BackTest Profit not consistent

• 02/26/2022 at 3:22 PM #188911
  PeterSt

  Dear All,

  Would anyone among you be able to explain how the BackTest Optimisation result (which is dummy in this example) does not resemble the Statistic coming from the BackTest period ?
  Please compare the 3209 with the 4438. I added up the individual profits per trade, and they resemble the 4438.

  Since not all that long (couple of weeks ?) I also see the results not being consistent; the Statics screen may be, but the Optimalisation form varies from run by run. This week it grabbed my attention more than before and I had the idea that the signal being “live” would influence, as in : the calculation is finished, but during the build of the result itself for the chart (EUR/USD) another trade is performed which influences the Statistic – not the Optimalisation any more. But today all is closed and it still happens.
  During the writing of this I also notice the number of trades not being equal. 18 vs 17. Plus one perceived loss can’t be 1200 (although it would be close).

  I would kind of bet that this happens since the same moment (upgrade ?) that the “level” you see in the 3rd attachment actively moves up and down (it tracks a virtual gain) while in such case it concerns a 1M historical backtest (result). So, nothing oughts to move there as the 1M chart won’t test forward as well. That does not bother me really (though it reflects the end result of the 1M backtest) but point is : a. it seems to happen since the issue of this topic and b. it looks very similar for the happening itself (the both end results (Optimisation and Statistic plus in this case the Chart) become inconsistent.

  If there is an explanation to this (great feature and such) then OK. But if not I should report it as a bug. And regarding this :

  We should notice (and PRT should know) that very often we don’t want to await the result of the Statistic, because the Optimisation tells all right away – while waiting some more precious minutes for the Statistic is a waste of time  – WHILE the Statistic seems to show the correct result.
  And what about the Optimisation to begin with ? it is quite hard to check the relative results (when not dummy).

  Thank you guys,
  Peter

  

  PeterSt

  Participant

  Topics: 70

  Replies: 2284

  Been thanked: 692 times
  02/27/2022 at 11:25 AM #188949
  PeterSt

  Via this means – the other way around – it shows more easily what actually is “lacking”;

  Here it is easy to see that the one losing trade (look at date/time) is not incorporated in the BackTest result (which has 100% profitable trades).
  Same conclusion as in the first post : the Statistics show reality; the BackTest result form lacks.

  I have this continuously (now I know how to look at it). It can well be that this is since I work with Limit orders (100% of cases) for entering the market …

   

  

  PeterSt

  Participant

  Topics: 70

  Replies: 2284

  Been thanked: 692 times
  02/27/2022 at 11:58 AM #188953
  GraHal

  PeterSt wrote:

  this happens since the same moment (upgrade ?) that the “level” you see in the 3rd attachment actively moves up and down (it tracks a virtual gain)

  Last Sunday I got a weird happening, may be similar to what you are writing about here??

  I was watching the results of optimisation appear in the Table and, for example, 4056 appeared as the top gain and then as more results appeared, the 4056 dissapeared and a lesser gain, for example, 3198 took its place as the top gain.

  I double and treble checked and I did have Gain selected and sorted as the top gain first.

   

   

  

  GraHal

  Participant

  Topics: 203

  Replies: 8389

  Been thanked: 2034 times
  02/27/2022 at 5:24 PM #188959
  PeterSt

  And I had EXACTLY the same. Including checking for the sort order. I can’t guarantee it, but I would say that this was last Sunday indeed.

  1 user thanked author for this post.

  GraHal

  

  PeterSt

  Participant

  Topics: 70

  Replies: 2284

  Been thanked: 692 times
• Author
  Posts