Odd / Annoying happenings while using PRT Platform

Forums ProRealTime English forum ProRealTime platform support Odd / Annoying happenings while using PRT Platform

Viewing 15 posts - 166 through 180 (of 224 total)
  • #208754

    Only thing I can imagine is that when the 10 Rejects for min distance happened (Mon 23 Jan) that 7 managed to ‘catch on / open‘, but as the Algo was ultimately Rejected / Stopped the positions remained open as they were not really tied to the Algo. and so were not dealt with by the Close any Trades settings?

    Either above or a similar scenario for Algos that get stopped for more than the allowed number of tries   … of which I get a spate of whenever there is high volatility.

    #208761

    of which I get a spate of whenever there is high volatility.

    Secret : when Things get tough, IG interferes manually. This could be related to your Monday Morning man.

    #208762

    As it is only Demo one would think … why would IG care?

    However, it may be related to overall server / network activity and so IG reign in Demo servers and networks in order for Live servers and networks to function correctly?  This, else the servers / network just cannot cope??

    Re another Rejection I get lots along with this error message … ‘another order may be in the process of being processed’ … Hello IG, what about Orders from all the other many users … you may as well let my few Orders through along with their many orders!!?? 🙂

    I wish I knew exactly why the various Rejections happen??

    #208763

    another order may be in the process of being processed

    I get this very often, in LIVE. I appreciate this as “normal” because I could be consuming the order myself (several Systems).
    At least a couple of times per day.

    #208765

    I could be consuming the order myself (several Systems).

    Yes I agree, but if another order is in the process of being filled on another System then that’s life, this is what happens!

    Above is same as another user trying to get filled in the same direction on the same Instrument … are we all in a queue fighting with one another to get filled first?? Not with at market orders surely, which mine mostly are … for just that reason.

    If stop and limit orders then I guess we are all fighting to get a certain price.  If we did all get that certain ideal price then IG’s books become unbalanced … I think I’ve answered my own question?? 🙂

    #208794

    Yes I agree, but if another order is in the process of being filled on another System then that’s life, this is what happens!

    What I see or feel is that the message tells you that you are waiting for yourself (thus I see that it never relates to others). My message (yours too I suppose) literally tells “If you want to talk about this, call support”.

    This is a typical message which (for me) appears one time only and immediately after (the order) always goes through.
    What you see below as a virtual example, will not be “the” situation, but you can imagine something with it. The real example will be at buying (getting in to the market) with two orders from you, at – or very close to – the same price. It will all be as you tell, GraHal, but it is you doing it to yourself (with two orders fighting for that same price) and it won’t be related to others. That is, not when this message occurs.

    All as how *I* see it, from empirical observation. And I never “called support to talk about it”.
    And oh, while this is live, I see no reason why this should not happen in Demo as well. Maybe even quicker ?

    1 user thanked author for this post.
    #208923

    Learning better and better how IG operates (call it manipulates), here’s another example of last night.

    Bars you see are 1 minute. Price (after Fed) rises relatively steeply with 200 points in 7 minutes.
    Bars used are 6 seconds (thus TF is 6s).
    Now each second an attempt is done to fill a Limit Buy order. This fails at the end of the bar. System is stopped etc. (mweh).

    I am fairly 100% sure by now, that during this stage, IG sets order filling to “manual”. There’s all sorts of hedging going on, and this is too much for the poor manual guys and gals. So they put all on hold by setting that point distance to a too crazy number, so my new order will never be filled. Wait, it is just rejected and therefore never filled.

    That the system is stopped is ProRealTime’s recognizing that this is never going to work – and that I perceivably handed an erroneous program to their server. However, the truth is different – IG sets that distance to something inordinate (this system ran for 7 days with xx trades).

    The sad truth is also that this system missed I-don’t-know how many good trades, all in the flow of a really great uptrend. This is – of course – not the intention of IG, but meanwhile a fact.

    I am now handing this situation to PRT because it seems to me that this can be “error trapped”, because it is bar-length related. Thus, do notice that this was thrown out after 6 attempts instead of the usual 10, just because by bar is 6 seconds of length only. Is it 1 minute ? then the number of attempt allowed should be at least longer than this length, hence more than 60 times (at a rate of one per second).
    In the end, I know from experience that this manual procedure at IG’s can last for over 3 minutes. So maybe it should be set time-related, the length of the bar regardless.

    1 user thanked author for this post.
    #208933

    Jolly Good Research Peter!!

    So above is on Live?  Following from what you say … why should it also occur on Demo (virtual money) and so IG would not lose real money no matter what happens??

    BUT … maybe IG make Demo as near to Live as they can else we would all be shouting … ‘How come I got Rejected on Live with Algo xyz (at 20:42 on 1 Feb 23) when Algo xyz (exact same code) traded fine on Demo (at 20:42 on 1 Feb 23).’

    #208941

    Yes, that is on Live.

    To answer your question with hopefully some sense :

    First off, indeed it will be so that all data will be handed to the Demo systems or else we won’t be able to Backtest in retrospect. I mean, it may be beyond your experience, but if one has passed some Live trades and later it must be justified, someone like me will add the appropriate graphing (say debug data) to the very same system, and watch what actually happened. This then of course has to be the same for data or else nobody will trust the backtesting. And as you know, this always *is* the same (with me it is) unless the System really running on Live (or Demo for that matter) is thrown out (a Backtest can’t be thrown out).

    But this was only one part of the answer. The other is that not all will be able to mimic, and this is how especially in Demo things often fail, which work out fine in Live. Maybe I didn’t analyse that decently, but I sure know that it does (and often told about it). Thus, would I bring such a Live system to Demo, then say 90% chance it will fail somewhere today. This, while the same system on Live would work for weeks (I could even run them in parallel to testify for that).

