Dukascopy Support Board
http://www.dukascopy.com/swiss/english/forex/jforex/forum/

Different Results and Runtime between vJForex and Historical Tester
http://www.dukascopy.com/swiss/english/forex/jforex/forum/viewtopic.php?f=141&t=51068
Page 1 of 1

Author:  vtechdir [ Mon 14 Apr, 2014, 01:17 ]
Post subject:  Different Results and Runtime between vJForex and Historical Tester

Dear Support,

I develop the strategy in vJForex and when I want to see it's performance I import it in HT. However both the results and the runtime are considerable different between the two environments. I select the source in vJForex and paste it in a new strategy in HT (I expect to have the same code).
Which one should I trust, given that if I will run the strategy it probably will run in an environment similar to HT?
How can I reduce the time it takes in HT, in vJforex it takes 11 minutes whereas in HT it takes 2 hours? On both environments strategies have been run on the same pair and period.
In HT, I can apply my template and I can check my assumptions against the desired indicators. I used the same 1 min and cubic spline in both. Even though I used 1 min and the indicators are correct, it does not execute the order until 2 candles later, what is the issue?

I attach the strategy, vJForex run results and the HT results.

This happens for every strategy I tested in both environments.
Also, is there a plan to introduce Renkos in vJForex?

Kind Regards

Attachments:
vJForex_Results.txt [1.63 KiB]
Downloaded 137 times
HT_Results.docx [290.52 KiB]
Downloaded 145 times
Long_and_Short_v17_Strict013.vfs [77.39 KiB]
Downloaded 84 times
vJForex_Results.txt [1.63 KiB]
Downloaded 100 times
Long_and_Short_v17_Strict013.java [55.54 KiB]
Downloaded 87 times

Author:  amine_chourou [ Wed 16 Apr, 2014, 10:11 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Greetings,

Thanks for these detailed information.
I tested your strategy in both Visual Jforex and Jforex HT over 1 month, processed on 1mn/Cubic spline prices and I've got exactly the same result in both reports. Entry and exit were processed identically and it took about 2mn in VJF and 1m31s in JF. I'm pretty sure that the test results would the same over 6 months

When the code is pasted in JF, you are right to expect the same code and strategy behavior. What I suspect is some inconsistencies using Jforex historical tester. In order to be 100% sure, please follow the below instructions and restest your strategy:
1- Close VJF and Jforex platforms and locate the following file: C:\Users\...\AppData\Local\JForex\.cache (the folder AppData could be hidden, make sure windows hidden files option is activated)
2- Clear the content of the folder ".cache" which is used by both Jforex and VJF for HT purpose
3- Clear the cookies of your browser and your Java cache
4- launch both platforms and test your strategy from scratch in both environments

For the time being, we did not assess the feasibility of implementing Price based periods - i.e Renko, P&F etc - in Visual Jforex, we'll keep you updated regarding this matter.

Thanks !

Author:  vtechdir [ Tue 22 Apr, 2014, 02:22 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Hi Amine,

Thank you very much for your reply. It did help and it also created some unintended consequences:

1. Had to download the data again and that takes time.
2. Noticed that in vJForex cannot replace the Pair in the configuration (even though I replaced it, it still ran with the original pair) It appears that the whole has to be modified.
3. I tried to find a good definition of Period and DefaultPeriod and still not happy, do you know a link?
4. It will be nice to have the ProfitFactor, % Long/Short wins in both, so that we can compare Apples with apples:)

Have a great week.
Thank you

Author:  amine_chourou [ Thu 01 May, 2014, 15:46 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Hi

Sorry for the late answer.
In order to 'force' a given strategy to subscribe to a period, you have to set the default period to be equal to the candle period (ask or bid). Also, it is possible to right click on the default period variable to edit it.

The concept of default period and period is quite sensitive in terms of use: In Visual Jforex the trading period is set by default to 10mn, you can indeed edit this value and set it differently but if, for example, you say in an IF block Default period is equal to 10mn (from the drop-down list of the second input) then this condition has absolutely no sens as the default period will be always equal to 10mn <--> 'Not equal' exit is in this case obsolete as will never be executed.
For accurate information related to the default parameters regarding the automated trading, please refer to our Wiki page under Strategy API section : https://www.dukascopy.com/wiki/#
Does the information available in the reports are not helpful ? (Please check the attached screenshot) did you checked the strategy reports out of Jforex historical tester as well ?

Thanks

Image

Attachments:
VJFReport.JPG [78.3 KiB]
Downloaded 481 times

Author:  vtechdir [ Thu 01 May, 2014, 22:38 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Hi Amine,

Thank you very much for your reply.
I like the reporting in vJForex: Profit Factor, Long/Short Profit Trades Amount %, Positions Total. They are a minimum of what it takes to evaluate a strategy.

However, when I run the same strategy in HT these are not available. It will be consistent across environments to have them provided in HT as well.
I believe it is possible and the effort could be minimal ?

Kind Regards

Author:  amine_chourou [ Fri 02 May, 2014, 14:23 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Hi there,

Well when I launched you strategy in HT I get the following VJF report:
Is that what you're referring to ?
Image

Attachments:
Reports.JPG [81.54 KiB]
Downloaded 318 times

Author:  vtechdir [ Sat 03 May, 2014, 04:54 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Hi Amine,

Yes, you're right , how can I get this report in HT?
I mean the JForex interface - as in the attached report, not the vJForex....

Thank you and regards

Attachments:
JForex_Reports.docx [699.08 KiB]
Downloaded 81 times

Author:  amine_chourou [ Mon 05 May, 2014, 17:00 ]
Post subject:  Re: Different Results and Runtime between vJForex and Historical Tester

Greetings,

Currently the only Visual report is as the one previously attached. In case there is a need of a Jforex report, then the same strategy has to be exported to Jforex and tested over there.

Thanks.

  Page 1 of 1