Updated to 1.8.1 (& MT950 -> MT971), results very different

[andy6898]

Yesterday I upgraded from V1.7.4 to V1.8.1, appear different backtesting results ??

Attached: before upgrading (V1.7.4 + Build950) and after upgrading (V1.8.1 + Build971)

http://imgur.com/a/eXWy0

Both use the same historical data files, use the same EA backtesting.

But backtesting result is a large difference ?? I do not know why ??

It let me know what to do, why different versions of different backtesting results ??

It is strange,,, This is obviously a big problem,

Please confirm that Tickstory immediately check for version 1.8.1 error bug ??

[tickstory]

Hi Andy6898,

Given that you have not regenerated the historical data files between upgrading from v1.7.4 to v1.8.1, we know that this isn't a data issue (there were no changes made to Tickstory that should account for any back-tester differences anyway). This means that there are several possibilities that could be affecting your results: A change in your EA or its parameters or a change in Metatrader Strategy Tester.
Are you sure that the 'spread' value in the Strategy Tester 'spread' box was explicitly set to '50' in both back-tests? If so, I would also be checking to see if there are any EA settings that are different between the two back-tests. Seems to me that the "average profit" for each trade in the v1.7.4 back-test was much lower, possibly indicating that the TP value was lower? Export your EA settings to make sure you base-line all possible variables. After that, it will be necessary to take a sample of the first few trades and understand why there are differences.

Hope this helps.

[andy6898]

Do you mean to say,, each upgrade must re-export the updated historical data ??

You mentioned the spread or EA settings ... These values are the same figures twice Backtesting.

That is all I used to set data to back-test the new version 1.81 are the same as the old version.

Why new version of the back-test results are different? I find it hard to understand??

Please do the same test with me, is there a Bug??


Best regard.

[fenky]

dear support:
im got same issue.........how to solve it?

[tickstory]

Hi,

Do you mean to say,, each upgrade must re-export the updated historical data ??
Please do the same test with me, is there a Bug??

This is not what we said - we said that since you are using the same data between MT4 Build 950 and Build 971, it can't be an issue that Tickstory has introduced.
Looking at the MQL forums about the latest Build 971 release, it appears that others are having similar issues:

https://forum.mql4.com/72722/page7

Unforunately if there are problems, we just need to wait for an official response from MT4. According to one user, they had some success trying a different broker terminal. A user on this thread also suggested (relating to optimisations):

It is because of conflicting values in the "cache" files. Just exit MetaTrader, delete all the cache files in the "tester\caches" directory and redo you optimisations.

I would suggest trying these solutions. If you are continuing to have issues, you might need to pinpoint exactly why there are differences in your EA and highlighting them on this forum so Metaquotes are aware of the issue. At this stage, it appears that affected users have found a workaround, so it hopefully shouldn't be necessary to revert back to MT950.

Please let us know how you go.

Thanks.

[netprofit]

Yesterday I upgraded from V1.7.4 to V1.8.1, appear different backtesting results ??

Attached: before upgrading (V1.7.4 + Build950) and after upgrading (V1.8.1 + Build971)

http://imgur.com/a/eXWy0

Both use the same historical data files, use the same EA backtesting.

But backtesting result is a large difference ?? I do not know why ??

It let me know what to do, why different versions of different backtesting results ??

It is strange,,, This is obviously a big problem,

Please confirm that Tickstory immediately check for version 1.8.1 error bug ??

I think your 950 results are probably wrong because of the 950 spread bug:
https://tickstory.com/articles/tickstor ... -released/

But if your 971 results are correct, i am not sure