Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!rpi!batcomputer!cornell!uw-beaver!ssc-vax!bcsaic!dwunsch
From: dwunsch@bcsaic.UUCP (Don Wunsch)
Newsgroups: comp.ai.neural-nets
Subject: Re: NN in Business DSS -- Information needed!
Keywords: business, decision support, market forcasting
Message-ID: <45587@bcsaic.UUCP>
Date: 20 Apr 91 00:43:32 GMT
References: <7452.280b6cb5@abo.fi> <2854@dsac.dla.mil>
Organization: Boeing Computer Services AI Center, Seattle
Lines: 23

The upcoming proceedings of the IJCNN-91-Seattle will include an
article entitled:

A commodity trading model based on a neural network-expert system
hybrid.  

by Karl Bergerson and Don Wunsch

The article is a fairly straightforward application of backpropogation,
with the twist that the neural net is actually used as a knowledge
aquisition tool.  That is, it doesn't merely try to predict the
market at all times using historical data, but intstead learns from
an expert's input how to make a prediction during a relatively rare
window of opportunity.  This is important for a practical system,
where it doesn't matter if your predictions are ultimately usually
right, if you lose all your money before you get to that point.
It's better to miss a few chances to make some gains to get
higher confidence for the trading moves that are made.

Mr. Bergerson is the corresponding author and can be reached at
Neural Trading Company
8045 20th Ave. NW
Seattle, WA 98117-4336