Better NN EA development - page 9
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hi
I like the spirit of contributions on this thread.
A few people want the information others are working on ,but they do not want to share their own resources.
Everybody should really be posting their ideas and then asking for other people to hand it to them.
It seems a few greedy people want to develop and sell the expert advisor for commercial exploitation ,the main reason for not sharing their work on this thread,whereas the others are freely sharing their thoughts
After all this is a free sharing forum and everybody who is making free contribution is expecting ,other people to share their work on this thread.Is it fair to free contributors?
OILFXPRO
Orders distributed in time
Barnix
Is this u?
Participant barnix - Automated Trading Championship 2006
Just curious!
OILFXPRO
Sorry to thread steal, but can an expert in NNs and the like check out https://www.mql5.com/en/forum/178418 and see if it's got any merit?
Why it didnet work?
Hi,
its great work , but non of the three EAs work, why?
its keep alerting with out any trade, may you tell me what is the problem please?
Another thing I can't make my mind up is whether adding technical indicators to the NN improve it....
Afterall, all TAs derive from price, so price in its purest form is the best predictive metric compared to any TA.
Another thing I can't make my mind up is whether adding technical indicators to the NN improve it.... Afterall, all TAs derive from price, so price in its purest form is the best predictive metric compared to any TA.
I am thinking you are right. There is a commercial software doing a NN salad of any kind of technical indicator with any parameters. It's name is "SafirXP"; it is very expensive and complex. People are working for years with it, but I don't know anybody who succed to make money with it.
Well, in computer science, we have something called "functional dependency"
A problem like a NN needs to broken down as follows:
Price is the direct results of something (Central Bankers activities)
Something directly determines price.
Several neural networks try to find patterns in price alone, which may have some respectible results. I believe "Better" uses price as its only input.
But to get the TRUE PREDICTIVE quality which we seek, we must find WHAT REAL WORLD ENTITIES ARE DRIVING PRICE. Sorry to get emphatic, but that's just how it is in the real world.
Unfortunately, banks do not move at the same time.... IE) One banks agenda moves on Monday, another bank moves on Friday..... so extreme noise takes place.
It is impossible to mindread those bankers.... impossible to quantify their thought processes without insider information, so we are always at a disadvantage.....
Some items I've looked at for inputs to a neural net are COT, CPI, interest rates, and unemployment. We must find WHICH reports have a strong linear correlation with price.... IE) which ones have predictive qualities. If the report does not, it should not be put into the NN.
This takes time and research.
Order:5161270 2007.11.27 04:50 sell
thanks barnix, nice move...
Good work
Hi Barnix,
I think if one try to predict the trend or whatever the inputs must be predictive. If one give the day of week as input it won't be able to tell us if the trend is up or down but it might be able to tell us that a large move is expected like every first Friday of the month for instance.
Hope this is usefull to you. Good work!
Regards
Derik
First time we need to generate an input for svm.
Something like this:
================================
Open Profitable Trade1 ...... MACD val1
Open Profitable Trade2 ...... MACD val2
Open Profitable Trade3 ...... MACD val3
Open Profitable Trade4 ...... MACD val4
Open Profitable Trade5 ...... MACD val5
Open Profitable Trade6 ...... MACD val6
================================
For generating this data we need a strategy tester
like in attachment.
==================================================
From Better's forum:
Profile Better - Automated Trading Championship 2007
From past comments, it appears that some folks in here are interested in the development approach used by Better. They might find this sample program and article draft I wrote helpful.
In this article, I provide and briefly describe a sample command-line C program that includes: parsing of MT4-exported CSV files, collection of statistical information, a simple trading system based on the collected statistics, a trading simulator, and trivial parameter optimization loops around both the statistics collection and the trading simulation phases.
Strategy Testing & Optimization in Standalone C Programs