I don't understand why some people go way out of their way to make a point that t/a works. please just shut up and lets not dilute the edge. It's a zero sum game and divulging what works and what does not is going to hurt you eventually.
Ron, Yes, SOM is Self Organizing Map, a kind of non-linear clustering. It somethmes called Kohonen network. See: http://www.scholarpedia.org/article/Kohonen_network Most commercial grade data mining packages include SOMs. If I recall S-Plus and JMP do. RapidMiner, which is free open source, does. What software application are you using? Much of the stuff marekted to traders like NeuroShell and BioComp are pretty bad at creating profitable trading systems. What kind of pre-processing and feature idnetification are you using for you NN? Many bad NNs are the result of errors in NN design or faulty pre-processing. Jerry
It's a double edged sword. Sharing information creates the opportunity to develop new ideas and methods, as is done in academic circles. Nobody is going to share all details and methods of something that actually works in real life. There are folks on some other ET threads who appear to be doing this, but I suspect they are mostly delusional, have other motives or just want attention. To your point, the other side is that if enough people used the same system it would no longer work. However that would likley need a lot of folks using exactly the same system in the same way at the same time in the same market. That isn't likely with human nature being what it is. Jerry
Is it so hard to believe that some of us are actually successful at implementing trading systems? Just because you don't posses the know-how, does not mean that others are also incapable. Face the reality; not everyone on this forums are losers. Portray that.
It's usually the ones who have failed at it and gone back to watching cute colored lines on a trading screen all day that have a chip on their shoulder.
TS equity curve. Nice curve fit. I can plug a macd, sma, and stochk together, and wait two weeks for all of the functional values to be found on 1 security. That's the kind of stuff that it takes.
Good luck with that..... I use no technical analysis of price in the determination of the entries and exits in my models. Here are some quotes to think about: "There is nothing new on Wall Street or in stock speculation. What has happened in the past will happen again, and again, and again. This is because human nature does not change, and it is human emotion that always gets in the way of human intelligence. Of this I am sure." --Jesse Livermore "The financial markets are naturally set up to take advantage of and prey upon human nature. As a result, markets initiate major intraday and swing moves with as few traders participating as possible. A trader who does not understand how this works is destined to lose money." --John Carter, Financial Trader "Fear and greed drive markets. That's why to succeed as a trader, you must learn to respect the two principal driving forces of price. To win consistently, you must put the odds in your favor by understanding when sentiment reaches an extreme at either end of the scale and take advantage of the markets at that time." --Boris Schlossberg, author of Millionaire Traders: How Everyday People Beat Wall Street at its Own Game
probably, it never intended for the wide spectrum of instruments, because after trying it on DIA, for example, we got the foll'g result: Testing Results -------------- 1) SuperBands, with Linear Regression Analysis 1500 Bars of Price Data were used in testing DIA. Loss $-928.74 Total Trades 19 2) Buy and Hold Results Loss $-171.88
So are we talking apples and oranges here or is there any objective reality to SuperBands? One party says it profitable while test resutls posted indicate otherwise.