The Ultimate AI EA Project - page 3
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
Hello.
I feel there is some form of misunderstandings on this discussion, which I wish to address.
In this project, I intend to create Neural Network libraries, find and share the best way to apply and use OpenCL in this domain (and share CL code for common functionalities if possible).
Please note, I will use these tools to up my trading strategy, and only provide performance statistics here. The project on the terminal will simply function as a (domain-specific) codebase
for developers who are interested in AI.
I intend to try and work with MQL despite it's limits on features and resources, and at least shed some light on what an AI enthusiast can do with 'the little' we have.
...What limits ?
What limits ?
Flexibility of shared projects, absence of machine learning and Neural Network libraries, inefficient work-around for compatibility with GPU's... These are necessary for complex computations that come along with AI.
Flexibility of shared projects
Do you mean from MetaEditor ? What concrete problems ?
, absence of machine learning and Neural Network libraries,
inefficient work-around for compatibility with GPU's...
This one I don't get it, can you elaborate a bit please ? OpenCL is a standard, you can check the compatibility of your hardware, and use an emulator if needed (of course in this case you will lose the benefits.
Do you mean from MetaEditor ? What concrete problems ?
Agree, for sure. But it's not a limit of the language itself.This one I don't get it, can you elaborate a bit please ? OpenCL is a standard, you can check the compatibility of your hardware, and use an emulator if needed (of course in this case you will lose the benefits.
The standard OpenCL library does not work with my intel GPU. I'd have to code the entire thing or use a different pc.
I'm not sure if I experience this issue alone. Is it possible to access an external GPU from within an MQL program? I have not seen any post, article or documentation stating so.
Programming languages like python have GPU libraries like PyCuda, which help a developer to get a working model without having to learn OpenCL first. I don't mean to compare MQL with python, but more support is needed in this regard.
About shared projects, how do I tell the changes made per user basis? I'd like to know when/who/what someone adds to the main source. It also seems that most people have no idea of how this 'projects' thing works. It is therefore so hectic to get a project up and running.
PS: I didn't mean that MQL language is limited in terms of syntax, code or anything like that. The language is fine and efficient, and that is why I chose it for this particular project over python.
Most brokers offer support for MTx, and that will most likely lead more people here in the future, when the interest for Algo-trading increases.
edit: I mostly utilize an external razor graghics card for my python projects. I'm wondering if this is possible using OpenCL.
The standard OpenCL library does not work with my intel GPU. I'd have to code the entire thing or use a different pc.
I'm not sure if I experience this issue alone. Is it possible to access an external GPU from within an MQL program? I have not seen any post, article or documentation stating so.
Programming languages like python have GPU libraries like PyCuda, which help a developer to get a working model without having to learn OpenCL first. I don't mean to compare MQL with python, but more support is needed in this regard.
About shared projects, how do I tell the changes made per user basis? I'd like to know when/who/what someone adds to the main source. It also seems that most people have no idea of how this 'projects' thing works. It is therefore so hectic to get a project up and running.
PS: I didn't mean that MQL language is limited in terms of syntax, code or anything like that. The language is fine and efficient, and that is why I chose it for this particular project over python.
Most brokers offer support for MTx, and that will most likely lead more people here in the future, when the interest for Algo-trading increases.
Ok, I was curious to know if there was real mql limitation, thanks for the clarification.
35 participants and counting. Thank you to all developers on board. Let's make this project successful.
Cheers!
Hey guys,
Sorry for the newbie question but, how can I join the "Artificial Intelligence (AI) - A reinforcement-leaning algo in pure MQL"? I would love to start playing with this!
I have 2 computers with Ryzen 9 3950X, each has 1 NVIDIA 2080 TI.
I can do things like playing with open source TensorTrade and mine cryptocurrency like Ravencoin (t-rex miner) but, I would love to start using the NVIDIA card with the Strategy Tester! Or with the Expert Advisor, anything lol !
Thing is, my MetaTrader 5 doesn't recognize the NVIDIA card at the Agents list, only the CPUs are there. What am I missing on MT5 so it can see the NVIDIA card (like this: https://www.mql5.com/en/articles/690 ) ?
Cheers!
This: https://www.forexearobots.com/open-cl-test-script-for-metatrader-5/ - looks like a lot of fun!
I'll give it a try! But, not sure how. LOL
This: https://www.forexearobots.com/open-cl-test-script-for-metatrader-5/ - looks like a lot of fun!
I'll give it a try! But, not sure how. LOL
Ultimate_AI is an ongoing public project under the MIT license. You can join and help by running the training and/or coding some modules, by right clicking on the project name inside MetaTrader5 projects tab.
Hello Nelson,
Thank you so much for your fast reply and help! I really appreciate it. You rock!
I'll give it a try tomorrow evening. Time to recharge now... ^_^
Thanks again buddy!