Discussion of article "Creating Neural Network EAs Using MQL5 Wizard and Hlaiman EA Generator" - page 4
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
On the highlighted point, don't lead the reader astray here, GA is not an implementation of NS, GA is an optimisation method.
On the contrary, I try not to lead the reader astray, so I don't take him into any rubbish, including genetic algorithm.
You probably didn't notice in the highlighted part that there is "or"(||), not "and"(&&), although in the second case there would be no serious error, because NS and GA have already merged, you haven't heard - NEAT, evolutionary neural network models, google it - it's an interesting and relatively new direction.
However, despite the fact that the theory of NS is interesting and informative, I propose to all those who wish to discuss it, to address to the appropriate threads on this forum, where in the context of already accumulated information, we can do it more constructively.
And in this thread I propose to concentrate on automation and creation of trading robots, as this topic is important enough and deserves a separate consideration, especially since it is indicated in the topic.
Perhaps someone already has neural network robots of his own development or downloaded and generated them using the method described in the article, let's compare, criticise, test, analyse, etc., etc.
.
I am not taking the reader to anything bad, including the genetic algorithm.
You probably did not notice in the highlighted part that there is "or"(||), not "and"(&&), although in the second case there would not be a serious error, because NS and GA have already merged, you have not heard of NEAT, evolutionary neural network models, google it - it is an interesting and relatively new direction.
However, despite the fact that the theory of NS is interesting and informative, I propose to all those who wish to discuss it, to address to the appropriate threads on this forum, where in the context of already accumulated information, we can do it more constructively.
And in this thread I propose to concentrate on automation and creation of trading robots, as this topic is important enough and deserves a separate consideration, especially since it is indicated in the topic.
Perhaps someone already has neural network robots of his own development or downloaded and generated them using the method described in the article, let's compare, criticise, test, analyse, etc., etc.
.
Aha, we don't want to talk in essence. Well, I didn't really want to.
Apaite your advertising topic yourself. Au revoir.
Yeah, we don't really want to talk about it. Well, I didn't really want to.
You can start your own ad thread. Au revoir.
Why I do not want to, on the contrary, I suggested to continue the discussion where it has already been conducted for several years, so as not to start here from scratch and not to repeat.
If of course you have something new on this topic, or you are just looking for a place to increase your rating by copypasting those NS discussions.
And in general - in fact, I believe that theoretical discussions of NS for traders are useful only for educational purposes and I am a categorical opponent of artisanal implementations, because IMHO the vast majority of negative experience of trading with the help of such implementations, in fact, have nothing to do with NS, and are due to trivial errors in the development.
That is, my attempt to slightly adjust the direction of the discussion is dictated by the desire to increase its usefulness for practical purposes.
.
Thank you all for your participation in the discussion and feedback, who is interested to see in detail the signals of the test EA Hlaiman EA Generator 007 -
Login : 1512007
Password : a3mlnkj
Server : MetaQuotes-Demo
I, for example, would love to watch a month in real life. Then there will be at least some notion of profitability of the system, but not on demo. Do not have an extra hundred quid to start 0.01 lot with 500 leverage ???? And a demo is a demo, no matter how it is and you can't trust it!
You are right, it will be possible to assess the real profitability of the strategy only in real trading conditions of a particular brokerage centre and we will try to launch such monitoring, but only after the development and debugging of an additional, high-frequency SignalHFT module, which was mentioned above.
In this case, for the forward test, preliminary evaluation of the strategy efficiency, in terms of price movement forecasting and accuracy in issuing trading signals, the real trading can only disturb the purity of experiments, due to worse execution and cancellation or delays in triggering orders than on the demo.
.
If we talk about evaluating the prospects of the most high-frequency trading (HFT) on Forex, for example, by extrapolation of known stock market analogues, IMHO we can identify at least three main components of effective strategy development - means of accelerating the work of trading algorithms, parallelisation of calculations and components for simultaneous work on different trading platforms.
Today, MQL5 already has functions for fast trading Async, OpenCL support, now MQ is adapting fast MQL5 on the MT4 platform, we are also developing new HFT engine components for Hlaiman EA Generator.
In particular, in addition to multi-terminal interaction tools for MT4 and MT5, new plug-ins for the integration of Dukascopy, FXCM, MBT, Ninja, LiveTrade will soon be available, allowing traders to create complex, multi-platform, multi-module strategies.
Elements of distributed computation and parallelisation are also present in the neural network robots in this article. In the conclusions section it is stated that the Expert Advisor and the script sent to the Hlaiman engine are executed asynchronously, as well as the possibility of running the terminal with the Expert Advisor and hlaim.exe with the plugin on different computers with network communication between them. If you have two networked computers, you can check this. To do this, run hlaim.exe on one of the computers, MT5 terminal on the other, and in the settings of the neural network robot specify the network name of the computer on which you run hlaim.
For proper operation, the computer where hlaim.exe is launched must have a directory with the same name as the MQL5\FILES directory of the terminal, as well as a neural network training data file corresponding to the Expert Advisor and the symbol.
is it possible to set up network training based on trader's past profitable trades?
that the network tracks the position, value and direction of indicators and oscillators used by the trader - with any deviation in value - for example 1-5% (customisable).
in this way you can identify profitable trades - for example, sort them by profit - and the Expert Advisor will recommend trades at similar moments.
I could use such an option.
And by the way, running the EA through successful trades of any trader with a positive overall result and accumulating the test results in memory with the subsequent upgrade of this memory into another neural network would also be appropriate.
implementation - sending reports and the algorithm of profitable trades to the central network - processing - selling or sending free updates to the brains of the grid.
are such variants possible? I would be very interested in participating in such a variant. if you are interested, please write me.
is it possible to set up network training based on trader's past profitable trades?
that the network tracks the position, value and direction of indicators and oscillators used by the trader - with any deviation in value - for example 1-5% (customisable).
in this way you can identify profitable trades - for example, sort them by profit - and the Expert Advisor will recommend trades at similar moments.
I could use such an option.
And by the way, running the EA through successful trades of any trader with a positive overall result and accumulating the test results in memory with the subsequent upgrade of this memory into another neural network would also be appropriate.
implementation - sending reports and the algorithm of profitable trades to the central network - processing - selling or sending free updates to the brains of the grid.
I would be very interested to participate in such a variant. if you are interested, please contact me.