The difference in backtesting on PRT Demo / PRT Live

• 01/20/2019 at 1:04 PM #89142
  Shitforbrains

  Hi,

  Has anyone else experienced a big difference in backtesting the same code on PRT Demo vs PRT Live? Have a look at the pics below. The code is exactly the same, tested on the same amount of units and with the same spread. How can this be, and which one should I trust? Probably the losing one. 🙂

  

  Shitforbrains

  Participant

  Topics: 4

  Replies: 27

  Been thanked: 0 times
  01/20/2019 at 1:47 PM #89146
  GraHal

  You sure you have tick by tick mode selected on the winning curve?? 🙂

  1 user thanked author for this post.

  Shitforbrains

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
  01/20/2019 at 1:48 PM #89147
  Vonasi

  My first thoughts are that it looks like one is tested with tick by tick on and the other with it off?

  1 user thanked author for this post.

  Shitforbrains

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  01/20/2019 at 1:49 PM #89148
  Vonasi

  Also is there any time conditions in the strategy and if so are both platforms set to the same time zone?

  1 user thanked author for this post.

  Shitforbrains

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  01/20/2019 at 2:11 PM #89149
  Shitforbrains

  Ah you’re right, tick by tick was not “ticked” on the live one, time zones are the same on both. But it’s still behaving weirdly on the live version, banking profits that are clearly losses when I zoom in on the candlesticks. Any idea how to minimize these?

  

  Shitforbrains

  Participant

  Topics: 4

  Replies: 27

  Been thanked: 0 times
  01/20/2019 at 2:19 PM #89150
  Vonasi

  Shitforbrains wrote:

  Any idea how to minimize these?

  Not without the code so as to know what the strategy is!

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  01/20/2019 at 2:42 PM #89151
  Shitforbrains

  Oh yeah, the code. 🙂

  It’s a simple Heiken Ashi strategy that opens a position based on the change of color.

  This is used on 1h tiumeframe between 0900 and 1800. (Not the same code used from pics in first post)

  DefParam FlatBefore = 090000 DefParam FlatAfter = 180000 Defparam cumulateorders=false xClose = (Open+High+Low+Close)/4 if(barindex>2) then xOpen = (xOpen[1] + xClose[1])/2 endif c1 = xClose>xOpen AND xClose[1]<xOpen[1] c2 = xClose<xOpen AND xClose[1]>xOpen[1] if c1 AND Not LongOnMarket then BUY 1 SHARE AT MARKET EXITSHORT AT MARKET endif if c2 AND Not ShortOnMarket then SELLSHORT 1 SHARE AT MARKET SELL AT MARKET Endif SET STOP $LOSS 26 SET TARGET $PROFIT 24

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

  DefParam FlatBefore = 090000   DefParam FlatAfter = 180000   Defparam cumulateorders=false   xClose = (Open+High+Low+Close)/4   if(barindex>2) then xOpen = (xOpen[1] + xClose[1])/2 endif   c1 = xClose>xOpen AND xClose[1]<xOpen[1] c2 = xClose<xOpen AND xClose[1]>xOpen[1]   if c1 AND Not LongOnMarket then BUY 1 SHARE AT MARKET EXITSHORT AT MARKET endif   if c2 AND Not ShortOnMarket then SELLSHORT 1 SHARE AT MARKET SELL AT MARKET Endif   SET STOP $LOSS 26   SET TARGET $PROFIT 24

   

   

  

  Shitforbrains

  Participant

  Topics: 4

  Replies: 27

  Been thanked: 0 times
  01/20/2019 at 3:17 PM #89158
  Vonasi

  Shitforbrains – please try to remember to use the ‘Insert PRT Code’ button when posting code as it makes it much easier for others to read. I have tidied up your post for you. 🙂

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  01/20/2019 at 3:35 PM #89159
  Vonasi

  I can’t see any weird trades like you suggest – do you have any screen shots to share with us?

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  01/21/2019 at 4:30 PM #89263
  Jakub

  Dear all,

  I beleive the difference is prompted by the new ProOrder engine (Multi Timeframe) which has been released only on Demo for now (it is available upon request on Live as well).

  They have a different ProBacktest execution in which the new engine is always right.

  Regards,

  Jakub

  1 user thanked author for this post.

  Shitforbrains

  

  Jakub

  Participant

  ProRealTime

  Topics: 0

  Replies: 70

  Been thanked: 24 times
  01/23/2019 at 2:18 PM #89439
  Shitforbrains

  Vonasi wrote:

  I can’t see any weird trades like you suggest – do you have any screen shots to share with us?

  Now that I do the backtest and look at it side by side with the system running, I can’t find the faulty trades. Hopefully I was just looking at the wrong trades.

  

  Shitforbrains

  Participant

  Topics: 4

  Replies: 27

  Been thanked: 0 times
  03/26/2020 at 1:09 AM #123342
  VN

  Hi Jakub, is the difference in the Multi time frame engine you mention above still present today ie in version 10.3? I have been backtesting a Multi time frame strategy in demo and I want to know if there will be differences in how the same strategy performs with the same backtest parameters in prt live.

  

  VN

  Participant

  Topics: 5

  Replies: 51

  Been thanked: 0 times
  03/26/2020 at 9:20 AM #123351
  Nicolas

  No difference due to MTF. Jakub was talking of a time period where the two versions were existing for clients: one with new engine that embed the MTF and the old one, which is thing of the past now.

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20828

  Been thanked: 4741 times
• Author
  Posts