I found a very interesting idea on Jurik's website: "What's the best application of a neural net in a trading system?" "Numerous users of neural net technology have reported that using a neural net to predict the future TRADING RANGE of a price time series is likely to produce better results than most other kinds of forecasts, including future percent-price-change, future price, and future volatility. "To forecast future price range, you need to create TWO neural nets. One to forecast the HIGHEST price of the next N days (minutes, ticks) and one to forecast the LOWEST price of the next N days. These two nets estimate the future trading range." http://www.jurikres.com/faq/faq_bcel.htm#application It occurred to me that this approach might apply to TA in general. Comments? I'd especially like to hear from experienced NN users. TIA
I have a software package for sale that creates neural networks. It works so good that i want to share my wealth with others for only 3000$.
The NN at Jurik's site is up to 93% less than your asking price. I am in no way advocating the usage of NNs for trading. I simply asked a question about the best usage of TA, and whether what Jurik says about NN usage is actually valid according to real NN users here. That's all.
I have yet to hear of one trader who has successfully used NNs to create a consistantly profitable trading system. I'm sure they have some application, but to date it doesn't seem anyone has any success applying them to trading. Runningbear
Runningbear: .A solid reference to that is "The Encyclopedia of Trading Strategies" by Katz and McCormick. :Al
As far as range forecasting goes, NN will not work better then simple statistic method. I have tried it and have seen other to wrestle with it without much of a success. If you will talk to Jurik, ask him if his NN can accurately forecast time of reversal. That would be more usefull .
Better yet, ask him why he hasn't become a zillionaire from trading this himself, instead preferring to sell NNs and indicators on a crappy website (looks like it was put together by a five-year old) rather than driving a Lambo (or two) and partying with hot chikkie babes with big mammaries (with questionable morals - of course - they are more fun - yeah yeah) - whilst his Neural monkeys earn him major cashola! One can only wonder...
The hardest part of designing a useful NN is to come up with the inputs to it. Once you have figured that out you its doubtful that you even need the NN itself.