A purely theoretical question for mathematicians. With the possibility of moving to the practical plane. - page 3

[Igor Makanu]

Сергей Таболин:

So I'll repeat my initial question:how do I calculate patterns in a series of values? For example +165, -240, +18, -378, +681, -115.... ?

Are you looking for true vertices of SZ or do you want to find an algorithm for generating sequences?

if a sequence of numbers are linear transformations, the NS will figure them out in no time, here is an example

https://www.mql5.com/ru/forum/5010#comment_329221

[Maxim Kuznetsov]

to answer, think about the very principle of building all sorts of "zigzags":

- (the current direction is downward)

- if the price is lower than the last extremum, the extremum is transferred to it

- if the price goes upwards and some factor (deviation/distance, etc.) is higher than the threshold - a new knee is added and the direction is changed.

Even from the general description it is clear that the zigzag itself as an indicator only measures a local "factor/principle of change of direction".

In some cases it is even possible to draw useful conclusions. For example "if the line has a slope, a turn-almost-now is probable".

But trying to derive a "formula" from a zigzag, it's simply not even practical - you have to take the factor mentioned and study it.

[Aleksey Nikolayev]

Сергей Таболин:

Unfortunately, back in 8th grade I gave up on the gold medal and completely abandoned my studies. Thank you for your explanations, they are certainly meaningful to knowledgeable people, but for me, they say nothing.

Therefore I repeat my initial question:how to calculate regularities in a number of values? For example +165, -240, +18, -378, +681, -115.... ?

Can anyone, without being overly snooty, suggest a formula for solving this problem?

Seek and find, push and shove.

[Vizard_]

https://radikal.ru/video/RSRGjZ9k94E

[[Deleted]]

Igor Makanu:

You should decide whether you are looking for "true vertices" or you want to find an algorithm of sequence generation.

If a sequence of numbers is a linear transformation, the NS will figure it out in no time, here's an example

https://www.mql5.com/ru/forum/5010#comment_329221

Let's put it a little differently.

This is not a zig-zag. I just visually connected the tops of "trends".

No knees at all.

I would like to calculate the interdependence of price movement on the marked segments.

Maxim Kuznetsov:

.................

But trying to derive a "formula" from the zig-zag, it's just not even practical - you have to take the factor mentioned and study it.

The zig-zag was cited purely for clarity. Sorry again, I'm not a mathematician, I'm not fluent in special terms.

Aleksey Nikolayev:

Seek and find, push and shove.

Thank you. I understand you. I expect a return of understanding - no more comments like that. OK?

[Igor Makanu]

Сергей Таболин:

Let's do it a little differently.

I would like to calculate the interdependence of price movements in the marked segments.

let's do it

You write the price movement - where on the history in the terminal is the movement?

there are only bars! Such verbal exercises create a desire to make the wish come true))

formulate your task as follows: here are the bars, look for ???? - Remember the old truth: the right question already contains half the answer

[[Deleted]]

Nikolai Semko:

Sergey, OK, I'll answer without any rhetoric.

There are an infinite number of formulas for generating the sequence you quoted.

...........

In order to create an exact formula of a regularity in which its future members are also taken into account, one cannot do without a time machine.

Your question can be rephrased as follows:

Can anyone, without too much conceit, suggest a working grail formula?

So the problem is in the last 4 points (....) of the sequence you give.

1. I don't need a formula to generate, but to find the relationship of an already existing series.
2. No, well, who told you about accounting for future members? You're like that anecdote: "You're my sweetheart.... .... ..... People! He called me a bitch!"
3. That word(grail) has almost become a dirty word around here. So, yes, I'm looking for it. And you're not? If not, what are you even doing here?

[Nikolay Bosuy]

I did something similar at one time... Only I was looking for "possible pivot points", the key word being possible. I was looking for the average of N rays in the history downwards and separately for N rays in the history upwards. I calculated the average and got the approximate value of the current ray down or up. All this works well in the flat or calm market, but in the long trend does not work, as well as most strategies to find the market reversal point

[Oleg Mamchenko]

I don't think algebra will help you here, just use a correlation formula.Pearson or Spearman correlation coefficient. They will tell you whether there is a correlation or not. To be able to calculate a series you need a hypothesis formula and then check for convergence or non convergence.

[[Deleted]]

In my view, finding a pattern is not the only challenge. But in finding it, there is the possibility of finding variation in that pattern. Or to look for the interrelationships of several.