Tony; You have been the only person to find this thread so far. I was hoping it might get frequented, but I hate to advertise it, I am not strong enough technically in that area. Possibly I will put the Merced information here. I understand the project has been under way for a couple of years and the HP data might be the first solidification of information. As they progress and find out what more/less they can do than their initial hopes some fine tuning will be in order. SInce they will be running at ~900 Mhz they are getting into new territory for an integrated CPU, but I think that CRAY has been there frequency wise and even higher frequencies but not at the density of this project.
Everything is ready to leap downhill on the slightest urging, as we all are anticipating the end will come soon. Smart money has gotten into a cash position already. Some say that traders are self perpertuating and we are on our way to infinity.
Are you related to Claud Viola of 241 Pizza?
Check out this New Scientist article.
Boom to bust
By Michael Brooks
greedy traders could be to blame for fluctuations in the stock market, says a physicist who has created his own virtual market. He concludes that political change and economic pressures have far less influence on the booms and busts than the combined passions of speculators.
Guido Caldarelli of Manchester University and his colleagues have created a computer program that simulates speculators trading with the aim of increasing their own wealth. Initially, stocks are randomly priced and traders buy at random. From then on, they buy and sell simply according to recent price history. After each buying round, the trader with least capital is wiped out and replaced by a newcomer.
As the simulation progresses, a complex pattern of peaks and troughs in the price of each stock emerges. The fluctuations appear to have a "fractal" structure--the shape of the fluctuations is the same, regardless of the timescale considered. Caldarelli has calculated the Hurst exponent of this fractal pattern, a measure that characterises its dimensions. The value is 0ú62--just 0ú03 short of a value published in 1995 for real stock market data (Nature, vol 376, p 46).
"The results are quite astonishing, even to us," says Yi-Cheng Zhang, one of the team. "For the pure speculation traders, the price movements show all the familiar bull runs, crashes and mini-crashes. Our ultrasimplified model has all the major characteristics of a real market."
The traditional economic view is that the laws governing markets are virtually impenetrable--driven by the different perceptions of every trader and forecaster, and swayed by complex political, banking and industrial factors. Mike Dicks, who studies the British economy for the investment bank Lehman Brothers, says he would not expect such a simple model to reproduce all the fluctuations.
"It's surprising that using only one or two variables, they are able to explain the high-frequency changes," says Dicks. "The underlying drive comes from the fact that everyone is trying to maximise their profits, but any model has to be an enormous simplification."
Caldarelli, who will report the results in Europhysics News, says the model best simulates foreign exchange markets, which are volatile and competitive. The interactions among traders in the simulation occasionally lead to a sudden crash--a digital Black Wednesday. "The crashes arise only as a result of collective trading activity, and there is almost no warning of their coming," says Caldarelli.
The researchers hope to use a similar model to find out if a central authority can stabilise the market and avert a crash. "It would be interesting to study what happens if we introduce a player whose aim is to balance the price," says Caldarelli. "We can probably demonstrate that a central bank or authority has to exist to ensure a stable market."
From New Scientist, 18 October 1997
c Copyright New Scientist, IPC Magazines Limited 1997 |
