How to find good value when using WF optimization?

• 04/25/2020 at 9:35 PM #128090
  Moda

  ok thanks @Francesco, i will double check

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  04/25/2020 at 9:37 PM #128091
  Moda

  full output attached…

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  04/26/2020 at 10:47 AM #128171
  Francesco

  In hundreds of rts that ive done, i’ve never seen a 20 5 and 21 5 having an higher score then 10 10.
  That’s why i told you to check it better and told that it was unusual @nonetheless

  Anyway, this could be an ecception, maybe synonymous of a good system 🙂

  3 users thanked author for this post.

  nonetheless, Paul, Moda

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  04/26/2020 at 1:50 PM #128203
  nonetheless

  Francesco wrote:

  never seen a 20 5 and 21 5 having an higher score then 10 10

  thanks Francesco. I never use 20,5 so wasn’t aware of any comparison between that and 10,10.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1235

  Been thanked: 978 times
  05/02/2020 at 6:28 PM #129480
  Moda

  Guys,

  Need eyes here, is this too much unbalanced results ?

  Cheers

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  05/02/2020 at 7:02 PM #129484
  nonetheless

  Well, that truly is an unusual result. Either you’ve written the perfect code or something has gone wrong in the VRT.

  Have you got this Excel sheet? #107211

  It automatically graphs the results so it’s easier to read.

  1 user thanked author for this post.

  Moda

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1235

  Been thanked: 978 times
  05/02/2020 at 7:07 PM #129486
  Moda

  Yes using the last version made by @Paul

  Here it is and the code is yours i’m working on MoD Dax/Dow

   

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  05/02/2020 at 7:14 PM #129489
  Francesco

  Having no results in the third one means both gain and % curve went negative, am i right?
  Also, the 100% gain curve in the 10 10 is nearly impossible, especially when in 20 5 and 21 5 go negative.
  There’s something inusual for sure 🙂

  1 user thanked author for this post.

  Moda

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  05/02/2020 at 7:16 PM #129490
  nonetheless

  Did you sort the average gain from low to high? It looks very peculiar. Almost impossible to get a 100% score there.

  If it’s my algo, then it’s definitely wrong! These are the results I got for DAX v2.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1235

  Been thanked: 978 times
  05/02/2020 at 7:22 PM #129494
  Moda

  Obviously, thanks for your time @Francesco

  It looks suspicious… for the 21/5 i did not run it

  Yes i sorted results … i found out sorry for the noise i made a mistake sorting data

   

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  05/02/2020 at 7:26 PM #129496
  Francesco

  If you want to share your code i can take a look for you

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  05/02/2020 at 7:32 PM #129497
  Moda

  sure i post it in the right place where it belongs in the MoD post #129503

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
  05/02/2020 at 8:14 PM #129506
  Francesco

  The problem was in the robustness tester code, replace it with this

  //Robustness Tester once j = 0 once flag = 1 if flag = 1 then j = j + 1 if j > qty then flag = -1 j = j - 1 endif endif if flag = -1 then j = j - 1 if j = 0 then j = j + random flag = 1 endif endif tradeon = 0 if opendate >= 20000101 then if barindex mod qty = 0 or barindex mod qty = j then tradeon = 1 endif endif

  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

  //Robustness Tester once j = 0 once flag = 1   if flag = 1 then j = j + 1 if j > qty then flag = -1 j = j - 1 endif endif   if flag = -1 then j = j - 1 if j = 0 then j = j + random flag = 1 endif endif   tradeon = 0 if opendate >= 20000101 then if barindex mod qty = 0 or barindex mod qty = j then tradeon = 1 endif endif

  I just did the 10 10 and the results are “normal”, so also 20 5 and 21 5 will surely work 🙂

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  05/02/2020 at 8:17 PM #129507
  Moda

  Many thanks @Francesco  😉

  

  Moda

  Participant

  Topics: 1

  Replies: 16

  Been thanked: 10 times
• Author
  Posts