a trading strategy based on Elliott Wave Theory - page 20

 
Prosto etomu indikatoru nado na mql4 i pribavit' filtr "shuma" iskazeniji amplitudi cen + kod pravki oshybo4novo opredilenija na4ala ods4iota...:)
 
Correct.....

Got it... :)


Agree - it's much more useful than just copying an algorithm. There will still be complications, but they are all technical in nature.

Good luck and good trends.
 
Since you don't need the parabola itself, you can approximate the derivatives straight away...

sounds rather strange...
A contradiction arises - how can one pay attention to the nature of the derivative, while stating that the form of the original function is irrelevant (and this idea has been voiced more than once).
After all, the derivative and the function itself are uniquely related (to within linear coefficients).
 
how you can pay attention to the nature of the derivative, while claiming that the form of the original function is irrelevant (and this idea has been voiced many times).

As far as I understand this idea, it is as follows.
When you approximate a price graph that resembles a parabola, for example, by a linear regression channel, you should get an error function, which will be a parabola with a peak located in the centre of sampling (and this already significantly reduces the amount of calculations!), i.e. it turns out that we do not care how and where we initially draw this initial parabola (of course within reasonable limits ;o)). And already for this resulting error parabola it is required to apply different methods of finding the confidence interval. And through this parabola, conclusions should be drawn about the channel, along which the price moves according to the sample. Knowing at least that the vertex of the parabola is in the middle of the sample, it becomes easier to find the corresponding approximation. And if those errors obtained for approximation of the error parabola from the first iteration, after the second iteration will be approximately equal in time distribution (the second derivative of the quadratic function is a constant), that is if they are similar to the normal distribution, then we consider that the approximation function of the initial price series is really 2, that is the function must be quadratic (parabola, or for example part of an ellipse - though in terms of logic ellipse is absurd!) And most importantly
The second part of the problem is to determine the confidence interval and the direction of the prediction itself. That is, we know in the second iteration the range (confidence interval) where the errors are after the second iteration (and it should be similar roughly to a flat channel, on the fingers). Then we place a sell order near the upper limit and a buy order near the lower one. Or, in the second case, we just calculate the probability of a trend reversal. And then we average these calculations over all found channels and define more precisely an average probability of a trend reversal and possible boundaries which price should pass to make all our constructions erroneous, i.e. we find where we should place a stop and probably with a reversal - Murray levels can help us with this a little. Also, on the other hand, if we look at the error chart of the second approximation, first of all, we know the exact confidence interval and secondly, we can say exactly where in the confidence interval we are at the current moment in time. That is, we can estimate how many pips from the current state the price can go before it reaches the boundary of the confidence interval. And accordingly, based on this we can think about the specific price for placing a Limit order. It should seem to correspond precisely with the real limit the price will reach.
Although, I could be wrong in my assumptions. Vladislav could correct me, because it is his idea and so far we can only speculate.
 
Correct. You see - I told you it's more useful to understand than to just copy.

Good luck and follow the trends.
 
Поскольку сама парабола Вам не нужна, то сразу можно аппроксимировать производные...

sounds rather strange... A contradiction arises - how can one pay attention to the nature of the derivative, while stating that the form of the original function does not matter (and this idea has been voiced more than once). The derivative and the function itself are uniquely related (to within linear coefficients).



This is only exactly true for analytic functions - that is, those that can be written down explicitly.
If you build an approximation, you get the function itself in error, and consequently its derivatives. Therefore, as long as you are in the same interval all "different" functions whose difference will not exceed the size of the confidence interval can be assumed to be the same.
The potentiality of the price field, on the other hand, provides an opportunity and method for reconstructing the function from the derivative. This is an inverse problem and it does not always have a solution, in contrast to the direct problem that always has a solution. Simply put, there is always a derivative, but there is not always an integral - this is true even of analytic functions. If it weren't for that, you wouldn't have to bother with the adequacy of approximations.

Good luck and hitchhiking trends.
 
Prosto etomu indikatoru nado na mql4 i pribavit' filtr "shuma" iskazeniji amplitudi cen + kod pravki oshybo4novo opredilenija na4ala ods4iota ...:)


In principle, now I have time - what I was doing already running in a trial - I'll look at the errors in the implementation of the machine ( because I'm sick and tired of trading by hand :) ). As long as I have time - so if you put the code in full (or send it to 4vg@mail.ru), I can help with the translation.

Good luck and good trends.
 
<br/ translate="no"> While there is time - so if you post the code in full (or email 4vg@mail.ru) I can help with translation.

Good luck and happy trends.


V principe sdes' iest' jest' expert s indikator v odnom, only ko kod staryj MT3 (ja neperevodil ~2500 strok v MT4, napizal drugoj) ;-]

Etot kod ja sdies' kinul 4toby kak vy imeli xot' 4toto dlia na4ala ... Takije liudi kak vy vsegda nuzny dlia etovo, polza budet vsem ;-]

Teper' o kode:

Algoritm prastoj: is4em na4ala ods4iota, markirujem kak na4alo Elliiot Wave 1, is4em osnovnyje 1-2-3 volny. Vinadleznosti ot trenda - UP ili DOWN doajem tak (napishu tol'ko pri UP, pri down invertirovanije):

