Known differences between PRT LIVE and PRT DEMO

Forums › ProRealTime English forum › ProOrder support › Known differences between PRT LIVE and PRT DEMO

  • This topic has 35 replies, 7 voices, and was last updated 4 years ago by avatarVN.
Viewing 15 posts - 16 through 30 (of 36 total)
1 2 3
  • 03/26/2020 at 5:46 PM #123487
    GraHal

    Yes I agree and lots of us have had the same annoying occurrence.

    Weird stuff happens that unless we worked in IT at IG we prob will never know why?

    Is slippage a factor in your example … you will never know?

    I’ve been manual trading a lot today and about 6 times when I clicked to Exit (both Long and Short) I got the message … ‘That Price is no longer available‘. As my Exit instruction was at Market Price then I can’t help thinking there are algorithms at work behind the scenes and in my case of not being able to Exit was due to a flash (milliseconds) widening of the spread or the minimum distance parameter or some other IG machination?? 🙁

    In short … there isn’t always an answer (not visible to us anyway) and you may drive yourself  bonkers trying to find an answer??

    I just though of one thing though we had a while back, I worked on with one guy … the calculation of Indicators were not exactly the same on Backtest and AutoTrade / Forward Test.

    Have you Graphed all your conditions? Ah, but of course you cannot GRAPH on Forward Test anyway so that trail dies there??  🙁

    03/26/2020 at 5:50 PM #123488
    VN
    Nicolas wrote:

    In that specific case, of an order X not sent the day Y at time Z with the strategy W… I don’t know! In fact, only a deep investigation by an IT guy with the server logs would help.

    yes there are a lot of variables in the mix here so I appriciate its not easy to give an answer. Is there a technical support team that can look into such a query and be able to debug this?

    03/26/2020 at 6:09 PM #123491
    GraHal
    VN wrote:

    Is there a technical support team that can look into such a query and be able to debug this?

    You may like to consider that a past member of this Community was made – by IG – IG Account closed due to taking up too much of IG time relative to the amount of business (revenue to IG ) he was putting the way of IG.

    Also IG’s view generally is that the PRT Demo Account is just that … a Demo Account to get used to how it all works before we use real money.

    So Demo Account issues get very little priority; I guess that is why the past member got the boot … he was expecting priority response??

    03/26/2020 at 7:04 PM #123507
    VN
    GraHal wrote:

    Yes I agree and lots of us have had the same annoying occurrence.

    Weird stuff happens that unless we worked in IT at IG we prob will never know why?

    Is slippage a factor in your example … you will never know?

    I’ve been manual trading a lot today and about 6 times when I clicked to Exit (both Long and Short) I got the message … ‘That Price is no longer available‘. As my Exit instruction was at Market Price then I can’t help thinking there are algorithms at work behind the scenes and in my case of not being able to Exit was due to a flash (milliseconds) widening of the spread or the minimum distance parameter or some other IG machination?? 🙁

    In short … there isn’t always an answer (not visible to us anyway) and you may drive yourself bonkers trying to find an answer??

    I just though of one thing though we had a while back, I worked on with one guy … the calculation of Indicators were not exactly the same on Backtest and AutoTrade / Forward Test.

    Have you Graphed all your conditions? Ah, but of course you cannot GRAPH on Forward Test anyway so that trail dies there?? 🙁

    I dont think its slippage in the cases I have looked at, but I can accept that could be a reason. In the backtest mode you can debug what is happening using the graph function and plot each variable/indicator at any point in time to validate what is happening. In the autotrade mode there is no way for one to see what is happening unless you are a developer and have access to the inside platform.

    You mention that the calculation of indicators is different between Backtest and AutoTrade mode? Is this True, because if it is then this could be the reason for the differnce.

    03/26/2020 at 7:31 PM #123516
    GraHal
    VN wrote:

    calculation of indicators is different between Backtest and AutoTrade mode? Is this True,

    I mentioned one instance that appeared to be different.  Any anomaly needs to be looked at on a case by case basis.

    Btw … if you use the Quote button then you only need select the part  of the post that you are commenting on (as I have above).

    How come your post above got posted 4 or 5 times??

    03/26/2020 at 8:29 PM #123521
    deletedaccount100622

    I had the same issues, opening Ig.com and PRT they had different prices and both were intermittently locking up. Not even a higher volume day today

    03/26/2020 at 9:24 PM #123522
    robertogozzi

    Please try not to use “quote” extensively.

    It must be used to emphasize some words or a sentence at most.

    In any other case just post what you want to say.

    This will help readers to better follow the topic.

    Thank you 🙂

    03/26/2020 at 11:26 PM #123533
    VN

    my browser wasn’t responding so the quote post got posted multiple times in error..

    03/27/2020 at 10:23 AM #123561
    Nicolas
    VN wrote:

    calculation of indicators is different between Backtest and AutoTrade mode?

    No! Differences might occur between indicator (ProBuilder) and automatic trading program (ProBacktest / ProOrder), because they use different engines. While it is rare, the best way to match your custom indicators are calculated the same between them is by using GRAPH or GRAPHONPRICE in ProBacktest. If the backtest match the indicators applied on the chart, then you can be reassured that ProBacktest/ProOrder will calculate your custom codes the same as ProBuilder.

    You are right about the fact that a log of what’s happening is crucial and this is also one of my priority development requests, but it requires the recovery of data from the broker on what is really going on with the orders sent by the platform, hope that this will be possible soon.

