ProBacktest shows different result

• 07/14/2020 at 10:09 PM #139287
  George65

  Hello,

  I was facing an issue today. I ran probacktest and automatic trading simultaneously. I was trading with US tech 100. Spread was set 2 points.

  Here is an example, an order which is showing very different result.

  Can someone help me, please  why is it? I think that testing should show about the same result (1-2 points could be tolerated).

  What do I miss here?

  Thanks

  George

   

  Probacktest					ProOrder Autotrading
  Entry date	exit date	type	nbr bars	abs perf	Entry date	exit date	type	nbr bars	abs perf
  14Jul2020 15:00:00	14Jul2020 15:00:00	long	0	GBP104,7	14Jul2020 15:00:01	14Jul2020 15:02:01	long	2	USD5,01

  

  George65

  Participant

  Topics: 11

  Replies: 48

  Been thanked: 2 times
  07/15/2020 at 9:37 AM #139315
  Nicolas

  Backtests are not suffering from real market conditions, spread and slippage could have some impact to the final results.

  I can’t help more without the code you are using, there might be things to improve into it.

  

  Nicolas

  Keymaster

  Topics: 83

  Replies: 20898

  Been thanked: 4866 times
  07/15/2020 at 12:22 PM #139331
  George65

  Hello Nicolas,

   

  First of all thanks for your reply! Of course I do understand, that there must be differences between test and real environment.  But you make your decision based on backtest results whether you let the algo run or not. If you have an algo that makes consistent profit in the test, but it can not make the same in the real life, then why is the backtest functionality? How can you optimize your algo, if test is not running like real life? Sorry, I understand, that it is a complicated issue from developers point of view. But what about a customer?

  First I made a mistake that I did not set the spread. Now I have a two point spread. In NYSE opening hours there is a 1 point spread in US100. So theoretically test should make worse results, right?

  Thanks for your understanding!

  best

  George

   

   

   

  

  George65

  Participant

  Topics: 11

  Replies: 48

  Been thanked: 2 times
  07/15/2020 at 1:05 PM #139339
  GraHal

  George65 wrote:

  14Jul2020 15:00:00 14Jul2020 15:00:00

  You sure above is correct … open and close in same second and makes £105??

  I guess it was a spike that retraced by the time the trade opened in REAL Live at 14Jul2020 15:00:01 ??

  

  GraHal

  Participant

  Topics: 222

  Replies: 9291

  Been thanked: 2460 times
  07/15/2020 at 1:13 PM #139342
  George65

  Hey Grahal,

   

  Thanks for your question. It can be a spike…I collected these data from PRT. Have to tell you, that all the closing dates are the same as opening dates for each order (if I remember well, can not check at the moment).

  Best

  George

  

  George65

  Participant

  Topics: 11

  Replies: 48

  Been thanked: 2 times
  07/15/2020 at 1:19 PM #139348
  GraHal

  Have you got Tick by Tick Mode enabled?

  

  GraHal

  Participant

  Topics: 222

  Replies: 9291

  Been thanked: 2460 times
  07/15/2020 at 3:39 PM #139364
  George65

  Thanks GraHal, very good question again :)! It was not enabled, because I got a warning message, to relaunch…

  Since I do not use target price in the code, only trailing stop (so there is no sequence issue)I meant that tick by tick is not necessary. But I can see that tick by tick mode is much closer to reality.

  Thanks for all of you for your support!

  Any tip I should pay attention to?

  Thanks again!

  Best

  George

  1 user thanked author for this post.

  GraHal

  

  George65

  Participant

  Topics: 11

  Replies: 48

  Been thanked: 2 times
• Author
  Posts