Fractals, fractal structures, their graphic images + Canvas - page 18

New comment
 
Roman Kutemov:
We already see this on history.
But how would one learn at least at the initial stage to detect them (at the stage of formation of the second fractal)?
Ps: I would turn the second fractal clockwise by 180 degrees.

The difficulty is that the market movement can be presented as a long one and then it may turn out to be a correction, i.e. the prevalence of one of the elements in the dominance order over the distribution space in its extreme values, the long Fr should "redraw" the short Fr.

 
Maxim Dmitrievsky:

usually looking for the end of such a structure, in anticipation of a U-turn. And the target is always on the centre (green dot). Doesn't always work, yes, I have to play with stops, re-set. I don't know how to automate it either.

Blue dot... yep, color-blind.

I've developed a super-fast method to calculate polynomials of almost any degree (in reality up to about 15, because beyond that I start having problems with lack of accuracy of type double) without a single cycle. With these polynomials (degree 2 and 3 are sufficient) we can easily find not only channels but also self-similar structures. And anything at all: flags, pennants, head-shoulders, etc.

Here's a 18Mb screenshot showing the speed of primitive channel finding with various settings. The speed of calculating all channels over the entire data history for all TFs is about 70 microseconds ( 15000 times per second) for 2nd degree polynomials. Each such calculation implies calculation of several thousand polynomials.


 
Maxim Dmitrievsky:

Yes, that's what I mean, it doesn't matter which arrow, as long as it's reversed.

In my opinion it is not the right approach to understand the structure of Fr, in the general construction there is a predominance (or correlation) of one trend over the other.

 
Nikolai Semko:

I have developed an ultra-fast method for calculating polynomials of almost any degree (in reality up to about 15, because beyond that the problems with lack of accuracy of type double start) without a single cycle. With these polynomials (degree 2 and 3 are sufficient) we can easily find not only channels but also self-similar structures. And anything at all: flags, pennants, head-shoulders, etc.

Here's a gif that demonstrates the speed of primitive channel finding with various settings. The speed of calculating all channels over the entire data history for all TFs is about 70 microseconds ( 15000 times per second) for 2nd degree polynomials. Each such calculation implies calculation of several thousand polynomials.


Cool. Now all that remains is to read some statistics on the patterns, and if it satisfies the request, then look for bargains.

 
Veniamin Skrepkov:

In my opinion it is not the right approach to understand the structure of Fr, in the general construction there is a predominance (or correlation) of one trend over the other.

This is only a particular case (reversal). In reality I just look at the chart and scroll through hundreds of combinations in my head, if I see something familiar and repeating (in any interpretation) then it is a fractal

Realtime optimization strategy )
 
Maxim Dmitrievsky:

Well, that's cool. I just have to count some statistics on the patterns and if it satisfies the request, then look for deals.

Yeah, there's a lot of opportunity.
Sleeping... It's 2:00 in the morning. I get up at 6:30.
 
Nikolai Semko:
Yes, there are plenty of opportunities.
Sleeping... It's 2 a.m. I get up at 6:30.

good night))

 
Also on the subject of fractals, where to look for them. Our body is essentially built according to a fractal form. That is, dna contains a certain formula and then this or that part of the body is formed under the action of a large number of iterations. The finger is made according to the same formula, but a different number of iterations are made. In fact, the body is not like dna form, but it is built according to the formula. But each cell of the body contains this formula, with info on what stage (iteration) that part of the body is at. The example is very crude and does not pretend to be scientifically accurate, but it reflects my view of the process. There is some formula, and the trades are iterations. With each new trade, the chart takes a unique shape peculiar to that number of trades. It is possible that the formula is the same for all markets, but for each market there is a unique identifier, from which the development starts and then iterations are counted. But as in the case of dnc, the formula must be contained in each minimum plot.
 
Nikolai Semko:

I have developed an ultra-fast method for calculating polynomials of almost any degree (in reality up to about 15, because beyond that the problems with lack of accuracy of type double start) without a single cycle. With these polynomials (degree 2 and 3 are sufficient) we can easily find not only channels but also self-similar structures. And anything at all: flags, pennants, head-shoulders, etc.

Here's a 18Mb screenshot showing the speed of primitive channel finding with various settings. The speed of calculating all channels over the entire data history for all TFs is about 70 microseconds ( 15000 times per second) for 2nd degree polynomials. Each such calculation requires calculation of several thousand polynomials.


Nikolai, do you trade by yourself with these calculations?

If so, could you please describe the trading results in general terms?

 
Nikolai Semko:
Sleeping... It's 2 a.m. I get up at 6:30.

You poor man. So it turns out you have a heightened need for colourful canvass ornaments from lack of sleep, the perception of which creates a surrogate sense of dreaming.

1...11121314151617181920
New comment