i know tradestation can't do this, but what is this concept called? why can't i give tradestation stock market data and have IT tell me what stock system will give me the best results? instead of me thinking of all the possible systems, why can't it be like a calculator and do the work for me? a computer could test ideas faster than i can. why aren't there software programs out there than go over data over and over and find the best systems for you?
GG... I'm not sure if your are asking the computer to develop the strtegy that will work best, or determone which of the multitude of known strategies will work best. Although tedious, Tradestation could actually do number the latter with some easylanguage programming. The former would require an AI module , but it would still have to have a basic set of "tools" with which to work.
yes, this is what i meant. i'm aware this isn't easy and it's probably something i will never even be able to do. i just think that it would be possible for a computer to analyze stock market data and come up with the best trading system it could. we all know even our own PCs can do a lot more calculating per second than we can. so why are we sitting here staring at charts and thinking of ideas? we should give a computer a bunch of basic principles and then have it tinker with them and find the best trading system. kind of like the SETI (spelling?) program. a bunch of computers are looking at data trying to pick up patterns coming from space. why couldn't something like that be done except the data is stock market data? again, i know this is not easy....just throwing it out there.
It's a cool idea. I'd like to have that program myself. Then we could just toss in some symbols and come back later and pick up the results. Then you could have the results for each symbol ranked from best to worst. "For GE, Mr. Gecko, the highest profitability resulted from trading the 30 minute chart and buying the first bar that closes above the previous bar's high after it has achieved a 32% Fibonacci retracement. Maximum momentum was achieved on average within 5 bars, Mr. Gecko, therefore a trailing stop of 3.86% allowed for maximum excurson 89% of the time." You have to have the program insert your name... for personability and such. I mean who wants a computer program that just spits out winning methods? O, I guess that would be me. Maybe the personability bit is unnecessary.
You're looking for a data mining program if I understand what you're saying. The best one I've ever seen was created and used by a large player. It told him what system (group of indicators) was working best on specific stks. at the moment. He was a short term player from India and could get all sorts of complex programs done cheaply. Wish I knew where to find him now.
What you're asking kinda reminds me of what I've heard about Neural Network technology. The networked processors, imitating a brain, can learn in real time from their mistakes, or something like that. I guess when these tools are actually able to calculate fear and greed and take advantage, they'll be totally superior to us...but then won't stock movements always be one step ahead and find a way to evade even these systems?
i work on something like this...one of my system work on pure statistic results. there couple parameters in this system. every morning computer recalculate optimal parameters for this system, parameters, that give you best results on given period of data from past. it's like system adjust itself for best performance))))) work for me.....