SUPERCOMPUTER MATCHMAKING FOR STOCK TRADERS?
SELLING A HUGE BLOCK OF stock at a good price isn't easy. Buyers get scared, so the price dips. You can hide your intentions by breaking the block into smaller sales, but that increases transaction costs, and word tends to get around. It's the same problem in reverse for buyers of big blocks.
Wall Street veteran William A. Lupien says supercomputers will come to the rescue. His company, OptiMark Technologies Inc. of Durango, Colo., has worked with IBM to develop what amounts to a discreet computer dating service for buyers and sellers of stocks. He says private, computer matching should stimulate more trading and produce fairer prices for both parties to a transaction--while taking billions of dollars a year away from players who exploit the price gaps in today's market. Brokers would get their regular commissions.
Here's how it should work: Players prepare a profile that says what they would consider a fair price per share for various quantities of a stock. For instance, someone needing a big block of IBM might pay slightly more per share to acquire it all at once rather than in dribs and drabs. The supercomputer compares the profiles of buyers and sellers, figures out where their preferences overlap, and then ranks all possible trades between all the buyers and sellers in order of how much ''mutual satisfaction'' would be created.
That takes a whole lot of arithmetic. For the Pacific Stock Exchange, which hopes to use the OptiMark system next year, IBM figures it will need an SP supercomputer with 20 to 30 processors built on the PowerPC chip.
EDITED BY PETER COY
Updated June 14, 1997 by bwwebmaster
Copyright 1996, Bloomberg L.P.