MetaTrader 5 Strategy Tester: bugs, bugs, suggestions for improvement - page 43

New comment
 

Forum on trading, automated trading systems and strategy testing

Optimization speed has dropped drastically on build 2284

fxsaber, 2019.12.23 05:43

The speed has dropped. Not by times, but by tens of percent.

2269

2019.12.23 06:37:25.504 Core 1  pass 0 returned result 102608.000000 in 0:00:06.628
2019.12.23 06:37:29.936 Core 1  pass 1 returned result 102608.000000 in 0:00:04.432
2019.12.23 06:37:34.322 Core 1  pass 2 returned result 102608.000000 in 0:00:04.385
2019.12.23 06:37:38.667 Core 1  pass 3 returned result 102608.000000 in 0:00:04.344
2019.12.23 06:37:43.060 Core 1  pass 4 returned result 102608.000000 in 0:00:04.393
2019.12.23 06:37:47.358 Core 1  pass 5 returned result 102608.000000 in 0:00:04.297
2019.12.23 06:37:47.358 Tester  optimization finished, total passes 6
2019.12.23 06:37:47.368 Statistics      optimization done in 0 minutes 29 seconds
2019.12.23 06:37:47.368 Statistics      shortest pass 0:00:04.297, longest pass 0:00:06.628, average pass 0:00:04.746


2284

2019.12.23 06:39:54.696 Core 1  pass 0 returned result 102608.000000 in 0:00:07.232
2019.12.23 06:40:00.028 Core 1  pass 1 returned result 102608.000000 in 0:00:05.331
2019.12.23 06:40:05.285 Core 1  pass 2 returned result 102608.000000 in 0:00:05.257
2019.12.23 06:40:10.506 Core 1  pass 3 returned result 102608.000000 in 0:00:05.220
2019.12.23 06:40:15.981 Core 1  pass 4 returned result 102608.000000 in 0:00:05.474
2019.12.23 06:40:21.226 Core 1  pass 5 returned result 102608.000000 in 0:00:05.244
2019.12.23 06:40:21.226 Tester  optimization finished, total passes 6
2019.12.23 06:40:21.236 Statistics      optimization done in 0 minutes 35 seconds
2019.12.23 06:40:21.236 Statistics      shortest pass 0:00:05.220, longest pass 0:00:07.232, average pass 0:00:05.626

 
Andrey Khatimlianskii:

To speed up testing 99% of EAs.

For the remaining 1%, a crutch can be inserted.

Yeah, there's a reason for that. But this deprives me of the right to choose, saying "we know how you'll be better, and if you don't like it, you can work your way around our limitation". What prevents me from adding a box in the tester's settings to choose the maximum number of bars to be used, I don't understand.

 

Build 2280. The datetime input parameter for the EA is not correctly displayed on the Parameters tab. We take an EA of the following type

sinput datetime TimeInterval=0;

void OnInit()
{
  Print((string)TimeInterval);
}

The parameter tab is copied and pasted

TimeInterval=D'01.01.1970 13:58:16'

As a result, the Expert Advisor sees the parameter incorrectly and displays it incorrectly as 1970.01.01 00:00:00. Although it sees it correctly in the parameters in the log. This is from the log:

RTS-3.18,M1: testing of Experts\test.ex5 from 2018.01.01 00:00 to 2019.12.24 00:00 started with inputs:
  TimeInterval=D'01.01.1970 13:58:16'
2018.01.01 00:00:00   1970.01.01 00:00:00
Or does pasting from the clipboard not understand the date format everywhere? Is it possible to teach it everywhere then? Thank you.

 
traveller00:

Build 2280. The datetime input parameter for the EA is not correctly displayed on the Parameters tab. We take an EA of the following type

The parameter tab is copied and pasted

As a result, the Expert Advisor sees the parameter incorrectly and displays it incorrectly as 1970.01.01 00:00:00. Although it sees it correctly in the parameters in the log. This is from the log:

Or does pasting from the clipboard not understand the date format everywhere? Is it possible to teach it everywhere then? Thanks.

The testing boundaries are always aligned at the start of the day.

 
MetaQuotes:

The test boundaries are always aligned to the start of the day.

This is not a test boundary. It is an EA parameter of datetime type.

 
traveller00:

Build 2280. The datetime input parameter for the EA is not correctly displayed on the Parameters tab. We take an EA of the following type

The parameter tab is copied and pasted

As a result, the Expert Advisor sees the parameter incorrectly and displays it incorrectly as 1970.01.01 00:00:00. Although it sees it correctly in the parameters in the log. This is from the log:

Or does pasting from the clipboard not understand the date format everywhere? Is it possible to teach it everywhere then? Thanks.

What did you mean by

TimeInterval=D'01.01.1970 13:58:16'

This is a notation for MQL program sources, not for set-files. You could have just seen how values in set-files are generated

TimeInterval=50296

 
Slava:

This is a notation for MQL program sources, not for set-files. You could have just seen how values in set-files are generated

TimeInterval=50296

Yes, I copy-pasted from the parameters and saw what form he expects to see. But it's not very convenient for a person to handle such values. That's why I asked if it's possible to introduce date formats support there, like in the source.
 
traveller00:
Yes, I copy-pasted the parameters and saw what form he expects to see. But it's not very convenient for a person to handle such values. That's why I asked if it's possible to introduce date formats support there, like in the source.
It is possible. Let's do it.
 
Slava:
It's possible. Will do.

Thank you very much.

 
traveller00:
Yes, I copy-pasted FROM parameters and saw what form he expects to see. But it's not very convenient for a person to handle such values. That's why I asked if it's possible to introduce date formats support there, like in source.

You can use a stringer for the time being.

1...363738394041424344454647484950...84
New comment