Machine learning in trading: theory, models, practice and algo-trading - page 3729

[Rorschach]

Maxim Dmitrievsky #:
What's there to try, symbolic regression?

"The key idea is simple: train a conventional machine learning model and then extract a mathematical formula from it. We use polynomial regression with Ridge regularisation to predict price and logistic regression for direction. SymPy then turns the coefficients into equations."

[[Deleted]]

Rorschach #:
"The key idea is simple: we train a conventional machine learning model and then derive a mathematical formula from it. We use polynomial regression with Ridge regularisation to predict price and logistic regression for direction. SymPy then turns the coefficients into equations."

Well symbolic regression. Next, I would be wary of reading articles like this (dangerous for the brain)

the most useful thing is just to know that there is symbolic regression :)))

[Aleksey Nikolayev]

It looks like Gemini 3 was trained on mql5.com - probably the first time the AI produced code that immediately compiled and ran without errors.

[[Deleted]]

Aleksey Nikolayev #:
It seems that Gemini 3 was trained on mql5.com - probably for the first time the AI produced code that was immediately compiled and launched without errors.

input bool getprofit = true? :)

https://xbox-dns.ru/ for more pleasant usage

[Aleksey Nikolayev]

Maxim Dmitrievsky #:
input bool getprofit = true? :)

if(getprofit) {MessageBox("BABLO?", "BABLO?"); Alert("BABLO!");}

Maxim Dmitrievsky #:
https://xbox-dns.ru/ for more pleasant usage

Thanks, I'll try to add it to my linux.

[Aleksey Nikolayev]

Maxim Dmitrievsky #:
input bool getprofit = true? :)

Asked to write a script with the most complete check of admission to trading. The result seemed to be based on examples from Short's book.

[[Deleted]]

Aleksey Nikolayev #:

if(getprofit) {MessageBox("BABLO?", "BABLO?"); Alert("BABLO!");}

On every tick :)))

[[Deleted]]

Aleksey Nikolayev #:
Asked to write a script with the most complete verification of admission to trading. The result seemed to be based on examples from Short's book.

Cloud 4.5 can do something there too. Asked to draw a simple panel on the canvas. Almost managed it.

[Renat Akhtyamov]

Aleksey Nikolayev #:

if(load_profit) {MessageBox("BABLO?", "BABLO?"); Alert("BABLO!");}

Thanks, I'll try to bolt it on my linux.

It seems simple.

but no, it doesn't work ;)

[[Deleted]]

Maxim Dmitrievsky #:

The trick of the markup is that it does not look ahead, i.e. without looking ahead. For example, signals using any oscillator (overbought/oversold). Then generalisation.

Therefore, the errors of the models are small, because they approximate the indicator in essence. But it is not possible to find such a magic oscillator for all pairs.

Attempt at a more detailed description of what was done:

https://www.mql5.com/en/articles/20343