Not logged in
Viewing 5 posts - 1 through 5 (of 5 total)

Forums>StrategyQuant>Application Support>Dukascopy data for GBPJPY not working on the QuantDataManager….

  • #238648|
    4 Posts

    Good day forum people. I am backtesting the following currency crosses on the latest Mt4 platform.

    1. EURUSD from 01-01-2015 to 11-02-2019 using Strategy Tester of mt4. Working perfectly

    2. GBPUSD as for 1. above working perfectly.

    3. USDJPY as for 1. above working perfectly.

    4. GBPJPY problem as follows. I first I click on Dukascopy-Add new Dukascopy data symbol.  I then I click on forex crosses to click on GBPJPY. I then click on download data for an existing symbol having ticked the GBPJPY box. I then set up data range as per the above settings in 1, 2, and 3. I then export the settings to csv which loads the data to mt4 as per 1, 2, 3 above. When I set up the Strategy Tester on mt4 for GBPJPY it simply kicks out no matter what I do. Obviously, I set the tester exactly as I have done for the other currency crosses but every time for GBPJPY it fails. Please help me as to what to now look for. My Question is this. Why are the other three currency crosses working perfectly but not the GBPJPY cross?

    A sample of open USDJPY is attached.

    My email address is

    1116 Posts

    Hello, did you try to reload the data? Delete first try to start from scratch? Did you try on a different MT4 installation?

    4 Posts

    Sorry for the delay in replying. Sorry, I don’t understand. Please explain, “Did you try on a different MT4 installation?”

    Yes, I have tried the following.

    1. Cancel and re-download data. Same result.

    2. Uninstall QuantDataManager reboot PC, re-install QuantDataManager. Same result.

    The journal states, “2019.02.17 10:01:08.318 TestGenerator: no history data ‘GBPJPY240’ from 2015.01.01 to 2016.01.05” If I try other periods, still the same see following different time periods.

    2019.02.17 10:08:03.847 TestGenerator: no history data ‘GBPJPY240’ from 2015.01.01 to 2015.12.31

    2019.02.17 10:09:23.004 TestGenerator: no history data ‘GBPJPY240’ from 2016.01.01 to 2016.12.31

    2017.01.01 to 2017.12.31 period only starts testing from 2017.11.07 at 20:00.

    Lastly, I set the backtest to start 2015.01.01 to 2018.12.31 The test starts at 2017.11.07, not 2015.01.01 I don’t know how to send screenshots of the events so using text is all I can do to describe what is happening. I sincerely believe there is a problem with the QuantDataManager program at present. Some of my trader contacts are not using it when I phoned them because of the same reasons I have given above. Now a friend of mine also tried on three currency crosses and he gave up because of data corruption when downloading using QuantDataManager. As a result, he is still using the version 4.6 tickdatadownloader program from Strategy Quant website which he has had for about 18 months and he has no problems with any currency cross data downloads including GBPJPY.

    I am sure that the problems WILL be sorted out soon. I must say the QuantDataManager when it is fully operational on ALL currency crosses is and will be the best. Many thanks for all the hard work that has gone into its development.

    What can be the problem?

    1116 Posts

    Ok will do some tests to check this on GBPJPY

    33 Posts

    Hey, I don’t know if this is related but I’ve also been having issues with exotic pairs recently. I had some strategies saved from a different computer so I had to re-download my data on another computer. When I went to test my strategies, most of them had similar results but my EURJPY strategy failed completely on tick data resolution testing. All tests from the previous computer were tested on tick data and passed. When I saw my EURJPY strategy fail on tick data testing in SQ3, I just chalked it up to my original tick data being bad and the strategy was trained on bad data.

    However, I later went and downloaded GBPJPY from the CDN network in the Data Manager and I could for some reason find strategies with several thousand trades when simulating on every resolution except for tick data testing. I’d find 400 strategies on a 30m and 1h timeframe with a 1-minute resolution but the second I tested them on real tick data, the results completely fell out the sky. strategies would go from having 14% max drawdown to having 100-2800% Maximum drawdown. Number of trades would go down significantly (thats the even weirder part because number of trades should stay constant). For the life of me I haven’t been able to figure out the reason. My data shouldn’t have any gaps in it. I think it might just be an SQ3 bug because SQ4 got different results with those strategies using 1-minute resolution. I haven’t tested GBPJPY with SQX yet because Im waiting for the next build release. But something seems off about the data for some reason.

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.