Discussing the strategy VECTORIAL DAX (M5)

• 10/04/2019 at 10:05 PM #109317
  fifi743

  the backtest starts on the 27th as in the picture.

  I will explain you in the weekend but in French because I am bad in English

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/04/2019 at 10:06 PM #109318
  Vonasi

  fifi743 wrote:

  I will explain you in the weekend but in French because I am bad in English

  No French in the English speaking forum please. It’s a forum rule! 🙂

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1811 times
  10/05/2019 at 6:49 AM #109322
  fifi743

  Good morning all. For those who have unpleasant remonstrances you keep them for you. I share my work to help you and a piece of code can serve you. To start I calculated the maximum return with a variable for the long and a variable for the short Line 21

  PP=positionperf(0)*100 if pp>ppf then ppf=pp endif

  1	PP=positionperf(0)*100 if pp>ppf then ppf=pp endif

  After I left the maximum yield for the long and I reduced the value to increase the gain

  if longonmarket and ppF>3.6 and pp>ppF-pp then sell at market endif if shortonmarket AND ppF>1.9 and pp>ppF-pp then exitshort at market endif

  1 2 3 4 5 6

  if longonmarket and ppF>3.6 and pp>ppF-pp then sell at market endif if shortonmarket AND ppF>1.9 and pp>ppF-pp then exitshort at market endif

  afterwards, I seek to reduce the losses and to maximize the gains while increasing the gains, the ratio loss gains and the% of winning position. It’s the longest and the most complicated. And I realized that line 45

  CondBuy2 = (pente > trigger) AND (pente < 0)

  1	CondBuy2 = (pente > trigger) AND (pente < 0)

  Ask me a problem so I looked for a filter to limit the long position openings

  // ajout d'un filtre pour achat MA=Average[4](close) CondBuy2 = (pente > trigger) AND (pente < 0) and high>MA and coeff>0.9

  1 2 3

  // ajout d'un filtre pour achat MA=Average[4](close) CondBuy2 = (pente > trigger) AND (pente < 0) and high>MA and coeff>0.9

  And I continued and not finished reducing losses and maximizing gains.

  If you have ideas I am a taker to avoid writing hundreds of lines.

  Next coming I will pass it in MTF In M 5 the code you know In M1 the verification of operations if they are well past.

  Because often backtest no problem but in reality it’s something else. Because all operations must pass. I’m waiting for your comment.

  Sorry it’s the translator

  6 users thanked author for this post.

  capgros, Edmond, josef1604, Paul, reb, Balmora74

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/05/2019 at 8:14 AM #109323
  AE

  Hi fifi743,

  thanks again for your idea.

  did you check your strategy in Demo or Live? Have you got the same results than in backtesting?

   

  1 user thanked author for this post.

  Gubben

  

  AE

  Participant

  Topics: 11

  Replies: 53

  Been thanked: 12 times
  10/05/2019 at 9:30 AM #109327
  fifi743

  Hello, no I did not test it because it is not finished.

  it’s like jumping in the water when you can not swim

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/05/2019 at 11:50 AM #109333
  Paul

  Hi fifi743

  You use a lot of market orders, which are influenced by slipppage & spread on every trade.

  Isn’t it more reliable (and perhaps not better) to define your price yourself, i.e. the close when you have a signal and then enter on limit else on stop or visa versa on the next bar?

  All is depended on the positionperformance, based the price you get from the broker. It should be based on the price you set yourself to work with regardless of slippage.

  Maybe it doesn’t matter, maybe a lot I don’t know. What do you think?

  Here’s a snippet I use.

