Discussing the strategy VECTORIAL DAX (M5)

• 10/29/2019 at 11:09 AM #111515
  GraHal

  I don’t even recognise that Import screen you are using.

  Also are you sure you are not trying to import an Auto-System as an Indicator/

  Have you ever imported an Auto-System successfully? Does the way you do it ever work??

  Are you doing it as attached via the Indicators and Trading Systems Window having clicked on the right hand side – Backtesting and Trading Systems??

  

  GraHal

  Participant

  Topics: 204

  Replies: 8401

  Been thanked: 2042 times
  10/29/2019 at 11:48 AM #111524
  Gubben

  The latest version has not executed any orders this week – is that correct?

  

  Gubben

  Participant

  Topics: 2

  Replies: 58

  Been thanked: 8 times
  10/29/2019 at 11:52 AM #111525
  GraHal

  Yes correct for V6-1  @Gubben

  I’ve not got V6-2 running yet.

   

  

  GraHal

  Participant

  Topics: 204

  Replies: 8401

  Been thanked: 2042 times
  10/29/2019 at 11:53 AM #111526
  Vonasi

  Just trying to get my head around the code and I noticed one odd thing. The section of code that sets the value of CD:

  for i=0 to 6 do if coeff[i]=100 then CD=1 break else CD=0 endif next

  1 2 3 4 5 6 7 8

  for i=0 to 6 do if coeff[i]=100 then CD=1 break else CD=0 endif next

  is quite late in the code and yet the condition CD is used in the first set of buy conditions. Surely this CD calculating code would be better placed before that possible first buy order otherwise you have to wait for one bar before those conditions can be met. Not a big thing on a five minute chart and maybe it is intentional?

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/29/2019 at 11:59 AM #111527
  winnie37

  @ Grahal

  yes, this way, as usual 🙂

  When i create a new code, it works, just the import does not work…. I ran the apple catalina update; Maybe a reason?

  Before finding a solution, could you be kind enough to post the entire code of the lastest fifi version (code and indicator)? Many thanks 🙂

  

  winnie37

  Participant

  Topics: 27

  Replies: 165

  Been thanked: 16 times
  10/29/2019 at 12:13 PM #111528
  winnie37

  Solved! I called Prt services; It was due to the Catalina update and security parameters….

  1 user thanked author for this post.

  GraHal

  

  winnie37

  Participant

  Topics: 27

  Replies: 165

  Been thanked: 16 times
  10/29/2019 at 12:41 PM #111530
  Vonasi

  Having now sat down and looked through the Vectorial Dax (fifi743’s version) code what concerns me about this strategy is that there are hundreds of very different sets of exit conditions – I think I counted 96 for closing trades and then there are all the position reversal sets of conditions which I have not counted yet but probably about the same number. That makes me think that a set of exit conditions has been created specifically for every possible entry in the history tested – otherwise known as curve fitting! This would explain why the robustness test equity charts all looked very similar. If every trade can find its own pre-selected exit or reversal condition amongst the hundreds on offer then it will close or reverse as perfectly every time a test is run.

  I might be wrong but I think an awful lot of work has been put into creating the ultimate curve fit where every trade has its own almost perfect exit condition from amongst the hundreds available. Forward testing will let us know!

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/29/2019 at 12:56 PM #111531
  AE

  Hi Vonasi,

  I understand your concern, in deed at first I thought like you. But, in 2 years the strategy did more than 760 operations and October is out of sample and its numbers are amazing. Lucky? Maybe but I consider very difficult to adapt more dan 90 rules for 760 operations during two years where 2018 was a year for shorts and 2019 a year for longs.

  I’m pretty sure that in future the results will be more weak than backtest but right now, October is been amazing and we need more time to know if this strategy is very solid or not.

   

  

  AE

  Participant

  Topics: 11

  Replies: 53

  Been thanked: 12 times
  10/29/2019 at 12:58 PM #111532
  AE

  Sorry, I mean in 3 years, not 2.

  

  AE

  Participant

  Topics: 11

  Replies: 53

  Been thanked: 12 times
  10/29/2019 at 1:16 PM #111534
  josef1604

  Anyone know why yesterday’s entry was not executed on the 28th as it appears in the bactesting?

  

  josef1604

  Participant

  Topics: 1

  Replies: 50

  Been thanked: 2 times
  10/29/2019 at 1:40 PM #111538
  josef1604

  This is the entry that was not executed yesterday in demo. The one that is marked in red. Does anyone know why he did not run?

  

  josef1604

  Participant

  Topics: 1

  Replies: 50

  Been thanked: 2 times
  10/29/2019 at 1:54 PM #111541
  fifi743

  Hello
  what is missing for it to work properly:
  holidays
  confirmation of orders to ensure that orders are well passed.
  I do not see how to do in one line stop, well it’s not my job.
  trading is like the weather, it’s never the same conditions

  1 user thanked author for this post.

  Wacko

  

  fifi743

  Participant

  Topics: 9

  Replies: 851

  Been thanked: 401 times
  10/29/2019 at 1:59 PM #111542
  fifi743

  I am going to modify the line said by vonasi.
  we lack history.
  but the problem is that in the story it is missing candles

  

  fifi743

  Participant

  Topics: 9

  Replies: 851

  Been thanked: 401 times
  10/29/2019 at 2:04 PM #111544
  GraHal

  josef1604 wrote:

  Does anyone know why he did not run?

  You could check using Cursor Details on the 5 Min Chart if  Price was close, but not close enough to be triggered allowing for spread being different in real than you / we have set in Backtesting?

  Let us know what you find out please?

   

  

  GraHal

  Participant

  Topics: 204

  Replies: 8401

  Been thanked: 2042 times
  10/29/2019 at 2:28 PM #111546
  GraHal

  GraHal wrote:

  Let us know what you find out please?

  From what I can see / determine … the Long Trade in backtest triggered at 12917.3, but in Real Live the Price for a 10 second bar at 08:40 (UTC + 0:00) did not get above 12917 as a high (see attached).

  Why a Trade did not trigger on a subsequent candle I do not know. The Long trade at 08:40 in backtest was a reversal from a Short so maybe the Trade had to be  a reversal or nothing??

  (I did above without thinking / very quick (out of curiosity) so it would stand a closer look (by anybody) if we think it is important enough??  Any comments on the method welcome?  We may all learn something?? 🙂 )

  

  GraHal

  Participant

  Topics: 204

  Replies: 8401

  Been thanked: 2042 times
• Author
  Posts