Message-ID: <23584280.1075845548071.JavaMail.evans@thyme> Date: Wed, 21 Jun 2000 10:43:00 -0700 (PDT) From: ted.murphy@enron.com To: rick.buy@enron.com, john.lavorato@enron.com, jeffrey.shankman@enron.com, mark.frevert@enron.com Subject: Factor Loadings Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-From: Ted Murphy X-To: Rick Buy, John J Lavorato, Jeffrey A Shankman, Mark Frevert X-cc: X-bcc: X-Folder: \John_Lavorato_Oct2001\Notes Folders\Controls X-Origin: LAVORATO-J X-FileName: jlavora.nsf I hope that the following clears up some of the confusion regarding the above-captioned process. Factor loadings are an input to Value-at-Risk (VAR). Generally, you can consider them the determinants of relationships between curves,and points on curves based on trader inputs. These pertain to all curves not just North American Natural Gas. There is a specific program which calculates these relationships which requires a considerable amount of processing time (3 hours per run) as well as another few man-days to evaluate the results. There is no "ideal" or recommended period in which these must be updated. More frequent is "better". Given the amount of time it takes to refresh, it was "agreed" a long time ago amongst RAC, Research, IT and Operations that it would be done every 2 weeks. No commercial input was received. The roles were to be simple as they are with most roles in and around VAR and other risk management controls: Research specifies the math IT codes the program Operations runs the program RAC evaluates the results The process has been run every two weeks on the second Thursday of every month since February (as it had been prior) and RAC has rejected the results because the results did not reflect the underlying math because of the complexity of the math and the continuing addition of new products and curves. This created the need to recode the application from scratch. RAC continued to agitate to get this done, particularly knowing that the pressure on the efficacy of VAR would be questioned as we went into the volatile period of 2000 ( e-mails from Rudi Zipter available upon request). The new code was implemented last Thursday and the results were accepted resulting in an approximate 10% difference in VAR. RAC had requested that the code be re-written on 6/10/99 @ 7:44 am with a target completion date of 8/2/99. This was discussed and accepted by Phillipe Bibi on 6/17/99 in a meeting in his conference room with Rick Buy, Dan Bruce, Jonathon Le, Ted Murphy, Bill Bradford, Debbie Brackett and Rudi Zipter along with the rest of our task list. We have resolved that the process will continue as stated and will communicate to you the results. There are dozens of other processes that are important to the calculation and interpretation of VAR that need to be implemented, enhanced, improved or rewritten altogether. For example, the jump-diffusion factors for North American Power have not been refreshed in 2000. The prioirty is dependent on the level of precision required. I will provide a comprehensive list of those processes in due course. Ted