    1 user thanked author for this post.
    avatar GraHal
    03/27/2020 at 3:53 PM #123622
    VN

    @Nicolas, merci again.

    For all intents and purposes the current set up the current AutoTrading functionality behaves like a black box . Visibility of what is happening for example being able to see the underying variables/indicators in real time is an absolute must for anyone serious enough to risk your own money on it so I applaud your effort in getting this developed.

    I am doing more testing as I write this and as I run the strategy in backtest and AutoTrading side by side, differences keep propping up. The AutoTrade seems to be opening positions whislt the backtest is NOT. This is not happening all the time but from time to time. From what I am able to see from graphing the variables used to evaluate the entry conditions on my backtest version (just as a proxy of what should be, but clearly isn’t happening with the AutoTrade version) is that there appears to be what I can imagine a time difference. I am running a 1 min candle chart and I can see from the variables that the backtest shows “no trade entry” at the exact point in time that the AutoTrade sent an order to the broker to fill which IG did so. One minute earlier however, from the backtest variables plotted I can see that it was “OK” to enter into a trade. If I use this example, then I can only think that the indicator variables calcualted by the AutoTrade code (which I cannot see) must be ever so slightly different to what the backtest is showing to produce the effect of a “time lag”. I just don’t know how I can move from here, as I would like to put the AutoTrade functionality to work with real money, but if I cannot explain part of its behaviour then how can I.

    03/27/2020 at 4:45 PM #123626
    GraHal
    VN wrote:

    One minute earlier however, from the backtest variables plotted I can see that it was “OK” to enter into a trade.

    Did you really mean 1 minute earlier (or 1 minute later) ??

    03/27/2020 at 11:27 PM #123665
    VN

    @Grahal, (and @Nicolas) on the particular example I picked the condition to enter into a trade (based on a number of indicators) was FALSE under the backtest, but it appears under the AutoTrade version it must have been TRUE hence why a trade entry was made. On the previous 1 min candle, the backtest showed me that the condition to enter a trade was TRUE before it became FALSE on the next candle which is when the trade entry was made…

    The entry conditions are made based on a number of indicators meeting certain paramaters. Without being able to see what those indicator values are are at end of each candle in the AutoTrading mode (just like one can see them on the Backtest engine by graphing them as they get calculated) it will be impossible to determine what is really happening. I am fairly certain however that for there to be differences in entry conditions (or exit for that matter) between Backtest and AutoTrading it will be down to different indicator closing values being produced as the code is the same between the two. The differences will most likely be driven from either differences in the indicator calculation methodology between AutoTrade and Backtest modes, differences in the market data (different prices) feeding those engines or differences in the timings of when that market data is fed to AutoTrade and Backtest calculation engine.

    I would expect a PRT developer to be able to validate if there is any truth in that last statement or at least explain how it works between the two modes. Is there anyone on these forums who has this level of understanding and can opine on this and in process help educate us all it would be most appriciated?

    03/28/2020 at 10:17 AM #123675
    GraHal
    VN wrote:

    On the previous 1 min candle, the backtest showed me that the condition to enter a trade was TRUE before it became FALSE on the next candle which is when the trade entry was made

    I would find it easier to read / understand if you use the actual times of True and False.

    Short one line statements / bullets help loads also when it is a critical point you are making.

    Are you  saying on Backtest that conditions were True (for example) at 09.46 but trade entry was 09:47?

    If Yes to above, then that is how PRT works … trades are executed at the open of the bar following the bar when conditions are True.

    03/28/2020 at 2:50 PM #123717
    VN

    I didn’t want to go into specifics on the timings as it wouldn’t change the fact that the AutoTrade mode is making trade entries at points when the Backtest running the same script concurrently to the AutoTrade script in real time is not. That is the main point here, the two modes are giving different results, and from the variables I can see from the backtest graph function I can validate that the backtest is doing the right thing. I cannot see what is happening with AutoTrade and why the trade entry was made, but knowing that the script looks at the number of indicator closing values it must be that those indicator closing values are different to what i can see from the backtest mode. That said here is the specific example i gave with actual candle times, the script checks for certain conditions (6 of them to be TRUE) before a trade entry is made:

    13:05:59

    Backtest – 5 of the 6 entry conditions are met to enter into a trade. No trade entry however is made as the script requires all 6 conditions to be TRUE for that to happen

    AutoTrade – have no visibility of what is happening at the end of the 13:05 min candle, but i expect that whatever it has calculated is different to Backtest, as on the open of the 13:06 candle there is a trade entry on my IG account, which makes me think that the 6th indicator which was FALSE in the backtest was probably calculated to be TRUE under the AutoTrade engine thus making 6 out of 6 conditions TRUE and thus entering into a position.

    13:06:59

    Bactest shows no entry based on the close of the indicators

    AutoTrade has nn open position running

     

     

    • 13:06:00 trade entry (I have no visibility as to why, but based on the Backtest I can see
    03/28/2020 at 2:54 PM #123719
    VN
    VN wrote:

    ignore the last bullet line “13:06:00 trade entry (I have no visibility as to why, but based on the Backtest I can see” from last message..

  • Author
    Posts
Viewing 15 posts - 16 through 30 (of 36 total)
1 2 3

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
prt personal account vs prt ig account
differnce between live & demo
little discrepancy between the code and the default prt pivot points
prt backtest in demo is very slow
prt v12 ig demo is here
6 jan 24 prt not available – too long to respond
can’t see procode or screeners in prt 11.
convert from tradingview to prt
converting twin range filter indicator from trading view to prt
create windows shortcut to start prt/ig

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