Message-ID: <29804654.1075856352380.JavaMail.evans@thyme>
Date: Wed, 27 Sep 2000 07:07:00 -0700 (PDT)
From: naveen.andrews@enron.com
To: ted.murphy@enron.com
Subject: Update: FFVols
Cc: vince.kaminski@enron.com, grant.masson@enron.com, 
	tanya.tamarchenko@enron.com, cassandra.schultz@enron.com, 
	wenyao.jia@enron.com, debbie.brackett@enron.com, 
	vladimir.gorny@enron.com
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Bcc: vince.kaminski@enron.com, grant.masson@enron.com, 
	tanya.tamarchenko@enron.com, cassandra.schultz@enron.com, 
	wenyao.jia@enron.com, debbie.brackett@enron.com, 
	vladimir.gorny@enron.com
X-From: Naveen Andrews
X-To: Ted Murphy
X-cc: Vince J Kaminski, Grant Masson, Tanya Tamarchenko, Cassandra Schultz, Wenyao Jia, Debbie R Brackett, Vladimir Gorny
X-bcc: 
X-Folder: \Vincent_Kaminski_Jun2001_2\Notes Folders\Discussion threads
X-Origin: Kaminski-V
X-FileName: vkamins.nsf

Ted,
         An update on the implementation for FFVols:
                  (1) In comparing 6 days of historical Var calculations 
(with that of the implied) for Agg-Gas, we have found that the Historical VaR 
calculations are consistently lower over this period, by roughly 17 MM.  The 
implied volatilities are much higher at this period, anticipating strong 
winter prices.
                  (2) At this time, the consensus is not to relase the 
historical implementation into production, and the official line to traders 
will be that the method is still in testing.  The historical VaR is 19.2MM 
and the implied is 37 MM for effective date of 09/25.    
                  (3) Further testing is in progress on a hybrid methodology 
(which I mentioned last week, whereby historical vols are scaled by the ratio 
of prompt to historical-prompt volatilities), to atleast  capture some 
implied/forward effects.  Tanya's analysis on a  fictitious portfolio 
indicates higher VaR numbers, but poorer backtesting in comparison to the 
historical approach.  This approach serves as an intermediate, and seems 
appropriate in periods such as the current one, wherein the historical 
numbers might be considerably lower than those of the implied.
               (4) Winston will start  testing using these hybrid vols, and 
if the results are deemed satisfactory, that will be the production 
methodology.

Of course, we will obtain all VaR numbers concurrently to serve as different 
indicators and beacons of risk.  The production number will hopefully be a 
sensible compromise of the different methods. 

Regards
Naveen