Page 1 of 2

wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 7:15 am
by mtigol
Hello,

tickStrory 1.7.4.0
Metatrader build 950

When running optimization, only the 1st optimization set returns results. the rest return with 0 values.

For example, have a look at the following screenshots
both.png

You can see two optimization sessions with variable value of CandleTRLimit.

When starts at 100, only this value is calculated correctly and the rest (110,120,130) return zero's.
When starts at 120, same symptom, but here you can see that 120 actually return a value, while the rest return zero's.

regular backtest works OK for all values.

Please advise
Thanks

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 10:29 am
by tickstory
Hi Mtigol,

It appears there is some issue with the MT4 Strategy Tester if this is the case. I would try a few different tests to see if you can work around the issue such as:

- Reduce the back-test date range (perhaps MT4 is running out of memory with your specific EA causing the next run to fail?).
- Place some debug information to see if your EA is actually running for subsequent passes. Perhaps there is some fatal error that occurs after the first pass?
- Test another EA to ensure it's not something thats occurring with all EAs (We've used MACD Sample and multiple passes are showing fine)
- Try rolling back to Build 910 and see if that makes any difference (there have been a number of users on the MQL forums suggesting they are having numerous back-testing issues with the latest versions such as slow processing, etc).

Hope this helps.

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 11:22 am
by mtigol
Hello,

10x for the prompt reply.

- The date range is already reduced.... about 25 days...
- There seems to be no helpful debug info in the Journal:

Code: Select all

 2016.02.08 13:09:42.218	TestGenerator: file "C:\Users\Mati\AppData\Roaming\MetaQuotes\Terminal\C08F8A0AE8F2EC29C56A136EC47A28D9\tester\history\EURUSD1_0.fxt" is read-only
2016.02.08 13:09:42.222	TestGenerator: symbol EURUSD period 1 model 0 from 2011.02.01 to 2016.01.31
2016.02.08 13:10:00.046	MACD Sample: optimization started
2016.02.08 13:10:06.328	MACD Sample: optimization finished in
2016.02.08 13:10:06.354	There were 6 passes done during optimization
- Here other EA such MACD provides the same results :(
- do you have links to earlier versions of metatrader? 910, 920 etc.

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 12:24 pm
by tickstory
Hi Mtigol,

Sounds a bit odd. Not sure what the reason for this could be. You can download Build 910 from here:

http://tickstory.com/files/Metatrader.update.910.zip

This build was just prior to a number of updates MT4 made, so hopefully it helps.

Regards.

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 2:46 pm
by mtigol
Unfortunately same results with build 910 (as well as when running the suggested MACD EA optimization MAtrendperiod as optimization parameter).

Any further suggestions?

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 2:53 pm
by mtigol
P.S. This happens only via tickstory launcher. when running the optimizer through the "regular" independent metatrader terminal - the optimizer shows results for all build versions (910, 950).

Re: wrong optimization - only 1st set calculated

Posted: Mon Feb 08, 2016 8:42 pm
by tickstory
Hi Mtigol,

Unfortunately we can't reproduce the issue on our side. How large are your FXT files? What date range have you exported?

Thanks.

Re: wrong optimization - only 1st set calculated

Posted: Wed Feb 10, 2016 6:20 pm
by mtigol
Hi,
6GB - about 9 years.

Re: wrong optimization - only 1st set calculated

Posted: Wed Feb 10, 2016 9:29 pm
by tickstory
Hi Mtigol - MT4 has a 4GB FXT limitation, so I would try to reduce the size of your FXT files to see if this is the cause of the problem. If you don't want to lose your current exports, you can simply move them to a temporary folder while you try exporting a smaller range (I would suggest just 1 month to test the theory).

Hope this helps.

Re: wrong optimization - only 1st set calculated

Posted: Fri Feb 12, 2016 4:26 pm
by mtigol
great! problem solved.

P.S. strange that the problem with bigger files appears only in optimization mode.
do you recommend remaining with the downgraded 910 build?