    The (most) stupid thing is that I know for a fact that IG switches to “manual mode” on their side, to hedge things. I suppose (but never asked) that at Fed’s decisions and such, a whole team is sitting and shivering in order to get things done (in time !). I could show you badly ending trades because they could not cope (slippage (ahem !) of more then 3 minutes).
    Yes, I know, unheard.


    A first response from PRT Support is also in already : what would be able to do to prevent form this ?
    IOW, it is not really easy to determine this fault situation which technically is no less than you being too far from the price they ad-hoc set. And back to your Monday Morning Man, to me this now looks as a similar situation : at Sunday night-whatever the market opens again, at what idiot distance which now is normal ?  If they are not there to hedge, they will lose money and so they set that even more idiot distance as a safeguard for themselves ?
    Side note on this latter : this should not bother us really, because it would be more rare that your system will trigger on an Entry for this usually one-bar excursion. But this depends on your system of course. All ‘n all that situation should be automated by IG and if that fails (for Reset), there’s always the MMM for backup.

    #208944

    IG switches to “manual mode” on their side, to hedge things.

    My understanding – after reading an ex IG employee – is that there is an auto-monitor of the balance of the overall book. When the balance exceeds a preset figure then sufficient trades are made (outside of IG books) to offset / bring the book back into balance.

    This then of course has to be the same for data or else nobody will trust the backtesting. And as you know, this always *is* the same (with me it is) unless the System really running on Live (or Demo for that matter) is thrown out (a Backtest can’t be thrown out).

    If we have Systems that are (never) not Rejected in Live then the Backtest performance will more approximate to Live performance.

    If we have lots of Pending Orders which are Rejected often in Live (for < minimum  distance for example)  then Backtest may look better?

    We are in agreement btw! 🙂

     

    1 user thanked author for this post.
    #208953

    If we have Systems that are (never) not Rejected in Live then the Backtest performance will more approximate to Live performance.

    Yeah. But in the end and IMO this can not be how a Backtest will influence the expected results in Live (or Demo for that matter). But now listen to this :

    *Because* things plainly fail regularly in Live, I by now lost something between 5K and 10K euros. Thus, be that a glitch in a StopLoss taken after 3 minutes only, or be that missed revenue like yesterday (which I obviously can’t even count-in), the Strategy is fine and it would comply to Backtesting fairly accurate (if not 100%). So envision, glitching/slippage can not be backtested, so small differences occur from that angle. But losing something like 1K on a trade of 20K (which really happened on 3 trades something like 6 weeks ago) is way out of line and can’t be worked with (can not be calculated-in). So you see, your text (see quote) is correct in itself, but it is also something like n/a.

    This morning I suggested (towards PRT Support) that the missed revenue because of a happening like yesterday is no sh*t, which really is not something that is on people’s minds (e.g. your’s, GraHal). This probably or possibly is because you just don’t have the experience in not gaining money because of a system thrown out, but I could show from yesterday’s further trades what the missed revenue could have been when I coincidentally would be more serious with my Autotrading and say the $ per point. So what will we do when this one system thrown out could be proven to have missed a more than nice new car ? Do we say “oh well, next time another time” ?

    #208954

    you just don’t have the experience in not gaining money because of a system thrown out

    I was onmarket yesterday in Live (at the time / Feds you refer to) and so I experienced the opposite!

    So what will we do when this one system thrown out could be proven to have missed a more than nice new car ? Do we say “oh well, next time another time” ?

    NO, we use at Market Orders and put price levels etc (which would have been part of stop and limit orders) as part of Buy / SellShort conditions.   This combined with 1 min or less TF can get us into or out of trades fast and better than getting Rejected and missing that nice new car! 🙂

    I’ve done above with a few Systems in the last 10 days or so … since we have been discussing this Reject for minimum distance annoyance!

    1 user thanked author for this post.
    #208957

    NO, we use at Market Orders and put price levels etc (which would have been part of stop and limit orders) as part of Buy / SellShort conditions.   This combined with 1 min or less TF can get us into or out of trades fast and better than getting Rejected and missing that nice new car! 🙂

    Well, this obviously is a good idea theoretically. Sadly, if you backtest this – say with my system or systems – over and over it is proven that this loses a lot of money. With more time I could show it – and obviously it depends on the further construction of the ordering system – but it easily differs half of the profit. For talks let’s say that this is 10K in 6 months of any system with minimum inlay (read  toying but for real). Now doing it as you suggests for Entry makes 5K profit of that. I have tried to turn this for the better countless times, but it never works out for the better.
    Of course I can’t know what I miss for systems thrown out at the Entry, but let’s say that net the Limit order is better.

    And oh, for the Exit I am already doing that because I refuse to pay 0.5p on insurance to IG for having a guaranteed SL (this 0.5p is what it costs, no matter whether it is needed or not – the text on IG’s site as well as the help text in PRT is misleading “is only charged when needed”. This is to read as : is charged when the SL is hit).
    I think I indeed started doing that per the result of our both discussion (probably in this topic) something like 6 weeks ago.

    Anyway, sharing things like these even lead to solutions for the better !

    PS: I hope you were not Short in there yesterday ?

    1 user thanked author for this post.
    #208958

    Have you ever thought about choosing a market that has a small minimum distance from the outset? So instead of Dow and Nasdaq, something like SP500? Minimum distance 1pip and the correlation is US market.

    #208959

    But phoentzs, it looks like you missed the point ?
    IG tweaks the lot so they have some rest to arrange for things.

Viewing 15 posts - 166 through 180 (of 224 total)

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