Author: Graham Laight
Date: 02:20:51 06/07/01
Go up one level in this thread
On June 06, 2001 at 21:33:11, Bruce Moreland wrote: >On June 06, 2001 at 17:50:55, Graham Laight wrote: > >>When all this has happened (say 10 - 20 years), if anyone with inside knowledge >>of a stock tries to trade on that knowledge, the signal will be picked up by >>everyone almost instantaneously, and the price will adjust to the new >>appropriate level. >> >>The rest of the time, the stock will remain flat. > >From what I understand, the software you described works by looking at the price >of the stock. There are people who believe that if you graph the price of a >stock over time, you can predict upward or downward movement. I think that the program allows you to inport more time-series variables than just the stock price itself. If it doesn't, then I'm sure that future programs will. >If enough people buy into this, what can happen is the movements occur because >they are supposed to. If the tea leaves indicate that the stock is going to go >down, people will sell it and the price will go down. > >If the stock doesn't actually move, the software can't do anything, unless not >moving is an indicator that it will go up or down. > >I think that what you are talking about is software that analyzes reality, and >based upon this analysis of reality decides what the stock price should be. As >I said, I don't think this is what that software is doing, but let's continue >with this. > >It is certainly true that if everyone believes the stock should be at 50, it >will be at 50. > >The problem is that people's beliefs change over time, because they get new >information and have bizarre emotional fits. Software may well pick up and interpret these signals more quickly than "manual" traders. >If fifty weathermen get together and agree that the day looks like it is going >to be sunny, it does not mean that the rest of the century will be sunny (or >rainy, if you are a contrarian). It means that based upon best evidence, they >believe that the weather will be so-and-so for some amount of time. > >There are lots of people who are company-weathermen, and they analyze the health >and future prospects of companies, and they issue reports, and people believe >these reports and this influences the price of the stock. This does not stop >the price of stocks from going up and down, as there are all kinds of chaotic >inputs into the stock price function. If these people change their outlook, a >rapid price change can be the result. 2 points on this: * a forthcoming significant report is likely to be signalled to the market before its release (by traders with some sort of inside knowledge of this report) * I don't deny that stock prices will still change - I just say that between (sudden) changes, they are likely to look flat relative to today's movements >These people don't quite predict the price. They predict the company's outlook, >which is a major influence upon the price of a stock. > >I think we are in about as much danger of seeing a perfect corporate health >predictor program as we are of seeing a perfect weather program. Weather is a >bitch, and I expect that corporate health + economic trends + other factors is >as much a bitch as the weather. > >It is already true that prices are influenced to a great deal by what various >experts think. I doubt that a proliferation of electronic experts will change >this much. > >However, the people who actually know anything about this aren't likely to hang >around the computer chess group. I see the future of AI trading software evolving in a similar way to the evolution of chess programs. In the early days of computer chess, the price was high and the quality was low. Gradually, those two variables have reversed - software that can play close to GM level is now quite cheap. As a consequence, Joe Public can sit at home and generate chess moves to GM standard. If he chooses to do this, the number of moves he can make in any given position is usually quite restricted. In the future, people trading in the stock market will be much more likely to make "intelligent" selections (because of the widespread availability of the software). Thus there will be less "noise" in the prices and indices. So there is, IMO, a lot of similarity between the evolution of trading s/ware and the evolution of chess computers. If I'm right, we in this group already know a lot of the future of AI trading. -g >bruce
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.