Machine learning in trading: theory, models, practice and algo-trading - page 2830

[mytarmailS]

Andrey Dik #:
you'd better keep quiet, you'll look much smarter or at least better mannered.

You should study the database BASE!!!!

What is local optimisation, global optimisation, types of functions, types of optimisation, types of optimisation, what kind of optimisation to apply to what function, etc....

Discrete optimisation, continuous optimisation, multicriteria, etc... what is the difference, what is the purpose, where to apply one and not to apply another....

You don't know the basic things!!!

Why should I keep silent if I have something to say on the matter, I am not a non-stupid person who just wants to say something.

[Maxim Dmitrievsky]

He's the one who's usually so excited about Googling. When he's googled something and he's happy to share it.

[mytarmailS]

You are not allowed to think, people like you are not allowed to think, just nod your mane and agree with everything they say.

[СанСаныч Фоменко]

Maxim Dmitrievsky #:

You have been suggested to test learning/optimisation on a few representative functions, this is good practice

If you think neural networks do it perfectly, you are probably wrong

There is a specific perception there, at the level of kargo-cult and belief in the divine R, bringing the gifts of civilisation.

I always have more faith in professionals who spend their whole life dealing with one problem, here optimisation and, in particular, gradient descent.

And the main sign of amateurs is to speak disparagingly about professionals. R is a professional language, the benchmark in statistics today. It's time to learn that, instead of writing all sorts of crap about "faith and kargo-cults".

[Maxim Dmitrievsky]

СанСаныч Фоменко #:

I always have more faith in professionals who have spent their whole lives working on the same problem, here optimisation and in particular gradient descent.

And the main sign of amateurs is to speak disparagingly about professionals. R is a professional language, the benchmark in statistics today. It is time to learn this, and not write all sorts of crap about "faith and kargo-cults".

Certainly you believe professionals, but you have not named a single one and have not given a list of works on this topic

It turns out to be another meaningless post. Please cite, please, also, where it is written that R is a benchmark in the field of statistics and what statistics has to do with it

For me R user has become on this forum a nominal name, with an implication, that's why I am laughing with it. It is only relevant within this thread and nowhere else.

For example, once long ago was on the course 1C, progi for accounting. And the course was led by one stinky (literally) student, who smelled sweat all over the audience and all turned their noses away. Then everyone laughed at variations on the theme. A similar situation happened here, which somewhat spoils the attitude towards the language. Impressions of the language 😀😀

[СанСаныч Фоменко]

Maxim Dmitrievsky #:
1. you certainly believe in professionals, but you haven't named a single one or given a list of works on this topic

2. It turns out to be another meaningless message. Please cite, please, also, where it is written that R is a benchmark in the field of statistics and what statistics has to do with it

As long as no facts are given, it all looks more like shouting of sectarians. R itself is normal language, not outstanding in any way. I would say that it is a bit superfluous, because it is similar to Python - a benchmark in statistics and machine learning and high-level programming in general. I don't understand why you should denigrate it in every post with such shouts and then urge to use it. Is this what professionals do?

1. Dick's question is a perfectly valid and correct question. I don't use NS, but I know for a fact that any function in any R package necessarily contains a reference to the author of the algorithm, and for serious algorithms, a reference to the article/book that describes the algorithm implemented in R. Since you are well acquainted with NS, if you were using R, you could search in R for the corresponding type of NS and find the corresponding reference where the corresponding algorithm is described, find a discussion on the algorithm, find out all the nuances of the professionals ... and answer Dick at the highest professional level, instead of mumbling something obscene.

2. R by name: the language of statistics and graphics. The essence of R is revealed by the rubric of its reference apparatus.

Here is a list of topics that R packages cover. One of the topics is machine learning.

Here is a list of packages related to MO.

Until a few years ago, one could find competitors to R among other specialised statistics languages. For example, SPPS, to date I have not found any. R has remained the only statistical language, is supported and moderated, has a huge number of mirrors, is included in Microsoft software.

3. comparing R with Python is absolutely wrong.

R is a specialised language. Python is a universal language. Python far surpasses R in the number of users, but the mass user of Python is web design. The fact that Python has statistics packages does NOT allow it to be classified as a statistics language. On this basis, C++, in which the packages used by both R and Python are implemented, can be classified as a statistics language. Due to its detailed rubric and references to algorithms of proposed functions, R can be used to study the theory and practice of statistics, while Python cannot.

[Aleksey Nikolayev]

mytarmailS #:

Alexei, do you know anything about optimising the noisy function?

I have not studied the question in detail. The idea seems to be simple, but there are a lot of technical subtleties in the ways of implementation.

[Aleksey Nikolayev]

Maxim Dmitrievsky #:
There is a complete search, and there is optimisation. It is needed to reduce the time to find an optimal solution. Since it is so, it is always a compromise. You can optimise with the stochastic gradient method and get a better result than through adam, but sacrifice time. And one has to choose. For some tasks, accuracy may be more important than speed, for example, to increase the expectation of TC.
.

Here it is interesting just to look at the visual results.

An important question is what to optimise. I would like to optimise meaningful criteria tied to profit, drawdown, volatility, etc.

Complete overshoot is the best way to optimise) Unfortunately not always applicable).

[Maxim Dmitrievsky]

Aleksey Nikolayev #:

An important question is what to optimise. I would like to optimise meaningful criteria tied to profit, drawdown, volatility, etc.

Complete overshoot is the best way to optimise) Unfortunately not always applicable).

I would quote the slogan "fight and search - find and hide".

set as a custom metric any criterion, in particular these standard ones. It will still optimise by logloss, but it will stop at these custom ones, which probably makes some sense

and indeed it does, because stopping in the same bousting is always based on some cast criterion like accuracy.

[mytarmailS]

Aleksey Nikolayev #:

I have not studied the question in detail. The idea seems to be simple, but there are a lot of technical subtleties in the ways of implementation.

There is also a question of what is meant by finding the maximum in a noisy function....

As I understood the definition - "optimising a noisy function" - it means that the function is complex and it is hard to find the maximum, gradient algorithms are not applicable, and so on... Roughly speaking, it's not a big deal, applied global optimisation algorithm and he went to look for the global maximum....

But I look at it differently, I want to find the maximum of a noisy function but with removed noise, not the global maximum in a noisy function, but the global maximum in a de-noisy function....

(And it' s not trivial, because the function is not known and the noise parameters are not known....