MT5 terminal updated today and the "Optimisation" window does not show up during the test - page 10
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
Yes, the idea of genetic optimization is perverted to randomizer. There are no words. Some breeder reads all this and tears his hair: here is the key to success - "it is no use continuing what has been started". Or in other words: work already done does not make any sense.
Something you're saying is not right, my friend.
Evolution (andgenetic algorithms) are based on selective memory of precisely random changes in heritable parameters. How without a randomizer?
And, indeed, it often turns out that 'the work done makes no sense' - that is how species die out. evolutionary achievements turn out to be insufficiently adapted to the environment, and further adaptation makes no sense (or rather, simply does not happen - all changes are weeded out by selection).
Yes, the idea of genetic optimization has been perverted to a randomizer. There are no words. Some breeder reads all this and tears his hair out: here is the key to success - "no point in continuing. Or in other words: work already done does not make any sense.
It's more like we're tearing our hair out amidst the detailed explanations right in this very thread.
Ban for stupidity.
Previously, in optimisation, after a restart the old passes were picked up and taken into account in the Next generation. And the graph monotonically updated highs with each pass and lows were also higher (i.e. the picture was left to right increasing). I am attaching a screenshot showing the startup after rebooting the computer.
At the moment the best result is 2.12. The last "Best result":
Next generation does not show the old higher values and they are noticeably less than before the reboot. Build 1816.
From what I've read in this thread, if I understand correctly, a randomizer is applied that re-runs the genetics.
So it turns out the old best runs won't count or is that not true?
P.S. Usually on this algorithm, with the same history after 6200 runs the optimization was accelerated, the remainder was optimized in much less time than the previous 6200 runs. This remainder took about two or so hours. It's now 7 hours after the 6200 passes and you can't see the beginning of the acceleration as in past builds when it's coming to a close.
This algorithm has been optimized many dozens of times and I know how long it takes to optimize. I agree yes passes up to 6200 passes are processed faster,
But after that there is no acceleration and search for max starts again after a pause.
You can easily press stop, see the results and then continue. In genetics, there will only be a continuation from the beginning.
A grfm updated optimisation results window is available to evaluate the results.
Renat, I will support my colleagues in the inconvenience of intermediate evaluation.
I want to see profit, Sharpe Ratio, drawdown, and number of trades to understand the interim results.
How may I understand it from the chart?
Previously, in optimisation, after a restart the old passes were picked up and taken into account in the Next generation. And the graph monotonically updated highs with each pass and lows were also higher (i.e. the picture was left to right increasing). I am attaching a screenshot of the startup after rebooting the computer.
I have the same picture.
The impression is that after suspending the optimisation starts again.
Renat, I will support my colleagues in the inconvenience of intermediate evaluation.
I need to see profit, Sharpe Ratio, drawdown, and number of trades in order to understand intermediate results.
How may I understand it from the chart?
You do not need it in the chart.
If the suspending would work normally, it would not be a problem. If everything is OK, I have pressed "stop", checked all required coefficients, estimated whether I should optimize further and, if everything is OK, pressed "start" again.
But to do so, the optimization had to continue from the point where it had been interrupted. Now there seems to be a problem with it.
The timetable doesn't call for it.
If suspension was working properly, there would be no problem. Press "stop", look at all required coefficients, estimate whether it is worth further optimization, and if everything is OK, press "start" again.
But, this requires that the optimization continues from where it was interrupted. There seems to be a problem with this at the moment.
If remote agents are used, such stops/starts will significantly slow down the process. It takes me up to 10 minutes to resume agent's work.
The solution would be to write all statistics to your file, then copy it and see what happens...It's the same for me.
The impression is that after suspension - the optimisation starts all over again.
Me too.
If remote agents are used, such stops/starts will significantly slow down the process, it takes me up to 10 minutes to resume agents.
The solution is to write all stats into your file, then copy it and see what's going on...No, well, it's understandable that 10 minutes to resume is not OK.
And the file is coming to you now (in the "profile" branch)...