Hidden divergence - page 45

New comment
 

These are not confirmatory signals but signals from strongly correlated indices. Feel the difference.

 

2/3

ЕСЛИ я правильно (то, что не "красиво" - факт) перенёс "мозги" FX5 в инклудник ТО
ЕСЛИ я корректно подцепил "предыдущие эксремумы ZigZag'а" к текущему бару ТО
{
  получаем файл DivStat_EURUSD_240.csv, в котором
   WhatSub    - некое название "типа фигни" из FX5, 0 - отсутствие "фигни"
   TimeCurBar - время выгружаемого бара
   TimeCurrEx - время окончания формирования "фигни" (обращаю внимание на разницу между TimeCurBar и TimeCurrEx
   TimeLastEx - время начала формирования "фигни"
   CurrCurrency - значение цены в точке TimeCurrEx
   LastCurrency - значение цены в точке TimeLastEx
   divCurrency - "оцифрованная половина фигни"
   CurrInd - значение индикатора в точке TimeCurrEx
   LastInd - значение индикатора в точке TimeLastEx
   divInd - "оцифрованная вторая половина фигни"
   TimeLastZZ - время предыдущего экстремума ZigZag'a
   ZZlast1 - первый буфер ZigZag'a (констатация)
   ZZlast2 - второй буфер ZigZag'a (если не ноль - минимум)
   ZZlast3 - третий буфер ZigZag'a (если не ноль - максимум)
}

-
about unloaded ZigZags.
At first I hoped to select the previous value in Access(surely everyone has it, unlike MS SQL), but then I decided that it's easier to unload
- about writing to a file - I decided not to make a fuss (quickly) and for the case without "bullshit" I did something like this:

...
   else
    FileWrite(handle, "0", TimeToStr(Time[iScript], TIME_DATE|TIME_MINUTES), 
              TimeToStr(Time[iScript], TIME_DATE|TIME_MINUTES), TimeToStr(Time[iScript], TIME_DATE|TIME_MINUTES), 
              0, 0,
              0,
              0, 0,
              0,
              TimeToStr(TimeZZ, TIME_DATE|TIME_MINUTES), ZZlast1, ZZlast2, ZZlast3);
...

	          



		          

			          

    Files:
  

      

          divstat.rar
 37 kb      

  




		          


	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
SergNF писал (а) >>
 
Proof variant. 
 
- At time X the KFOR is formed
 
- Looking at the previous value of the ZigZag. If the Minimum, we are in the UP Trend. If the Maximum, we are in the Down Trend.
 
- If after KFOR is formed through n bars, the next "ZigZag point" appears, we consider KFOR  has completed its work.
 
- We calculate the "only right" number of correct events and compare it with 50%.
 

 
- ZigZag is not the best ... filter called "TREND", 
 
 
BUT. It's about time we started!
 
 
Maybe at least you'll get this thread back on track.
 
It is not a question of what (trend and KFOR tools) to use (you can use ordinary МА and ordinary oscillators. As an example see the picture below), but how to determine whether it has worked or not.
 The idea - if after the formation of KFOR through n-bars the next "ZigZag point" appears, then we can say that KFOR trigger  ed seems quite reasonable to me. Who is implementing it?
 

 

	          




		          
	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
I formulated the rules for manual trading  Strategija torgovli. I will give the password to anyone who is interested in these rules.
 
For those who promised me something I will give them for what they promised.

	          




	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
3 not of 3
 
So I will try to "digitise" Geronimo's definitions (from page eight) in "my" terms:
 
- "Hidden" Div-Con on a bullish trend :
 
divCurrency = Price[last peak] / Price[previous peak] < 1
 
divInd = Indicator[last spike] / Indicator[previous spike] < 1
 
Bullish trend - 3rd ZigZag buffer at previous "extremum" > 0
 
- "Hidden" Div-Con on a bearish trend 
  
divCurrency = Price[last peak] / Price[previous peak] > 1
divInd = Indicator[last peak] / Indicator[previous peak] > 1
Bullish trend - 2nd ZigZag buffer at previous 'extremum' > 0
'
Preliminary ZY
.
I still don't understand the phrase "the troughs in the indicator are deeper than the troughs in the price chart
", i.e.
 divInd > divCurrency?
'
Next
.
Let's link the file DivStat_EURUSD_240.csv to Access and write the query:
SELECT WhatSub, TimeCurBar, TimeCurrEx, TimeLastEx, CurrCurrency, LastCurrency, 
divCurrency, CurrInd, LastInd, divInd, TimeZZ, ZZlast1, ZZlast2, ZZlast3, 
IIf([WhatSub]="0","",
   IIf([divInd]<0,"Нетрадиционная ДК",
      IIf([divInd]<1 And [divCurrency]<1 And [ZZlast3]>0,"СДК бычья",
         IIf([divInd]>1 And [divCurrency]>1 And [ZZlast2]>0,"СДК медвежья",
         "Простая ДК")
      )
   )
) AS ПризнакДивергенции
FROM DivStat_EURUSD_240;
The concept of "Non-traditional DK
" is a version 2.1 of the indicator that draws "bullshit bars" on the indicator chart "through 0".
s2101 wrote about 'that too', but not for the Geronimo
 variant
'
How and where to do and post-processing hasn't worked out yet.
'
ZS. I changed the query

	          




		          
	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
4 out of 3
 
Solved it in Excel (I couldn't figure out how to write it in Access' "dialect" of SQL).
 
