No or incorrect display of drawdown in V12

Forums ProRealTime English forum ProRealTime platform support No or incorrect display of drawdown in V12

Viewing 15 posts - 76 through 90 (of 109 total)
  • #231902

    I called – not an easy process getting through to anyone. The 0800 support line said the office is closed and to refer to the website for opening times … the website said the support was still open for another 40 minutes …. so I called a landling number I have, held for a while and it just went dead. Next time I got lucky and spoke to a very helpful person who followed up the case number and said my email of Saturday had been received and was with the tech team. She said they will email me an update. So I pleaded with her that this would be more than the usual meaningless “they are working on it” and might give us a bit more info on the V12 update that Nicolas referred to and whether it should fix this double counting issue. If so, then we could all stop worrying about it, spending time on it and bugging them for updates – and use workarounds until they are ready to roll it out.  Let’s see what I get back from them.

    2 users thanked author for this post.
    #231923

    Well done Batty!  A Big Thank You from us All!

    Was that IG or PRT you spoke to?

    1 user thanked author for this post.
    #231947

    Thanks GraHal – it was IG, via their main switchboard 🙂

     

    1 user thanked author for this post.
    #232215

    Hi all,

    I discovered this issue when backtesting a system and also discovered this thread thanks to GraHal a few days ago. Today I made two pics of the issue as encountered. This probably looks like the one already described here but I would like to focus on the graphs. One is drawn above the price graph and the other can be drawn on demand in the detailed report. While the first always appears incorrect to me, the second always appears correct and the figures giving the number of trades and the gain in the detailed report might be randomly correct or doubled.

    Am I wrong when I say one equity curve is always labelled with doubled figures while the other one is always correct ?

    (V12 version of PRT – papertrading)

    Gabriel

    1 user thanked author for this post.
    #232236

    Thanks Fgats, you have your displays set up differently to me but it seems to be showing the exact same problem i.e. double counting the number of transactions, resulting in incorrect values for any other value that uses the number of transactions as part of its calculation formula – the Total Gain / Loss for example. Whereas any result that is a ration is alwys correct – e.g. Percentage Wins / Losses.

    This also lends credence to the thought that there are other users out there expeiencing the same problem. I suspect it’s quite a lot more and that some have not noticed it.

    Unfortunately, despite having been promised over a week ago that I would get a response to my latest email and the further information I sent, I have heard nothing.  Not very impressive really …


    @Fgats
    – are you using PRT Complete, via IG?  Or are you are direct client of PRT?  Picking up on a point that GrHal asked, maybe this problem is specific to PRT Complete via IG?

    1 user thanked author for this post.
    #232244

     

    Thanks Batty for your reply !

    I am not using PRT complerte via IG or other broker.   I have a subscription to PRT directly for a V12 version (real time) in order to experiment with paper-trading before choosing a broker.

    Have you made the same observations in relation to the two equity curves that I mention above? one being always correct while the other is always not correct and this regardless of whether the figures in the detailed report are themselves correct or not ?

    Gabriel

    #232245

    I had another look at your charts Gabriel. I think you are referring to the equity curve in the top left of each of your screenshots.  If so, I can see what you mean.  This particular thing seems different, with my reports, the equity curve is always correct.  I use the equity curve as a quick reference to show is the Detailed Report generated has double counted or not.  Perhaps if your PRT version is directly via PRT the bug is manifesting itself in different ways …. or mabye according to how users have set up their reports.  It’s interesting to know however that the same underlying problem is affecting both IG Clients and non IG Clients.

     

    #232248

     

    ok so my observations about the curves are irrelevant.

    I don’t have a broker and am only in contact with PRT so it is clear to me that the problem is a bug located in the program itself and it is difficult to understand why it takes so long to correct it .

    Gabriel

    #232261

    I have a subscription to PRT directly for a V12 version (real time) in order to experiment with paper-trading before choosing a broker.

    Batty (I know the quote above is not from your text), do you use the Demo version from IG for your Backtesting ? If Yes, the commonality could be that both you and Fgats use a Demo kind of version from PRT. I myself like to attest that the Papertrading version from PRT-IB is rather worthless (this is a separate PRT version). The Demo version from IG  I hardly use so I can’t tell about that.

    It still puzzles me why I never see this issue. This counts for V11 and V12, but always used with PRT-IB (I don’t develop within PRT-IG lately).

