Chatter about the MT5 strategy tester - page 22
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
Indeed! It would be nice to turn off the optimisation mode after it's finished.
Once again, you start checking individual passes and come across the same problem.
When clicking on optimization results all is OK, a single pass is launched. And after you have clicked on them a few times, it already remembers that you are in single test mode.
And after that, if you change something manually in parameters, and press start, it starts optimization. ((
It would be a good idea to distribute tasks to local agents in smaller portions when parameters are fully enumerated. For example 1-2% of all tasks. When one of them is free, give them the same amount more.
Now the situation is like this. All agents were given 704 tasks, and since they all have different durations depending on parameters to be optimized, 2 agents finished the job, and 2 agents got the longest tasks. In 10 hours they solved 41%. I'm afraid that optimization at such a pace will be over in another 15 hours. It would be better to load all agents.
I understand that issuing tasks in batches is better to reduce time on related processes. Transfer of quotes, etc., but on local agents it all happens quickly, and on local network too, for them it would be good to make smaller portions when handing out tasks.
It would be a good idea to distribute tasks to local agents in smaller portions when parameters are fully enumerated. For example 1-2% of all tasks. When one of them is free, give them the same amount more.
Better yet, have a common job queue which is accessed by the released agents.
But I'm afraid this is not allowed by the architecture of the cloud.
The best thing is to have a common job queue which is accessed by the agents who are freed.
But I'm afraid this is not allowed by the architecture of the cloud.
So I propose to keep optimal algorithm for cloud, and for local agents and local network to distribute tasks in small batches or 1 as you suggest (in terms of speed it's probably the best).