I got the following formula
 
=ЕСЛИ(O2<>"";
  ЕСЛИ(СУММ(СМЕЩ($L2;0;0;Q$1;1))/L2=Q$1;
   "Продолжилось";
   "Изменилось");
  "")
 
And for value 3 (within 3 next H4 bars of EURUSD pair) from 21.06.2004 16:00:00 till 07.04.2008:
 
- "CDC is bullish".
 
21 in total
 
"Continued" 19 units
 
"Changed" 2 units
 
- "KFOR bear".
 
Total 30 pieces
 
"Continued" 27 pieces
 
' 'Changed' 3 pieces
 
'
 
:)
 
I will begin, naturally, "with myself" (look for errors - small amount of "bullshit" + probably "mistake sign", although Geronimo in one place "trend continuation", in another - "reversal" ), but . we're getting tired. ;)
 
SZZ. If I will not be lazy, I will look for the way of "unloading" the "waves" from 2004.
 
SZU. It is true that in absolute terms, the concepts "Continued" and "Changed" have not been considered.

	          




		          
	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
Last one not out of 3 :(
 
The most recent "KFOR" + "Changed" is line 4568:
 
- KFOR is bullish.
 
- Bar formed on 25.05.2007 12:00:00
 
- Right time of "bull bar" formation on 25.05.2007 4:00:00
 
- Time of previous "extremum" ZigZag 25.05.2007 0:00:00 (i.e. 3 bars of "operating time" after the "extremum", on the next bar after the "extremum" the formation of "bullshit" has finished)
 
- The time of the next ZigZag "extremum", 25.05.2007 16:00:00 (i.e., on the next bar of the "operating time", after the "bull" was drawn "retroactively", the ZigZag has drawn its maximum)
 
 
Looking at the "chart":
 
 
I.e. ' SOMEONE 's got it right.
 
'
 
ZS. Speaking of indicator parameters (NOT recommended):
 
//FX5_Divergence
extern int    fastEMAext=12;
extern int    slowEMAext=26;
extern int    signalext=9;
//ЗигЗаг
extern int ExtDepth=12;
extern int ExtDeviation=5;
extern int ExtBackstep=3;
 
'
 
Added!!!
 
I.e... SOMEONE is quite correct... in the 'algorithmic' sense, i.e. with the assumption that one of the factors/indicators described by TC is substituted for ZigZag!!!!! That it (ZigZag) is nonsense is clear to me, but I wanted a "quick" price tag.

	          




		          
	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
SergNF писал (а) >>
 
For value 3 (within the next 3 H4 bars of the EURUSD pair) from 21.06.2004 16:00:00 to 07.04.2008:
 
- "CDC is bullish".
 
21 in total
 
"Continued" 19 units
 
"Changed" 2 units
 
- "KFOR bear".
 
Total 30 pieces
 
"Continued" 27 pieces
 
"Changed" 3 pieces
 
 
For value 1 (within 1 next D1 bar of EURUSD) from 21.06.2004 16:00:00 till 07.04.2008:
 
- "CDC is bullish".
 
Total 2 units
 
"Continued" 2 units
 
"Changed" 0 units
 
- "KFOR bear".
 
None
 
'
 
For value 5 (within 5 consecutive M30 bars of EURUSD) from 21.06.2004 16:00:00 to 07.04.2008:
 
- "CDC is bullish"
 
Total 169 units
 
"Continued" 140 units
 
"Changed" 29 units
 
- "KFOR Bear"
 
None.
 
'
 
For value 10 (within 10 and next M15 bars of EURUSD) from 21.06.2004 16:00:00 till 09.02.2007 (did not fit into Excel...)
 
- "SDK bullish".
 
A total of 162 units
 
"Continued" 98 units
 
"Changed" 64 units
 
- "KFOR bear."
 
Total 233 pieces
 
"Continued" 125 pieces
 
' 'Changed' 108 pieces
 
'
 
For value 30 (within 30 and next M5 bars of EURUSD pair) from 21.06.2004 16:00:00 till 05.05.2005. (it did not fit into excel):
 
- "SDK bullish".
 
Total 153 units
 
"Continued" 22 units
 
"Changed" 131 units.
 
- "KFOR bear."
 
Total 200 pieces
 
"Continued" 26 pieces
 
' 'Changed 174 pieces
 
'
 
ZS. I have no H1 EURUSD history at all on 'this' computer.
 
The fact that the parameters, ZigZag in particular, need to be changed is understandable.
 
'
 
Now that's it.

	          




		          
	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
SergNF писал (а) >>
 
For value 5 (within the next 5 M30 bars of the EURUSD pair) from 21.06.2004 16:00:00 to 07.04.2008:
 
- "SDK bullish".
 
Total 169 units
 
"Continued" 140 units
 
"Changed" 29 pcs.
 
 
Do I understand correctly that this is the most interesting part?
 
Could you please, Sergey, share the FX_5 Divergence_v2.1 indicator you have on your chart?

	          




	          

		          

	          

          

        

        


            

          

              
	          

		          
		          
                
		          



                 
	          


	          

		          
Xadviser писал (а) >>
 
Am I right in assuming this is the most interesting part?
 
 
Not really. "An even number of errors can sometimes give the right result"
 
The suspicion is the low number of "combinations" in 4 years.

	          



		          

			          

		          


	          

		          

	          

          

        




  


  

    

      

        
      

        
1...383940414243444546474849505152...66      
        
    

    

          
            New comment