    Doesn’t someone has a regularly failing .itf to share, so others can test it and help finding the issue ? (I surely would like to help)

    1 user thanked author for this post.
    #232263

    Hi PeterSt

    See my post :  https://www.prorealcode.com/topic/anomalie-du-rapport-detaille-dun-backtest/

    Here to show the issue, I did a backtest with << (M15 SP500 LowBuyHignSell >>). This itf-program can be downloaded from the PRT library.

    Gabriel

    #232265

     

    I forgot to mention that I am using a complete V12 version of PRT (with real time) and not a demo-version.

    Gabriel

    1 user thanked author for this post.
    #232266

    Hmm … So phoentzs also suffers from it. But this is his strategy.

    @phoentzs
    , do you recall that this one is now bugging you too or still ? Or is it easy to test whether it does at this moment ? (and we know, you work with PRT-IG).

    #232267

    @Fgats – I think your observations about the curves are probably relevant. It seems to be a very similar symptom of the same underlying problem. My reports are set up differently to yours so that might be why I don’t see this. As to why it takes  so long to correct … who knows?  This thread is months old now and started with a number of different problems with the Detailed Reports – which were also reported bia the correct PRT tech support function in the platofrm. Some weeks ago, kindly Nicolas notified us that there had been an update and some of these issues appear to have been fixed at that time. But the double counting issue either started then, or bacame apparent then. Eventually I submitted a new tech support request and Nicloas said it was being looked into, but that there is also an update pending, with no date on this update as the tech teams priortities are eleswhere and that this is only affecting a few users.  They were directed to this thread which has many screenshot examples – and it is clearly affecting more users than are just on this thread. However, everything has gone dead from IG / PRT since then … it seems impossible to get any kind of update or guidance as to how or when this might be solved.  I don’t think this is the way things should work … we’re trying to help after all!


    @PeterSt
    – the double counting problem is also appartent on my Live account – V12 Complete IG. I just ran 6 individual backtests using different systems and it happened three times. I also wish there was some commonaility we could identify as to when it happens and when it does not, but it appears to be completely random. The six tests I just ran were one after another, just one at a time with each one completely closed before I started the next one. So the test environment for each was identical excepting for a very small difference in the time that each test was started i.e. same PC, same operating system, same chart, same instrument … same everything.

    1 user thanked author for this post.
    #232272

    Maybe in between the lines and certainly not to stoke any fires, if there are fires at all : Both the IB and IG versions are from the same date at this moment – March 28. This is rare in the first place because both versions are completely different, or at least bear completely different elements. The sheer fact that PRT seems to have stopped updating – because why stop doing that in an environment which is most clearly not trouble free (V12) ? Thus, it won’t be so that there are no updates, but something is holding those back. Say some severe issue we won’t know about.
    I’ll add to this that I now see that V11 has been updates as well, to a version from March 19. I don’t know exactly what the previous version was for V11, but I think it was July 2023. So that too is odd.

    My message would be : the issue can have been solved long gone, but ProRealTime might be unable to update currently.
    Does the issue of this topic bears prior it with them ? hard to tell because as usual we are in the dark with everything (like what changed and such). I think today someone said that somewhat longer ago other issues with the Detailed Report from Backtesting were solved. Well, that could be the moment that for me the V12 Detailed Reports started to show data again, while for 6 weeks I did not have any data. Just nothing. But I seemed to be the only one because I posted this in another thread and nobody screamed “me too !”. Meanwhile I could not backtest for 6 weeks. And this too was readily confirmed by PRT “we recognized the issue” and “usually it may take 3 weeks to update”. That became 6 weeks. No priority I guess. At least that would be true to some extent because if I would be the only one … then it can’t bear priority.

    ProRealTime works with a ticketing system. Anyone you ask will look in that system and see a status “is being worked on”. That is what they (support) read outloud to you. They won’t call someone to ask. What the ticket says is the truth (I even believe that). But being worked on is just that. There is not status like “will work on that from of April 15” or something.

    #232281

    @Gabriel, thank you for that link.
    Did you create – or change any optimization variables ? If so, what are the values of them to use ? … I feel it is better if you post your .itf of it, then I would not have these questions and a couple more.

    Additionally, how many backtests must I run before I can be sure I can NOT copy the behavior ? Please don’t say 1000. 🙂

    And regarding small time differences (mentioned today), does it also happen during the weekends ?

    Thanks !

Viewing 15 posts - 76 through 90 (of 109 total)

Create your free account now and post your request to benefit from the help of the community
Register or Login