Discussion of article "Neural network: Self-optimizing Expert Advisor" - page 2
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
the article is interesting in the application of the BUT library
ALGLIB Free Edition (download):
delivered for free under open source licence (GPL)
offers full set of numerical functionality
single-threaded, without extensive low-level optimisations
copyleft requirement (GPL) does not suit most commercial applications
if I understand correctly - the free edition is single-threaded ..... and this puts a big fat cross on this library, I'm waiting for someone to finally add tensor flow to metatrader 5.
further unfortunately it is not clear what this library is applied to at all ...
where the input and output data is not clear, the topic of settings is poorly disclosed, the code present should be removed because it is just functions that need to give something to the input and get the result, I think it was enough to describe the input and output parameters - which by the way is almost not done,
examples are not useful at all, why are there examples of "binary-decimal converter", "determinant of prime numbers"??????
if one wants to write scientific articles, one should write in appropriate journals, and here it is trading ..... how can all this be applied to trading and in practice????
the article is not clear at all
the article is interesting in the application of the BUT library
ALGLIB Free Edition (download):
delivered for free under open source licence (GPL)
offers full set of numerical functionality
single-threaded, without extensive low-level optimisations
copyleft requirement (GPL) does not suit most commercial applications
if I understand correctly - the free edition is single-threaded ..... and this puts a big fat cross on this library, I'm waiting for someone to finally add tensor flow to metatrader 5.
further unfortunately it is not clear at all what this library is applied to ...
where the input and output data is not clear, the topic of settings is poorly disclosed, the code present should be removed because it is just functions that need to give something to the input and get the result, I think it was enough to describe the input and output parameters - which by the way is almost not done,
examples are not useful at all, why are there examples of "binary-decimal converter", "determinant of prime numbers"??????
if one wants to write scientific articles, one should write in appropriate journals, and here it is trading ..... how to apply all this to trading and in practice ?????
the article is not clear at all
Screwing in Tensorflow is not a problem (only via R). But you don't think it is the most advanced package today, do you? There are plenty of packages with deep neural networks nowadays. If only we knew how to use them.
No, I don't think so, you can use microsoft or caffe, it doesn't matter.
For me, there are several parameters that this package should fulfil:
1. free
2. multi-threaded and works with video cards, it is highly desirable the ability to use distributed networks, because one CPU is not enough, no matter how it is done.
3. no R/pyton layers, direct integration of dlls, code in C#
4. the package supports all modern types of neural networks and training methods
5. there are examples and well-developed documentation (preferably in Russian)
No, I don't think so, you can add microsoft and caffe, it doesn't really matter.
For me there are several parameters that this package should fulfil:
1. free
2. multi-threaded and works with video cards, it is highly desirable the ability to use distributed networks, because one CPU is not enough, no matter how it is done.
3. no R/pyton layers, direct integration of dlls, code in C#
4. the package supports all modern types of neural networks and training methods
5. there are examples and well-developed documentation (preferably in Russian).
Good luck
Is it possible to give an example for any simple indicator like RSI, CCI, MACD without using any kind of scripts
in order to find optimal values ... for example, you can take any built-in expert like "MACD Sample.mq5" and find optimal parameters InpTakeProfit, InpTrailingStop say once a week on Saturday.
the topic has died out, but this is the most promising direction.
Proof that the market is unpredictable. Because it is not right or wrong
Very good work and well explained.
I have a question, since I am starting to look at the networks I have the doubt if I could create a network to optimize a robot in which I participate but I do not have access to the code.
That is to say, can I create a network that makes optimisations to the robot I have by modifying its parameters by analysing the history?
I don't know if it was very clear... and besides, the answer seems to be negative hehehehe.
Best regards!
Hi, very elucidative.
Just to help, I work with AlgLib is so many platforms, and MQL is one of the most defective, like it's almost impossible to work with training tasks due to extremely poor resource management.
In this case I would use MLPSerialize, much easier to store the Network, and reliable.
Thanks anyway.