Fractals, fractal structures, their graphic images + Canvas - page 14

New comment
 
Igor Makanu:

Not to lie, I've been studying your indicator - Channels - for about half a year and, what's interesting, I've never even launched it in MT, but I look at pictures and videos quite often

Last month I did a job - I needed to draw the trend lines on the ZigZag tops with a certain algorithm - in the tester they have pictures just like your channels - the price either bounces from the trend line or not - it's so beautiful.....

i thought about studying your indicators today.... but somehow it seems that my ZZ trends are just an illusion - your Channels are just an illusion.... We have to decide not what we see as "patterns" but what's really going on and what can be used to build TS - nothing in principle: the channel and trend lines are based on 2 points and all that will happen is one more touch of the channel line (trend line) or no touch of it - what I mean, the maximum is the third point, but not the 4th or the 5th point of contact with this channel (trend line) - please note that it often happens and the third point of contact with the line will not be a quite frequent event as well. I.e. statistically there cannot be any regularity here

imho, it is an illusion that there is a forecast of the future price movement in the continuation of a channel or trend

My algorithm is predictive, i.e. the channel has not yet been formed but it has already been detected.

 
Uladzimir Izerski:

My channel construction algorithm is predictive, i.e. the channel is not yet formed, but it is already detected.

same thing :-)

it's not as complicated as it seemed.

 
Maxim Kuznetsov:

same thing :-)

It's bullshit, because it's not as complicated as it seemed.

It's not complicated, but it's important to know the development of future events. It's more important for me to know what's going to happen, not what was.

You can look at the story endlessly, but you won't make any real money from it, only tester money...

 
Andrey Azatskiy:

Well, averages and SPR are the basis of statistics one can say; however trading systems based only on averages - i.e. almost no profitable ones (I mean if one uses textbooks clasics)...
I have not traded with hands for a long time, but the wave patterns can be seen in the market, and if we speak of fractals, this is the first thing that comes to mind - because the basic idea of waves is essentially a figure in a drawing.

Let me repeat myself( I've already said ithere ).

Of course the whole price history (including past and future) of any instrument is a fractal with a relatively small formula.

But to calculate this formula is a mega-complex task that nobody, not even Perelman, can do, but only God).

Clearly structured fractals won't do here.

A coastline is also considered a fractal. But it is a "random" fractal structure. It cannot be predicted by modern mathematical methods. The same is with the price.

 
Igor Makanu:

Not to lie, I've been studying your indicator - Channels - for about half a year and, what's interesting, I've never even launched it in MT, but I look at pictures and videos quite often

Last month I did a job - I needed to draw the trend lines on the ZigZag tops with a certain algorithm - in the tester they have pictures just like your channels - the price either bounces from the trend line or not - it's so beautiful.....

i thought about studying your indicator today.... but somehow it seems that my ZZ trends are just an illusion - your Channels are just an illusion.... We have to decide not what we see as "patterns" but what's really going on and what can be used to build TS - nothing in principle: the channel and trend lines are based on 2 points and all that will happen is one more touch of the channel line (trend line) or no touch of it - what I mean, the maximum is the third point, but not the 4th or the 5th point of contact with this channel (trend line) - please note that it often happens and the third point of contact with the line will not be a quite frequent event as well. I.e. statistically there cannot be any regularity here

imho, it is an illusion that there is a forecast of the future price movement in the continuation of a channel or trend

Yes, the above handicraft has some serious drawbacks due to which its practical application is very problematic. But it clearly shows that there is a finite number of channels in the history and it is even possible to derive the logic of precedence.

Most of the shortcomings have already been solved.

It is wrong to look at one single channel for a forecast. But you can when you look at everything as a whole. Even the probability of a breakdown of a lower channel can be discerned by analysing the "behaviour" of the higher ones.
 

Well, look at Wolf waves (the implementation is in articles by Dmitry Fedoseyev) or Adverse tactics, where "channels" work somewhat differently, but, like, sometimes work, depending on the situation

i.e., repetitive patterns, how much they repeat statistically, i don't think anyone has studied them

In general, the chart is one big fractal, and if you look at it that way, the present will be influenced by absolutely all the history, from the very beginning... at different time levels different influences that intersect. I don't know how reliably it can be described by channels, because if you take stepped ones, you will find that there are also a lot of interesting coincidences, and so on... you need statistics

 
Nikolai Semko: But it clearly shows that there is a finite number of channels in the history and it may even be possible to derive a logic of precedence.
It is not correct to examine just one individual channel for making a forecast. But it is possible to do it when you take the entire history as a whole. Even the probability of breakdown of the low channel may be considered when analyzing the "behaviour" of high channels.

I thought about it, but it all looks nice, but to create a TS we need specific numbers - how many times this and how many times that - statistical observations

If I'm not mistaken, you draw them on canvas - how labor-intensive is the process of determining the intersection of lines of your indicator with a bar? When working with conventional indicator buffers it's a simple comparison of real numbers - the indicator buffer and the price, while when using canvas, what is available?


Maxim Dmitrievsky:

If you look at it that way, the present will be influenced by the entire history from the very beginning... at different time levels there are different influences that intersect.

that's the problem - the problem is that what you said is written on every "fence" in the internet ))) - I have also read many articles on fractals and complex self-similar structures, but they seem to me to be just writing ))))

According to my observations, similarity on graphs appears in splashes of valatility - it is not infinite and has finite values - i.e. in essence the lengths of the ZigZag edges, but if you draw ZZ on a continuous chart, without taking into account trading activity, the ZZ edges will be different, and if you draw I think, that it is necessary to divide a chart into hours with high and low valatility, as well as hairpins in the news = they should be removed from the chart - they don't contain current information - the information was either much earlier or it is just an inside information, which does not reflect price in the previous history

 
Igor Makanu:

I thought about it, but it all looks nice, but to create a TS we need specific numbers - how many times this and how many times that - statistical observations

When working with conventional indicator buffers it's a simple comparison of real numbers - indicator buffer and price, while using canvas, what is available?

I don't quite understand the question. What are the difficulties?

In my iCanvas class there are functions for this

double  X(double bar)          //The X coordinate by the bar number. The bar number must be of type double, otherwise, the bar will be interpreted as time.
double  X(datetime Time)       //The X coordinate by the time.
double  Y(double Price)        //The Y coordinate by the price.
double  Price(int y)           // Price by the Y coordinate
double  Bar(int x)             // bar number by coordinate X 
datetime  TimePos(int x)   
double    Close(int x)     
double    Open(int x)     
double    High(int x)      
double    Low(int x)    
 
Nikolai Semko:

I don't quite understand the question. What could be the difficulty?

In my iCanvas class, there are functions for this

OK, thanks! I.e. your indicator can be taken as a basis for a channel indicator - there is access to data, it is not just a "drawer"

I just haven't looked at your code, but I watch your work regularly

 
Igor Makanu:

OK, thanks! I.e. your indicator can be taken as a basis for a channel indicator - there is access to data, it's not just a "drawing machine"

I just did not look at your code, but I regularly watch your creative work.

Thank you.
If you mean MQL4, it works. It's just very slow and not on kanvas. I managed to speed up the same algorithm about 1000 times faster, but it's still my intellectual property.

I just noticed that it has been downloaded 25228 times in the Russian part of the forum and 53422 times in the English version. Pleasant as hell, although then (in 2012) I just started experimenting with pattern recognition.

1...7891011121314151617181920
New comment
Reason: