I looked at this when I was a TS4 user several years ago. I think he calls it fuzzy logic. I completely understand the fuzzy part, the rest I have no clue. http://www.sirtrade.com/products.htm
bob111, i'd like to ask you what you would say to the people who claim an entry is not that important and that only about 10% of your time should go towards the entry? many people here say to keep it simple. your system does not seem simple, since you are calculating things every morning. although i could be wrong, maybe what you do is simple. anyway, knowing what you know, do you think you can trade simple? i certainly can't answer this question cuz i ain't raking in the money. basically, i'm wondering why people are telling people to keep it simple.....yet you are calculating optimal parameters on a daily basis and you said it's working for you? i'm not trying to cast doubt unto what you do, i'm just tryin to get to the bottom of if trading can be simple or not.
GG, This is an offshoot of Genetic Algorithms and Neural Nets. In a way, it is cuve fitting in that it is not hard to find a system that will fit a pretty good curve to any data. Then, doing simple mathematics, go long on valleys and go short on plateaus. FWIW, Wealth-Lab allows one to test many systems on a dataset. TradeStation has an add on from RINA that I belive allows one to do something similar, albeit with much more work. nitro
trading must me simple, the way to get in it is not))))))) example- one of parameters in my system (pair trading) is average of intraday % difference between two stocks....for how many days back? system start from N of days, and increment them, add to other parameters (just couple of them, i do keep things smple), then recalculate P/L for number of days, then increment this average, recalculate again.and so on....nothing complicated. just some calculations, on which you can spend a week with calculator)))))that what it does........it doesn't mean the system change this average every day.
I've never tried the product, so don't take this as an endorsement but: Michael Harris describes a similar idea in his book " Stock Trading Techniques Based on Price Patterns." From what I remember he has the computer go through thousands of data bars to look for patterns. A simple (not real) example might be [1](L<L[1]), [2](H >H[1]), [3](c>o). if you spot this pattern then buy on close. He sells a software product that sifts through bars of data to look for 3 to 6 bar patterns. After reading his book I wrote to Michael a few times to clarify a few ideas. He was very helpful and patient and always replied. Hope this helps
Try Omnitrader. A guy I worked with used to use it so I don't know everything, but it did what you what For every market it tested which indicators were working best for that market. Disclaimer: Yes, we are fitting to past data, not looking for a necessarily "good" system, just one that is working at the moment. Anyway, hope this helps. MUChris