1) Jesli UP - is4em za period 350 cen minimal'noj price dlia EW1 na4ala (sdies' nuzen kod flata), patom maximal'noj ceny za period say, 350 cen (5/35 x 10 , takije cifry v EW indikatorax) - eto u nas v na4ale EW2 na4alo.
2) smotrim ni uxodit cena slishkom iz predelov shuma, jesli da - is4em minimal'noj ceny posle EW2 na4ala i markirujem kak EW3 na4alo. Cena ni dlozna padat' >3/4 EW1 dliny(sviazka s Fibonacci ~61.8%) i nidolzna byt menshe 1/4 EW1 dliny (Fibo ~32.8%)
3) EW3 dolzna dostignut' vyshe EW1 (jesli nedastigli "current" prices), jesli perevarot i padajet nize konca EW1, zna4it, ploxoj ods4iot, smotrim v periode interval pabolshe vakuju vsio taki storonu trend, i delajem vsio zanovo (nizabudte o flate weekly i monthly, ili zaciklitsia!).
4) jesli imejem EW3 na4alo, i eta cena v istoriji idut za granicy EW1 v dline EW1 x 2(ods4iot s EW2 konca) , zna4it imejem EW3.
Takoje dlia 1-2-3, dalshe 1-2-3-4-5:
5) smotrim kokda budet perevarot s vyxodom iz ramok shuma, lovim na4alo EW4 posle perevarot.
6) Jesli imejem na4alo EW4, cena dolzna padat' ni menshe 23.6% dliny EW3 , no i nidolzna peresekaca s koncom EW1(!)
7) Jesli imejem EW4 na4alo, i cena uidut v verx vyshe shuma ramok, i EW4 vpisysyvajetsia v 6) , na4inajem ods4iot EW5
8) jesli imejem EW5 na4alo, volna dolzna byt' dlinoj ne menshe dliny EW1 (!) i zakon4itsia vyshe na4ala EW4.

Vot takoj principe moj algoritm, ideji i MQL4 kod sdies' bylo by kruto! ;-]

P.S. dlia opredilenija shuma ja v tom indikatore ispolzoval kvadratnoje standartnoje otklonenije v intervale EW1 na4ala-current, jesli imejete variant po lu4e, daite kod MQL4 ;-]
 

Пока есть время - так, что если выложите код полностью (или шлите на мыло 4vg@mail.ru), то могу помочь с переводом.

Удачи и попутных трендов.


V principe sdes' iest' expert s indikator v odnom, only ko kod staryj MT3 (ja neperevodil ~2500 strok v MT4, napizsal drugoj) ;-]

Etot kod ja sdies' kinul 4toby liudi kak vy imeli xot' 4toto dlia na4ala... Takije liudi kak vy vsegda nuzny dlia etovo, polza budet vsem ;-]

Teper' o kode:

Algoritm prastoj: is4em na4ala ods4iota, markirujem kak na4alo Elliiot Wave 1, is4em osnovnyje 1-2-3 volny. Vinadleznosti ot trenda - UP ili DOWN doajem tak (napishu tol'ko pri UP, pri down invertirovanije):

1) Jesli UP - is4em za period 350 cen minimal'noj price dlia EW1 na4ala (sdies' nuzen kod flata), patom maximal'noj ceny za period say, 350 cen (5/35 x 10 , takije cifry v EW indikatorax) - eto u nas v na4ale EW2 na4alo.
2) smotrim ni uxodit cena slishkom iz predelov shuma, jesli da - is4em minimal'noj ceny posle EW2 na4ala i markirujem kak EW3 na4alo. Cena ni dlozna padat' >3/4 EW1 dliny (sviazka s Fibonacci ~61.8%) i nidolzna byt menshe 1/4 EW1 dliny (Fibo ~32.8%)
3) EW3 dolzna dostignut' vyshe EW1 (jesli nedastigli "current" prices), jesli perevarot i padajet nize konca EW1, zna4it, ploxoj ods4iot, smotrim v periode interval pabolshe vakuju vsio taki storonu trend, i delajem vsio zanovo (nizabudte o flate weekly i monthly, ili zaciklitsia!).
4) jesli imejem EW3 na4alo, i eta cena v istoriji idut za granicy EW1 v dline EW1 x 2(ods4iot s EW2 konca) , zna4it imejem EW3.
Takoje dlia 1-2-3, dalshe 1-2-3-4-5:
5) smotrim kokda budet perevarot s vyxodom iz ramok shuma, lovim na4alo EW4 posle perevarot.
6) Jesli imejem na4alo EW4, cena dolzna padat' ni menshe 23.6% dliny EW3 , no i nidolzna peresekaca s koncom EW1(!)
7) Jesli imejem EW4 na4alo, i cena uidut v verx vyshe ramok shuma, i EW4 vpisysyvajetsia v 6) , na4inajem ods4iot EW5
8) jesli imejem EW5 na4alo, volna dolzna byt' dlinj ne menshe dliny EW1(!) i zakon4itsia vyshe na4ala EW4.
9) kazdaja "failed" volna - ploxoj ods4iot, smotrim pri takom slu4aje interval pabolshe i delajem vsio zanovo.

Vot takoj v principe moj algoritm, ideji i MQL4 kod sdies' bylo by kruto! ;-]

P.S. dlia opredilenija shuma ja vom indikatore ispolzoval kvadratnoje standartnoje otklonenije v intervale EW1 na4ala-current, jeseli imejete variant po lu4e, daite kod MQL4 ;-]
 
It seems to me that you just didn't understand my sentence. The point is that the code posted here is not complete. I think some of the code was "cut off" due to restrictions on the size of programs that can be posted on the forum. I don't feel like restoring your system, because I'm trading on my own. But I can help you with translation, but to do so the program must at least compile - you have to take something as a reference. That's why I wrote, if there is old code, but it compiles, then I can help with the translation - and then you can think about what to do with it. If you don't need a translation, then there's no need to bother with it. If anyone needs it - I can do it (again, for this purpose the program must be compiled). I have my own trading system.

Good luck and passing trends.