Machine learning in trading: theory, models, practice and algo-trading - page 2951

[mytarmailS]

Roman #:

The difference is that the whole meaning of the concept of "machine learning" is lost.
The point is that a moving average should be trained once over a long period of time, and the model should produce accurate results.
Such a model fits into the concept of "machine learning".

And retraining every five minutes fits into the concept of "constant adjustment to the current situation".

First of all, that's not what my question was about.

Secondly. Who has defined the rules of what is considered machine learning and what is not, where can I find them?

[Roman]

mytarmailS #:
First of all, that's not what my question was about.

Secondly. Who has defined the rules for what is considered machine learning and what is not, and where can I find them?

There must be some logic in reasoning.
For example, the same moving average.
Why? To use machine learning to find a moving average,
if the moving average can be calculated mathematically.
Also think about the difference.

[mytarmailS]

Roman #:

There has to be some logic to the reasoning.
Using the same moving average as an example.
Why would you use machine learning to find a moving average,
if the moving average can be calculated mathematically.
Also think about the difference.

Oh my god, well that was an example if I replaced the moving average with a multidimensional dynamic space in a sliding window.

Would that make you feel better? It doesn't change the point

[Roman]

mytarmailS #:
Oh my god, that was an example if I replaced the moving average with a multidimensional dynamic space in a sliding window.

Would that make you feel better? It doesn't change the point

I see that you don't get the concept of what I said.
I don't want to continue the conversation on this topic.

[mytarmailS]

Roman #:

I see you don't get the concept of what I'm saying.
I don't want to continue this conversation.

It's your concept, not a common one... That's why I don't get it, and rightly so....

Tell all the algorithms with online learning, robots, reinforcement learning, etc.. that they're not machine learning, because you've invented that machine learning is ad hoc learning....

[Maxim Dmitrievsky]

The level of abstraction of a techie allows you to draw conclusions in 7 hours, not 7 years :)

Exploration, using RL as an example, works (has predictive value) only in slowly changing environments, not jumping. Otherwise you really get a lagging MA.

[mytarmailS]

Maxim Dmitrievsky #:
The level of abstraction of a techie allows you to draw conclusions in 7 hours, not 7 years :)

Exploration, using RL as an example, works (has predictive value) only in slowly changing environments, not jumping. Otherwise you really get a lagging MA.

I don't see a contradiction, rather the opposite, my same thoughts but in different words

[Evgeny Dyuka]

Renat Fatkhullin #:

I have shown my achievements - they are visible to everyone. But yours are quite bad - you have only comments on this forum.

Do you have anything to show, any code publicly available? Do you have anything tangible?

If I may, I have a similar counter question.
(this is not about your business, but specifically about the topic of MO).

[Sergey Golubev]

Evgeny Dyuka #:

The models are already a year old and I don't plan to retrain them yet. The neural network predicts several crypto trading pairs and publicly signals its forecasts with the duration of the forecast.
After each signal, after the signal time expires, a screen of performance is published.
Everyone is given a free trial period to make sure that the signals and reports correspond to each other.
Is this normal?

Don't publish Telegram channels here.
There was a dispute with mytarmailS (not with you), for example, he is Renat:

So, like I said, you're a programmer....
And how to train networks, how not to train, what data to feed, how to feed, you have no idea...
So don't talk about what you have no idea about and don't criticise what you are far from

and then, after Renat left the thread - this post #29493 (and there was a dispute there not with Roman), where mytarmailS declares himself the Teacher of
(i.e. MQ should learn from him).
But then in a private message he wrote me all sorts of nasty things about mql5, about this Internet resource, that he will not post any codes/scripts here, but will only make posts, that is - talk (because, as he says - here is "miserable mql", and everyone should learn from him, and so on).

Banned the user mytarmailS for a day.

----------------------.

I mean, he's gone mad too ... (first at JeeyCi, then at mytarmailS ) ... what is this branch?

[Evgeny Dyuka]

Sergey Golubev #:

You don't need to post your Telegram feeds here.

so ban them already, it's high time
you have a business here and you are annoyed that people write wrong things instead of spending their efforts on developing MO in MQL5 environment.