The Sultonov system indicator - page 70

 
Nikolai Semko:

I tried to find out too, but Yusuf never gave me any answers

How and why twist and turn, see post above. Why Y appears and many things. You don't get it - over there.

 
Yousufkhodja Sultonov:

How and why twist and turn, see post above. Why Y and many other things appear. It's not clear to you - there.

Where above? Above 70 pages.

the links to the posts can be copied like this


 


Nikolai Semko:

where above? Above 70 pages.

https://www.mql5.com/ru/forum/307935/page69#comment_11199181

Or, forcedly:



Системный индикатор Султонова
Системный индикатор Султонова
  • 2019.04.02
  • www.mql5.com
Уважаемые форумчане, в качестве основы стратегии будущего индикатора рассмотрим и обсудим следующую гипотезу: Цена текущего бара зависит от 4-х зна...
 

I thought you had

Y = Ydiscount

 
Roman Shiredchenko:

Again, not to be taken as rubbish, glad to read you.

Hi Roma!
I'm happy to read the "old timers" of the forum too)
 
Mikhail Dovbakh:
Hi Roma!
I'm glad the "old timers" of the forum are here too)

Glad, glad, glad, glad, for the "oldies"... :-)

Eh, "old-timer..." :-)

 
You should understand that the problem you are describing here is different from the one under discussion.
I would use, as Nikolai suggests, the time index of the series point.
After all, this thread discusses series x from the same source, only series x2 in your presentation is the same as series x1, with zero value removed but k-th value added. Or it refers to x4, I never understood(
In this case the system will simplify, both analytically and for the readers' understanding)
As for Y, this is an estimated (predicted) value with calculated coefficients and actual values of x - although you keep saying that there are no predictions here)
Add indexes and you yourself will find it easier...
 
Nikolai Semko:

I thought you had

Y = Ydiscount

That's how it should be, that's how it is in the exel code and the MKL indicator code. What is the problem?

 
Yousufkhodja Sultonov:

This is the way it should be, this is the way it is in the exel code and the ICL indicator code. What is the problem?

Forum on trading, automated trading systems and strategy testing

Sultonov's system indicator

Yousufkhodja Sultonov, 2019.04.02 22:26

It follows from the principle of Gauss ISC:

If Үdiscount = a0 + a1x1 + a2x2 + a3x3 + a4x4,

We must minimize the difference of the sum of squares U =Σ(Y - Үdiscount)^2 =Σ(Y -a0 + a1x1 + a2x2 + a3x3 + a4x4)^2 = minimum

Universal minimum condition :

1. dU/da0 =0;

2. dU/da1 =0;

3. dU/da2 =0;

4. dU/da3 =0;

5. dU/da4 =0.

After differentiation we obtain the system given inhttps://www.mql5.com/ru/forum/86249/page3 , since,Σa0 = na0. There is no limit to the number of input data, even if there are a million data. Simply, there will be a million minus 5 systems, each to be solved for each selected period.

That's why the sum signΣ, the parameters X and Y and their sums will appear everywhere, and also the fact, that the quantity of the equations must be no less than 5, by the principle the more, the better, but no less than 5! I have settled on this number.


The problem is this (highlighted in yellow).

IfY = Ydimensional, thenΣ(Y - Ydimensional)^2 is always 0

 
Mikhail Dovbakh:
You must realise that the task you are describing here is different from the one you are discussing.
I would use, as Nikolai suggests, the time index of the series point.
After all, this thread discusses series x from the same source, only series x2 in your presentation is the same as series x1, with zero value removed but k-th value added. Or it refers to x4, I never understood(
In this case the system will simplify, both analytically and for the readers' understanding)
As for Y, this is an estimated (predicted) value with calculated coefficients and actual values of x - although you keep saying that there are no predictions here)
Add indexes and you yourself will find it easier...

Otherwise, as I've done, it's impossible to handle a single series of price data. There is no way to simplify the system. Y is from that series of already existing data, differs from it only in the fact that, it's the price of a new bar that just opened. Where is the history or the forecast? Where to add indices? Where do you put the prices themselves? You get heterogeneous data - you mix the warm with the heavy. As a result, you will not obtain results about neither warm nor heavy properties of the object, except to make Gauss roll over in his grave with amazement. It is another matter to take different 4 instruments. Then, each of the tools will enter into the system from its row, but, it already from other line of processing and has no relation to the present case.