ECT, Jim 
Schwieger/HOU/ECT@ECT, Jeffrey A Shankman/HOU/ECT@ECT 
Subject: VaR

John,

We have  been working for the last few days on VaR related issues.
The focus is on Jim Schwieger's  storage book as of 5/25 and 5/26
where we had some counterintuitive results. This book is a good
candidate for a systematic review of the VaR process.

It seems that the problem arises from forward - forward vols used by the VaR 
system. You can see in the attached spreadsheet that the VaR, on a cumulative 
basis,
jumps on Jan 04, when an abnormal FF vol hits a relatively large position.
This FF vol is also much different from the previous day number producing a 
big
jump in VaR.
This row (Jan 04) is in magenta font in the attached spreadsheet. Please, look
at column D.

The abnormal FF vol may result from one of the two factors:

 a. a bug in the code. We are working with the person in IT who wrote the
                     code to review it.

 b. a poorly conditioned forward vol curve ( a kink or discontinuity in
                    the fwd vol curve will do it). One solution I can 
propose, is to develop for
                    the traders a fwd-fwd vol generator  allowing them to 
review the fwd vol curve
                    before it is posted. If it produces a weird fwd-fwd vol,  
it can be smoothed.

Can you meet at 4 p.m. to review our findings?


Vince