ForumSections

Machine learning in trading: theory, models, practice and algo-trading - page 1580

New comment
Maxim Dmitrievsky:

Well, cointegration.


we usually discover this happiness after the fact

so it is of little use, except to tick the box - these rows WERE cointegrated quite recently

[Deleted]
Boris:


we usually discover this happiness after the fact.

so it's not much use

there's nothing else out there, I think.

Maxim Dmitrievsky:

I don't think there's anything else.

that's what we're talking about.

sad

or here, for example, are 2 rows, each made up of their own increments, or differences, as others say

or this

or this

or this


Anatolii Zainchkovskii:
Boris, a synthetic is just the sum of financial series. The only thing you can do on synthetics is to narrow the variance indefinitely. Hence, I suggest continuing to observe the diverging curves of the original series, not synthetics. You will see the same picture, with the same expanding curves.Alas, I can't tell you how to trade this expanding pattern. But I can suggest that if some part of the curvatures began to change the directional vector, most likely some other part should also change its vector of directionality.
It would be more interesting to find out when this expanding cloud of curves has a strong offset from the center, then you can collect statistics and try to identify something useful. But this is not the MO, and therefore I apologize for throwing in information off-topic.

it has a place, though not very strong, but it seems quite stable

and that's why the question arose about investigating possible overtraining

Colleagues,

I congratulate you with the coming new year, I wish you that next year you will receive only generalized models adequate to the market. Good luck!!!!

And for your refreshment. The hit of this new year


The CatBoost video is more for programmers :)


Aleksey Vyazmikin:

The CatBoost video is more for programmers :)


Interesting and not very clear :-(

I'm in ahtubing bros. Today is a day off and I'm missing the last symbol prices in the market overview. Is there any way to update them through tick emulation or something. I don't want to count without last prices :-(.

I searched the whole forum but could not find anything alas

[Deleted]
Evgeny Dyuka:

I'm new here and the thread is big, if it's not hard to give me a sammari - is it possible to predict quotes using neural networks?

I'm very deeply into it, I work with a Google colab + TensorFlow + Keras + Telegram (I signal the results for myself).

In short, the theme is: who starts working, he leaves and does not tell anyone. Who does not begin to work - just go away.

I came to the fact that the stat analysis is fuerst, and then the cherry in the form of MO on top. Personal opinion. Perhaps there are effective ways to do stat analysis on automatic via AI, I have not succeeded.

not statistical analysis, but econometrics fuerst, that's more accurate. NS is just a "why not?" type of thing.
1...157315741575157615771578157915801581158215831584158515861587...3749
New comment