Difference in live trading and simulation

• 02/02/2017 at 2:07 PM #23601
  tob107

  Hi,

  I am fairly new at this, so I just made a simple MA crossover to start with. The results were great so I took in live yesterday.

  This morning the algorithm took its first position, which was a loss of 11,4 points. But when simulating (or back-testing) the algorithm now, ProOrder indicates that the trade was a profit of 8,7 points. Please see the attached screenshots.

  The spread trough IG is 2 points, so that should be the issue.

  Can anyone please explain this?

  Thanks in advance!

  Tobias

  

  tob107

  Participant

  Topics: 7

  Replies: 22

  Been thanked: 15 times
  02/02/2017 at 2:21 PM #23608
  Joachim Nilsson

  Have you added spread in your backtest? Have you tested it with PRT v10.3 (tick by tick)? Have you become a victim of the “0 bars issue”?

   

  

  Joachim Nilsson

  Participant

  Topics: 4

  Replies: 62

  Been thanked: 18 times
  02/02/2017 at 2:22 PM #23609
  Nicolas

  Did you made tick backtests with version 10.3? Could you please share the code? Otherwise, it would be difficult to help more 🙂

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20828

  Been thanked: 4738 times
  02/02/2017 at 2:55 PM #23622
  tob107

  Hi,

  The code is found below.

  I use PRT 10.2, so I haven’t checked the tick by tick back-test.

  The spread is included in the simulations. What is the “0 bars issue”? Perhaps that could be it since it closed on the same candle.

  // Definition of code parameters DEFPARAM CumulateOrders = False // Cumulating positions deactivated DEFPARAM FlatBefore = 071500 DEFPARAM FlatAfter = 215500 indicator1 = Average[34](close) indicator2 = Average[36](close) NoC = CurrentMonth - 1 //Adjust for February //indicator3 = Stochastic[14,3](close) // Conditions to enter long positions c1 = (indicator1 CROSSES OVER indicator2) IF c1 THEN BUY NoC CONTRACT AT MARKET ENDIF // Conditions to exit long positions c2 = (indicator1 CROSSES UNDER indicator2) IF c2 THEN SELL AT MARKET ENDIF // Conditions to enter short positions c3 = (indicator1 CROSSES UNDER indicator2) IF c3 THEN SELLSHORT NoC CONTRACT AT MARKET ENDIF // Conditions to exit short positions c4 = (indicator1 CROSSES OVER indicator1) IF c4 THEN EXITSHORT AT MARKET ENDIF // Stops and targets SET STOP pTRAILING 2

  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 29 30 31 32 33 34 35 36 37 38 39 40

  // Definition of code parameters DEFPARAM CumulateOrders = False // Cumulating positions deactivated DEFPARAM FlatBefore = 071500 DEFPARAM FlatAfter = 215500   indicator1 = Average[34](close) indicator2 = Average[36](close) NoC = CurrentMonth - 1  //Adjust for February //indicator3 = Stochastic[14,3](close)   // Conditions to enter long positions c1 = (indicator1 CROSSES OVER indicator2)   IF c1 THEN BUY NoC CONTRACT AT MARKET ENDIF   // Conditions to exit long positions c2 = (indicator1 CROSSES UNDER indicator2)   IF c2 THEN SELL  AT MARKET ENDIF   // Conditions to enter short positions c3 = (indicator1 CROSSES UNDER indicator2)   IF c3 THEN SELLSHORT NoC CONTRACT AT MARKET ENDIF   // Conditions to exit short positions c4 = (indicator1 CROSSES OVER indicator1)   IF c4 THEN EXITSHORT  AT MARKET ENDIF   // Stops and targets SET STOP pTRAILING 2

  I know it looks weird with the narrow span on the MA’s and tight stop loss, but since the results were great, I couldn’t resist trying it live.

  Cheers,

  Tobias

  

  tob107

  Participant

  Topics: 7

  Replies: 22

  Been thanked: 15 times
  02/02/2017 at 4:20 PM #23631
  Nicolas

  You’ll find many topics about the “0 bars” issues around here. Since version 10.3, prorealtime embed a new backtest engine that solve this behaviour of the way trades were considered if the target and stoploss price levels were in the same bar.

  I made a video about it: https://www.prorealcode.com/blog/video-tutorials/tick-tick-backtest-engine-probacktest/

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20828

  Been thanked: 4738 times
  02/02/2017 at 8:14 PM #23687
  Henrik

  it Seems to be the trailing stop set to 2 in code but 10 in live.

  Regards

  Henrik

  

  Henrik

  Participant

  Topics: 9

  Replies: 97

  Been thanked: 19 times
  02/02/2017 at 8:30 PM #23690
  Joachim Nilsson

  If you´re trading DAX I think the smallest possible trailing stop you can use is 5points

  

  Joachim Nilsson

  Participant

  Topics: 4

  Replies: 62

  Been thanked: 18 times
  02/03/2017 at 9:38 AM #23746
  tob107

  Hi,

  Thanks for your replies. I thought the trailing stop at 2 points seemed funny, but that was the result I got when optimizing it. I’ll adjust it and try again.

  Henrik, do you have any explanation for why the stop is 10p instead of 2p?

  So basically, the backtest can’t be trusted in PRT 10.2 without tick by tick backtest?

  Tobias

  

  tob107

  Participant

  Topics: 7

  Replies: 22

  Been thanked: 15 times
  02/03/2017 at 1:09 PM #23774
  Joachim Nilsson

  You can´t even trust v10.3 tick by tick backtest but it´s better than 10.2

  

  Joachim Nilsson

  Participant

  Topics: 4

  Replies: 62

  Been thanked: 18 times
• Author
  Posts