Message-ID: <1580501.1075863444911.JavaMail.evans@thyme> Date: Thu, 20 Sep 2001 09:31:11 -0700 (PDT) From: zimin.lu@enron.com To: j.kaminski@enron.com Subject: RE: Bid-offer Algorithm Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-From: Lu, Zimin X-To: Kaminski, Vince J X-cc: X-bcc: X-Folder: \VKAMINS (Non-Privileged)\Kaminski, Vince J\Inbox X-Origin: Kaminski-V X-FileName: VKAMINS (Non-Privileged).pst Vince, Thanks. Right now we will not do anything with him yet. Zimin -----Original Message----- From: Kaminski, Vince J Sent: Thursday, September 20, 2001 10:57 AM To: Lu, Zimin Subject: RE: Bid-offer Algorithm Zimin, We can talk to him about the feasibility study. The man problem is managing too many partners and paying for what is a consulting service. A lot depends also on where the relationship with Prediction Co goes. Vince -----Original Message----- From: Lu, Zimin Sent: Wednesday, September 19, 2001 11:08 AM To: Kaminski, Vince J Cc: Lee, Bob Subject: Bid-offer Algorithm Vince, Mary Nordstorm in RAC forwarded me this message from her friend who is a consultant. I think it does not hurt us if we prob his algorithm without telling him anything. Zimin -----Original Message----- From: Nordstrom, Mary Sent: Wednesday, September 19, 2001 10:52 AM To: Lu, Zimin; Lee, Bob Subject: FW: Izzy Nelken Over the past year, I've been doing a lot of work to one of the largest Chicago market making firms in equity options. The project has to do with the "Market Microstructure". With high degree of accuracy we can predict where the stock price is going to go in the next 15 seconds to 60 seconds. This technique can NOT be used to predict where the stock will be tomorrow or even in the next hour. As these are only short term predictions, the changes in the stock are usually small. Therefore, this technique is of no use to day traders. However, suppose you need to buy stocks anyway. For example, if you need to delta hedge an option position and buy some stock, the model might tell you to wait a few minutes and save a few pennies. Every time the stock moves and the trader has to modify the position, they consult our model and decide on their action: buy now or wait. Of course, this also works for "sell now or wait". Indeed, we found out that they save about 3-4 cents per stock trade. The market makers typically trade hundreds of millions of stocks a year so this is quite interesting to them. We also have a project with one of the major US banks to apply this to foreign currency (forex) markets. At this stage, it would be interesting to adapt this project to other markets. For example crude oil comes to mind. The project involves two stages: a) A feasibility study to check if this is possible to do. b) Building a real time system to give signals to the traders. I think we can do step (a) independently without any commitment on your part. We would then apply our models and see if we can come up with anything interesting. We have access to "tick data" through a commercial service called "e-signal". I believe that it is reasonably good. If you have access to better tick data then it would be helpful for us to obtain several days (perhaps two weeks) of tick data (bids, asks and trades which are time stamped). Izzy Nelken Super Computer Consulting, Inc. 1070 Westfield Way Mundelein, IL 60060 USA (847) 837-0200 izzy@supercc.com www.supercc.com