Pro Order. Server error message
Forums › ProRealTime English forum › ProRealTime platform support › Pro Order. Server error message
- This topic has 9 replies, 3 voices, and was last updated 4 years ago by Nicolas.
-
-
02/05/2020 at 4:48 PM #118756
Hi,
I have been receiving ” Server. error. strategy. probacktest. error. missing_data” message when attempting to run several of my more complicated backtests in Pro Order, though able i have being able to run simple backtests in Pro Order.
Thanks
02/05/2020 at 5:15 PM #11876002/05/2020 at 6:02 PM #118768Hi Nicolas,
I’m running version 10.3. I have 10+ backtests running in pro order and was re-launching backtests individually after doing minor modifications to some of them in the code. So it is possible to stop an functioning backtest but unable re launch the identical backtest again. Also slow backtesting over the past days.
Thanks
02/06/2020 at 9:36 AM #118802This error message is not clear because it hasn’t been translated into a human comprehensible language 😉 PRT is deploying new error messages system and some of them are still not yet associated with translation in each language.
Anyway, this error is about something wrong server side and you should send a technical report about it by using the report tool you can find in the help section menu of the software.
02/06/2020 at 10:36 AM #11881302/12/2020 at 2:53 PM #119426Hi – I have the very same, but for V11.1 – 1.8.0_45.
Please notice : this is a Paper account for PRT(/IB) itself (not IG).
Also notice : I don’t post this to solve my problem (I suppose I may work it out) BUT similarities may show with the original post/problem.The Strategy runs fine in BackTesting, but it fails for just running the Strategy “live” (Automatic Trading) as soon as it enters the Market.
I am pretty sure it is related to a recently added Indicator which in itself is simple :
1234567891011121314DTSignal = 0MyDT01 = ExponentialAverage[DT01Avg](Close) + 0.0001MyDT02 = ExponentialAverage[DT02Avg](Close) + 0.0001If MyDT02 crosses over MyDT01 then // Black over Red. Up-spike indicates Exit market for Long.DTSignal = 0.5endifIf MyDT01 crosses over MyDT02 then // Red over Black. Down-spike indicates Exit market for Short.DTSignal = -0.5endifRETURN DTSignal as "DTSignal"This code is called like this :
12DT01AVG = 16 // 10-02-2020,PS.DT02AVG = 32 // 10-02-2020,PS.1DTSignal = CALL "DTTrigger"[DT01AVG,DT02AVG] // 12-02-2020,PS, For Testing !(the parameters default to 11 and 10 respectively)
and putting this on top of my code it does NOT go wrong. However, when the order enters the Market (I just attempted a Short) it fails immediately. Please notice that the call at the top of the code remained in there for dummy, but that at other places it is called for real and that I don’t even know whether it is called.
1234if not JustTraded thenDTSignal = CALL "DTTrigger"[DT01AVG,DT02AVG] // 09-02-2020,PS.if LongOnMarket then...12if not MustSell thenIf HasBeenHigh and DTSignal > 0 thenI expect that something just conflicts somewhere and that it may relate to the parameter naming or something like that. Ragarding this, please notice that the same Indicator(s) also run separately for visuality :
DT01 (parameter defaults to 18):
12AVG = DT01AvgRETURN ExponentialAverage[AVG](Close) + 0.0001 as "DT01" // This one is running ahead of DT02.DT02 (parameter defaults to 28):
12AVG = DT02AvgRETURN ExponentialAverage[AVG](Close) + 0.0001 as "DT02" // This one is running behind DT01.Key could be that the DT01Avg and DT02Avg are being used everywhere. Thus, in the “visuality” Indicator(s) (DT01/DT02), in the calculating Indicator (DTTrigger), thus also in the Parameters of each of them and in the calling main code.
Crucial difference between the Paper program and the Backtesting program, is that the same variables are there (Backtesting) used as parameters again for varying the Backtests. If this would be the key to success somehow, then obviously *that* can not be done in the real program (never mind it runs in Paper mode).
Lastly, I just tried the code with commented-out related lines (this is the assignment of the DT01Avg and DT02Avg variables plus the call to DTTrigger.
The triggering line of code was also deactivated like this :123456789if not MustSell thenIf 0 then // HasBeenHigh and DTSignal > 0 then // 12-02-2020,PS, Deactivated for Test.MustSell = 1if graph0 thenTrigger = Trigger + 0.0002//graphonprice(((Trigger))) COLOURED (255,0,255) as "Sell to exit market because HasBeenHigh and Gain Trend urges for Trail-up"endifendifendif(with the note that many of these same sequences of code exist under here, with the difference of the + 0.0002 being 0.0003 etc.)
and further down the line :
123if MustSell then // 28-05-2018,PS, This If instead of the combined sh*t.Sell at Market...And now the code just works.
Please remember that the Call to the indicator can be on top of the program and that it works for each bar(-call).Hints are welcome of course because testing is crazily tedious (wait for ages until your strategy triggers).
And as said, I hope it helps the OP seeing similarities and find his problem.Best regards,
Peter02/12/2020 at 4:59 PM #119451Just a follow up on the strategy error message, I have had a reply to the problem from PRT support and it seems to be simply just not enough preload bars in some of my strategies . I think i had my backtests set at 5000 and it has been recommended to set at 10000 and now all is working. I guess it can be easy to miss especially if you make custom indicators involving long periods and then use the Call function in the strategy. Though there is/was a specific error message about insufficient amount of preload bars.
02/13/2020 at 12:09 PM #119552All right – I found my issue too …
123if not JustTraded thenDTSignal = CALL "DTTrigger"[DT01AVG,DT02AVG]EndifThe above is the part which is not allowed … it is the conditional thing. So I thought to be smart and only apply the call when it is necessary. But *that* does imply error.
It thus seems that any indicator call is only allowed to be unconditional in the program. Say at the top of it.Can anyone confirm this ?
On a side note it would be logical to have it unconditional, assumed it it is to graph something (otherwise “holes” would emerge in the graph). But I am not graphing this Indicator …
Peter
02/13/2020 at 12:19 PM #11955402/13/2020 at 1:30 PM #119568 -
AuthorPosts
Find exclusive trading pro-tools on