ProOrder Trades not Triggering

• 12/05/2018 at 3:15 PM #86357
  gone_mexico

  Hi there,

  Can someone please help me understand why some trades don’t trigger on ProOrder when exactly the same code does trigger a trade (as expected) in Back Testing mode.

  Surely the same code should generate the same trades unless the broker rejects a trade for some reason.

  Any pointers would be much appreciated.

  Phil.

  

  gone_mexico

  Participant

  Topics: 4

  Replies: 8

  Been thanked: 0 times
  12/05/2018 at 3:22 PM #86359
  robertogozzi

  Custom trading hours might be the reason?

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 17782

  Been thanked: 4430 times
  12/05/2018 at 3:37 PM #86361
  Vonasi

  …or different spread in the live data compared to the fixed spread you use in back testing.

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1812 times
  12/05/2018 at 4:48 PM #86366
  gone_mexico

  I don’t think so. This happened at 9:06 (UK Time) on the 4th of December on the DAX.

  My systems were running and had already placed a trade correctly an hour earlier.

  Under ‘Time Zones and Trading Hours’ (for Indices – DAX) I have ‘Use Trading Hours of this Market’ checked. This is 00:00 to 00:00.

  I did wonder if the ProOrder servers were running on a different timeframe to the time frame being reported through the UI. My code does have logic which is based on the ‘Time’. Do you know what timeframe is used by the ProOrder servers and whether this is the time returned by the Time command? (rather than the time of the market or the time of my charts)

  

  gone_mexico

  Participant

  Topics: 4

  Replies: 8

  Been thanked: 0 times
  12/05/2018 at 4:51 PM #86367
  gone_mexico

  I don’t think it was the spread. My system should have place an order to sell the DAX at 11,403 when processing the 9:03 3 min bar.  i.e., just below the bar being evaluated.

  The next bar (9:06) opened at 11,408.7, so it should have triggered my order (11,403( when it moved down to close at 11,395.7 irrespective of the spread (which is usually .5 or 1 pt during core hours with IG).

  

  gone_mexico

  Participant

  Topics: 4

  Replies: 8

  Been thanked: 0 times
  12/05/2018 at 6:24 PM #86372
  robertogozzi

  TIME is the time when a bar closes and matches your TZ; as for DAX when you, in UK, open a trade when TIME shows 090000 you open a trade when its trading time is actually 100000.

  But, unless ProOrder and ProBackTest don’t use the same settings, this should not make their performances differ!

  1 user thanked author for this post.

  gone_mexico

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 17782

  Been thanked: 4430 times
  12/05/2018 at 8:11 PM #86376
  TempusFugit

  Just a few days ago I posted that something similar was happening to me:

  https://www.prorealcode.com/topic/differences-in-backtest-results/

  And just yesterday I sent a ticket to the PRT support  complaining about a trade in Live that shouldn´t have opened and as in backtested it didn´t. I lost quite a lot in this trade by the way.

  With this concern in mind I have now several systems with everything the same in Live and Demo at the same time to check the differences, and I can tell there are differences in the trades they make. Also with the backtest.

  All this makes me wonder if I can trust PRT backtest results. I read here people that says that this differences shouldn´t occur…. but have you checked as I am doing? Because I don´t think I am the only one that this happen (and gone_mexico also)

   

  1 user thanked author for this post.

  gone_mexico

  

  TempusFugit

  Participant

  Topics: 41

  Replies: 232

  Been thanked: 65 times
  12/06/2018 at 9:32 AM #86399
  Nicolas

  Always look at your orders list and check the failed orders. Most of the times, the problem is about pending orders placed at wrong price because of the minimum setup distance allowed by the broker.

  1 user thanked author for this post.

  gone_mexico

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20874

  Been thanked: 4801 times
  12/06/2018 at 11:50 AM #86425
  TempusFugit

  Nicholas, if you are talking about mininum stop distances I can tell you is not my case.  And anyway how can that explain the difference in trades between Live and Demo with everything the same?

  Have you tried to run the same system (with everything the same) in Live and Demo at the same time? I am not saying that all the trades are completly different,  but there are trades that don´t match, enough to worry me at least. For example, I just check one of the systems I am running in both Live and Demo, in the last month there are 3 trades that don´t match between them, from a total of 17 trades in the month. And for making it worse there are also differences in trades for backtesting in the same period for both Live and Demo!!!!

  

  TempusFugit

  Participant

  Topics: 41

  Replies: 232

  Been thanked: 65 times
  12/06/2018 at 12:00 PM #86428
  gone_mexico

  I’ve not experienced what you mention (trades triggering in live that didn’t trigger in back testing). That does sound scary.

  My only scenario was something not triggering when it should have done.

  I think my scenario was actually a school boy error on my part. Remember the max position size you set when you start up your systems in live? Well, I had set this two low it seems. My assumption is that this is what stopped my trades from triggering. This is a useful safeguard feature. However, I don’t like the way it’s implemented as you have no way of seeing that the safeguard has prevented you from taking a position that your code is trying to take.

  Will let you know if I still have issues after re-starting my systems with a higher parameter for the maximum position size.

  

  gone_mexico

  Participant

  Topics: 4

  Replies: 8

  Been thanked: 0 times
  12/06/2018 at 2:18 PM #86437
  Nicolas

  TempusFugit wrote:

  how can that explain the difference in trades between Live and Demo with everything the same?

  I’d like to give you a precise and definitive answer for each of your order difference. But it has already been talked many times, demo accounts are not as good as real accounts to make forward test, yeah I know that’s a pity, but that’s how it is! and for that specific cases, the only people to blame is the company that run your code on its own server, aka your broker .. and the market to not be as precisely replicated on a demo environment 🙂

  And for the backtests differences: spread, minimal distance, slippage, price rounding, latency, trailing stop step, etc. Every order in backtest is placed at the precise desired calculated level you want, that’s impossible in real time. And we are not talking about potential bugs and weird behaviors that tend to be dramatically reduced since members of the forums are making reporting to PRT, through the website and directly within the platform.

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20874

  Been thanked: 4801 times
• Author
  Posts