Machine Learning in ProOrder ProRealTime

• 03/12/2020 at 10:31 AM #121911
  GraHal

  Correction re my reply to Jan’s question

  GraHal wrote:

  adjust the setting of the variables automatically given the structure of price movements ?

  It is adjust the setting of the variables automatically in relation to the results of the last 3 trades?

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/12/2020 at 1:29 PM #121920
  GraHal

  GraHal wrote:

  Trade4 would be used together with Trade3 and Trade 2 to repeat above.

  Yes above is nearly correct … the last 3 Profitable Trades on a rolling basis (makes good sense) … due to code below at Lines 83 and 84 …

  For i = 1 to Reps Do If positionperf(i) > 0 Then

  1 2	For i = 1 to Reps Do If positionperf(i) > 0 Then

   

  1 user thanked author for this post.

  Jan

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/12/2020 at 7:19 PM #121988
  Jan

  Grahal,

  thanks a lot for your explanations.

  

  Jan

  Participant

  Topics: 10

  Replies: 90

  Been thanked: 29 times
  03/12/2020 at 11:38 PM #122007
  GraHal

  Does anybody know what the *100000 is doing in the code below …??

  I know positionperf is a %, but to then x by 100K it ends up making StratAvgA = figures that bear no relation to the strategy average profit which is more like 50 for Paul’s 1 second TF System?

  See the attached at the red arrowhead … 570 ??

  Then on the next trade  (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1) = – 810

  Are we sure the Halgo is working correctly??

  StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1)

  1	StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1)

   

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/13/2020 at 12:13 AM #122011
  Jan

  Good evening Grahal,

  As far as I understand it, StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1)  as coded in https://www.prorealcode.com/topic/machine-learning-in-proorder/page/3/  count the total profit of 3 trades of Strategy A and averages it ,

  but the 100.000 seems to much to me.

  You can graph PositionPerf to check whether you really need this fraction to multiply by 100.000

  By the way, in the code of Juanj I can not find a strategy at all, I miss StratAvgA

   

  Question of Grahal about Part of the Juanj code

  For i = 1 to Reps Do If positionperf(i) > 0 Then WinCountA = WinCountA + 1 //Increment Current WinCount EndIf StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1) Next StratAvgA = StratAvgA/Reps //Calculate Current Avg Strategy Profit //Graph (PositionPerf(1)*countofposition[1]*100000)*-1 as "PosPerf1"

  1 2 3 4 5 6 7 8

  For i = 1 to Reps Do If positionperf(i) > 0 Then WinCountA = WinCountA + 1 //Increment Current WinCount EndIf StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*-1)*-1) Next StratAvgA = StratAvgA/Reps //Calculate Current Avg Strategy Profit //Graph (PositionPerf(1)*countofposition[1]*100000)*-1 as "PosPerf1"

   

  

  Jan

  Participant

  Topics: 10

  Replies: 90

  Been thanked: 29 times
  03/13/2020 at 12:32 AM #122012
  Jan

   

  StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*–1)*–1)

  As Reps = 3 ,  is this line of code summing  the last 3 PositionPerf x resp. the last 3 count of position x 100.000 to the stratAvgA

  

  Jan

  Participant

  Topics: 10

  Replies: 90

  Been thanked: 29 times
  03/13/2020 at 10:16 AM #122033
  GraHal

  I would agree Jan that is what is intended, but surely the multiplier should be the instrument price (as in DJI 21,600 currently)?  Reason: positionperf is the % gain of postion(n).

  As an aside … this is interesting for me to see how you guys (far far better than me at coding) do not find it easy to interpret another author’s code?

   

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/13/2020 at 11:01 AM #122041
  GraHal

  GraHal wrote:

  Reason: positionperf is the % gain of postion(n).

  Correct myself again … when I use positionperf in the optimiser, I do always convert the fractional part (in my head) to a % to get a better feel for the gain relative to the TF … to gauge if it has reasonable expectancy to deliver on a regular ongoing basis.

  Anyyywayyyy … a positionperf = 0.1 when converted to money would be 0.1 x 25600 = £256 (if on DJI at 25600 at £1 per point and lot size = 1).

  So how / where the eff does the  x 100,000  get into below …

  StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*–1)*–1)

  1	StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*–1)*–1)

   

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/13/2020 at 11:12 AM #122043
  GraHal

  Jan wrote:

  in the code of Juanj I can not find a strategy at all

  A basic strategy is in the post below …

  Paul wrote:

  I created a basic strategy on a second timeframe for testing

  But below is a more complex strategy that began on this Topic but got split-off to it’s own Topic … best read the whole Topic to get the full picture?

  Paul wrote:

  for testing

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/13/2020 at 12:41 PM #122058
  GraHal

  GraHal wrote:

  would be 0.1 x 25600 = £256

  save anyone correcting me … that should have read … 0.01 x 25600 = £256 (1% of 25600)

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  03/13/2020 at 1:00 PM #122060
  juanj

  Hi Everyone, apologies for not being able to answer all the questions!

  I am a full time trader and as you can imagine the markets have kept me very occupied the last week.

  I will attempt to make some time tonight to address some of the unanswered questions here.

  @GraHal thank you for the assistance here. With regards to:

  StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*–1)*–1)

  1	StratAvgA = StratAvgA + (((PositionPerf(i)*countofposition[i]*100000)*–1)*–1)

  This quite simply calculates the “capital” value of the performance multiplying the points value with the position size * the index/forex decimal point value (so yes it can be different from one market to the next)

  100000 was used for the ZAF40 market

   

  2 users thanked author for this post.

  Paul, GraHal

  

  juanj

  Participant

  Topics: 51

  Replies: 505

  Been thanked: 170 times
  04/09/2020 at 1:45 PM #125433
  Bard

  Hi GraHal, I was wondering why Pauls’s RSI results were the same with and without heuristics though? Particularly as Franceso’s results improved a lot: https://www.prorealcode.com/topic/machine-learning-in-proorder/page/3/#post-121230

  

  Bard

  Participant

  Topics: 138

  Replies: 814

  Been thanked: 99 times
  04/09/2020 at 2:38 PM #125448
  GraHal

  The HAlgo is a bit tricky to get working correctly, maybe – in Paul’s version you refer to – the HAlgo wasn’t working?

  A few of us spent some time on it, but seems we put it on the back burner for a while? Other things  pulling on our time I guess?

  Be great if  you could come up with a System where it clearly makes a big difference? A single variable System may be good?

  The HAlgo can be duplicated and used on 2 (or even more variables)  but it needs some dedication to get it all hanging together fully functional.

  I did it with 2 variables and planned to try 3, but for no good or bad reason, my momentum slowed / fell by the wayside.

  

  GraHal

  Participant

  Topics: 211

  Replies: 8760

  Been thanked: 2176 times
  04/09/2020 at 3:08 PM #125451
  Bard

  Interesting results @Francesco! How do you measure / do a robustness test?

  

  Bard

  Participant

  Topics: 138

  Replies: 814

  Been thanked: 99 times
  04/09/2020 at 3:26 PM #125452
  Francesco

  Bard wrote:

  Interesting results @francesco! How do you measure / do a robustness test?

  Take a look here https://www.prorealcode.com/topic/day-month-year-strategy-robustness-tester/

  You have to read and study most part of the pages in order to understand how it works.

  Thanks to Vonasi’s work 🙂

  1 user thanked author for this post.

  Bard

  

  Francesco

  Participant

  Topics: 32

  Replies: 439

  Been thanked: 200 times
• Author
  Posts