  If orderhandling=2 then you have to set your own orderprice after a signal, and replace all tradeprice(1) to orderprice (only in your code the trailingstop)

  if orderhandling=1 then pperf=(positionperf(0)*100) orderprice=tradeprice(1) //graphonprice orderprice elsif orderhandling=2 then if longonmarket then pperf=((close/orderprice)-1)*100 elsif shortonmarket then pperf=((orderprice/close)-1)*100 endif //graphonprice orderprice elsif orderhandling=3 then //manual orderprice=12370.7 if longonmarket then pperf=((close/orderprice)-1)*100 elsif shortonmarket then pperf=((orderprice/close)-1)*100 endif //graphonprice orderprice endif

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

  if orderhandling=1 then pperf=(positionperf(0)*100) orderprice=tradeprice(1) //graphonprice orderprice elsif orderhandling=2 then if longonmarket then pperf=((close/orderprice)-1)*100 elsif shortonmarket then pperf=((orderprice/close)-1)*100 endif //graphonprice orderprice elsif orderhandling=3 then //manual orderprice=12370.7 if longonmarket then pperf=((close/orderprice)-1)*100 elsif shortonmarket then pperf=((orderprice/close)-1)*100 endif //graphonprice orderprice endif

   

   

   

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  10/05/2019 at 1:36 PM #109344
  fifi743

  thank you paul i will test

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/05/2019 at 5:16 PM #109357
  Paul

  I did implement it, hopefully correctly

  the code is still very good!

   

   

  1

   

  4 users thanked author for this post.

  capgros, josef1604, Balmora74, fifi743

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  10/06/2019 at 12:59 AM #109377
  Paul

  In test1, because condsell/condbuy, it creates signals on multiple bars after each other and then it changes the orderprice too.

  to make it stick I used below and other way around.

  if not onmarket or shortonmarket[1] then orderprice=close endif

  1 2 3

  if not onmarket or shortonmarket[1] then orderprice=close endif

  The code works too if limit orders are used. This creates now near similar results to the original code. So far it looks reliable.

  1 user thanked author for this post.

  fifi743

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  10/06/2019 at 11:25 AM #109396
  fifi743

  good-morning Paul ,
  yes condbuy / condsell produces several signal at a time.
  I search for a filter to limit the conditions.

  2 users thanked author for this post.

  josef1604, Paul

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/08/2019 at 10:52 AM #109602
  capgros

  Hi guys,

  Thanks for your amazing work.

   

  I have made some 200k tests in the different versions I found more interesting. From 24/11/2016 to 08/10/2019 (today)

  I don´t know why, but I don´t get the same results as you do on the last 2 versions (fifi & fifi´s Paul edit), could you maybe help me understand why?

  From left to right: v4.2p  / v5.0p / fifi´s / fifi´s Paul edit

   

  

  capgros

  Participant

  Topics: 0

  Replies: 11

  Been thanked: 4 times
  10/08/2019 at 4:03 PM #109630
  Paul

  @capgros

  hi, basically it are different strategies now. From my upload I would focus on 5.0p and on fifi43 latest upload. (you picked the first one)

  test1 was not correctly coded, 4.2p is too old.

  If you want to compare then use the same positionsize /posmax.

  1 user thanked author for this post.

  capgros

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  10/08/2019 at 4:41 PM #109632
  Paul

  a comparison 200k bars limit vs market, 1 position

  Yesterday/today results were different because of 0.8 entry points. Strangly limit had a better entry but was closed quickly.

  Included the correction on previous upload.

  2 users thanked author for this post.

  Balmora74, josef1604

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  10/08/2019 at 8:36 PM #109640
  fifi743

  Hello,
  here is the comparison between my last version and paul’s.
  less DD more gains.
  in my last version I included supports and resistances weekly and daily.
  I still have to include the round numbers
  example 11500
  there can be a rebound

   

  2 users thanked author for this post.

  Paul, josef1604

  

  fifi743

  Participant

  Topics: 10

  Replies: 913

  Been thanked: 426 times
  10/08/2019 at 9:52 PM #109658
  josef1604

  fifi743 wrote:

  Hola,

  aquí está la comparación entre mi última versión y la de Paul.

  menos DD más ganancias.

  en mi última versión incluí soportes y resistencias semanales y diarias.

  Todavía tengo que incluir el

  ejemplo de números redondos 11500

  , puede haber un rebote

  Thank you very much for your work! The v2 you have hung is not the same as the .mod, right? the * .mod version gives me different results to the latest version v2. We wait with intraga the code. Thanks again for your work and Paul’s

  

  josef1604

  Participant

  Topics: 1

  Replies: 50

  Been thanked: 2 times
• Author
  Posts