Optimisation in the Strategy Tester - page 17
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
Any speed improvements - kudos!
But I'd still like to move on from the matrix application to the processing of the story according to the roughly described technology.
Can you tell me what is the reason for the limitation in the tester of 10496 passes during optimization?
Bild last 450, chose full brute force and genetic optimisation max 10k passes.
Expert Advisor from the standard delivery of the terminal ...2EmaTime.
Any speed improvements - kudos!
But I would like to switch from mathematical application to history processing using the technology described above.
This is not just a toy, first of all it is used for working out the possibilities of speedup,
Secondly, it can already be used for simplified calculations on history (open here and close there, the difference in pips is a profit, without real order openings).
And this is something.
First of all, we will work out the possibilities of acceleration on it,
Secondly, it can already be used for simplified calculations on the history (like here you opened, here you closed, the difference in pips is profit, without real order openings).
And that's something.
And it works on any cunning synthetics too.
--
Too bad you can't write scripts for this thing. I'll have to make my own.
Yes, and it works on any tricky synthetics too.
--
Too bad you can't write scripts for this thing, you'll have to make your own.
Than to make your own GA,
Write a C++ tool that captures the tester's events and button presses instead of a script, that's a script for you.
I understand that soon the number of parameters will increase, and there will be enough for your ideas.
MetaTrader 5 Client Terminal build 470
- Terminal: Fixed drawing of graphic object "Fibo Fan".
- Terminal: Fixed control of displaying scales in the "Chart" graphical object.
- MQL5: Fixed restart of an Expert Advisor when changing an account.
- MetaTester: Added support for navigating through the chart from the keyboard during visual testing.
- MetaTester: Changed rates for visual testing for different regulator positions.
- MetaTester: Added support for custom indicators in a template during visual testing.
- MetaTester: Added output of information on the change of an agent settings in its log.
- MetaTester: Changed location of the Data Window in the visual testing.
- MetaTester: Introduced a general limit on the number of testing and optimization parameters - 1024.
- MetaTester: Fixed release of resources during testing and optimization.
- MetaTester: Fixed the cleanup command behavior. Now all agents, including those working in the MQL5 Cloud, get it.
- MetaTester: Fixed formation of paths to the data in the visual testing mode.
- MetaTester: Fixed displaying of the chart scale during visual testing.
- MetaTester: Fixed display of the tick chart during visual testing.
- MetaTester: Fixed downloading of the modified EX5 library.
- MetaTester: Fixed operation of an agent when addressing non-existing agents.
- Fixed messages on the forum and the crash logs.
- Updated the documentation.
Automatic update is available through the LiveUpdate system.Download the MetaTrader 5 client terminal from the following link: https://download.mql5.com/cdn/web/metaquotes.software.corp/mt5/mt5setup.exe
This is really "something". But why 1024? - Why not 1'000'000?
The question is how to use it now. Handwritten in the optimiser's settings is problematic. But if you could specify 2-dimensional array as input, mmm..., like this:
That's really "something". Only why 1024? - For a nice number? Why not 1'000'000?
Organisational arrangements have been made in relation to cloud agents. There has been an audit of the control of memory usage by test agents. There were limits before, but they were not coordinated between optimization and testing. We also brought the documentation into compliance.
It has been estimated that 1024 is more than enough, despite the ever increasing complexity of the Strategy Wizard.
But it is a limit on the total number of parameters, not on the number of simultaneously optimized parameters, which remained 64
Organisational arrangements were made in relation to cloud agents. The control of memory usage by testing agents has been audited. Limitations were in place before, but they were not coordinated between optimization and testing. The documentation was also brought in line.
It has been estimated that 1024 is more than enough, despite the ever increasing complexity of the Strategy Wizard.
But it is a restriction on the total number of parameters, not the number of simultaneously optimized parameters which remains 64.
Oops. The transfer of network training from a homemade calculator to an in-house tester is cancelled, which is a pity.
Hello moderator!
I have a question: why do all three pictures attached here show a different number of optimizer passes? What is the reason?
According to the first picture the number of passes should be 817. However, on the second and third picture their number is much lower.
I have a question. Why do all the three attached here pictures show a different number of optimizer's passes? What is the reason?
According to the first picture the number of passes should be 817. However, in the second and third pictures the number of passes is much lower.
It only makes sense to use genetics if the number of passes is well over 10000. In your case there are only 817 passes and they need to be run in full brute force mode.
Genetics operates by modelling gene sequences to produce efficient individuals. This requires a certain number of populations, usually at least 10,000 passes.