Machine learning in trading: theory, models, practice and algo-trading - page 1041

[Yuriy Asaulenko]

Maxim Dmitrievsky:

continue...

Continued in the textbooks.))

[Maxim Dmitrievsky]

Yuriy Asaulenko:

Continued in the textbooks.))

try to understand what I wrote and imagine, for starters. I'm not talking about classic acf.

[Dmitriy Skub]

Alexander_K2:

No, well, I could be wrong, Dimitri - I'm just not able to compare everything. I just saw the periodicity of ACF on EURJPY in Erlang of the 30th order and got happy... I did not see it before...

When you get tired of banging on the ice, think about the fact that both price/time scales are not linear in essence. This is if approached from the perspective of pure algo-trading (without understanding the market).

[[Deleted]]

Maxim Dmitrievsky:

Regarding the ACF - I continue to stick to the opinion that returns should be inverted mirror-like after a certain point, and look for the best point where there is a clear periodicity in the acf on both pieces. Then predict something based on the past inverted returnees. But so far, of course, I haven't done it myself :)

Errors, by the way, will also be very significant, but prediction will not be so random as with all these autoregressions and garches. + There you have to modify the model specifically for different situations.

That's silly...

No wonder you don't know or understand what an autocorrelation function is, what it gives, what it can and can't do, what it's used for -- i.e. its scope of application, and the tasks assigned to it.

s.

have you figured out what approximation means?

[Maxim Dmitrievsky]

Oleg avtomat:

That's silly...

No wonder you don't know or understand what an autocorrelation function is, what it does, what it can and cannot do, what it is used for -- i.e., its scope of application and the tasks assigned to it.

Olezhek, go graze in your own waving thread. This one is for intelligent, intellectually mature individuals. On Asaulenko too, do not look, he has long since ...

[[Deleted]]

Maxim Dmitrievsky:

Olezhek, go graze in your own thread with mushrooms. This one is for intelligent, intellectually developed individuals. Don't look at Asaulenko either, he's been...

I'm twice your age, and you keep your homosexuality to yourself.

[Maxim Dmitrievsky]

Oleg avtomat:

Listen, dummy, don't be rude. I'm twice your age, and you keep your gayness to yourself.

You're older than me and you write like you're three years old, Oleza. Late adulthood, I guess.

[Igor Makanu]

Alexander_K2:

2. the ACF in the sliding window must be:

here is a good article, with examples https://www.mql5.com/ru/articles/292

[mytarmailS]

Hello!

Is there an expert of the fractal theory among the participants, can someone explain how it can be used in the forecasting of series, well, we have counted the Hearst, calculated the fractal dimension of the series and then what to do with it?

Is it possible, for example, with the help of the fractal theory to make it unnecessary to look at many timeframes?

Is it possible to find the same patterns on different timeframes using this theory?

[Grigoriy Chaunin]

You can't. Check currencies for the Hearst index. It clearly shows the randomness of the market. And what can you think of in a random market? Only martin. But on the other hand, there are inefficiencies in the market of different times of existence. And people make money on them. And this is not a randomness. That's the direction in which inefficiencies should be searched for. I would like to automate this process. But I cannot feel what to start from. Neural networks are not suitable for this. They need ready-made patterns for learning.