>Both of you are right in the fact that this can't be an easy thing to do and can be quite unpreditictable. The concept I'm going after is to create a "game" out of the chaos.
>
>If I can get the basic structure so that it can constantly get updated prices, then comes the logic that tracks price changes and decides to buy or sell.
>
>Several of my friends are going to create their own logic for the program to tell it when to buy or sell. We'll track these programs for several weeks and compare our results. The idea is that we might stumble (I doubt it!) upon a simple system that would work over a large period of time. The testing would probably take year.
>
>This is just a test and kind of a crazy contest for us to try. I'm sure the results will be that it is truly impossible. Although if we do figure something out, we would use it ourselves. Marketing it for others would probably make it useless.
>
>What do you think?


I've always kind of wondered if there were "trends" (given the chaotic nature of the market) that individual stocks generally followed. For example, does stock X generally increase at least n% before any downward change. The nice thing about a computer handling the tracking is that you put in the buy/sell prices you're "guessing" at (based on this %) and let it cook over a few weeks. It could handle calculating the profit made (minus fees, etc) and give you a net change at the end of the testing period. Obviously, there are a bunch of other parameters you'd need to set and track, but you get the idea. The next step would a system that would use feedback to adjust all these parameters automatically.

It should be kind of interesting to see what you can make of it. The other thing that really interests me is neural networks and the area of emergent behavior. I think the stock market would make an interesting test of this technology. I'd be surprised if it hasn't already been applied here, considering CA's Neugents product that uses this type of technology to predict network failures and trends.