Machine learning in trading: theory, models, practice and algo-trading - page 1186

 
Igor Makanu:

and? what difference does it make which programming language the developer wrote in? imho,python, almost always, is an interpreter, for games performance is important, so the use of python is more of a disadvantage for game developers than a plus

another issue is that as a modeling system, Python can be attractive because the language is actively developing (promoted), not tied to the platform, Python supports all possible types of programming: Structured, OOP, functional programming - this point is very important when simulating, there is no need to "reinvent the wheel" and then port this "wheel" to other platforms - you take and write mathematical formulas as you see in the tutorial, and Python itself will understand that it is matrix multiplication or work with the function

Python is seen here as a ready-made tool for NS/MO, pretty much all the tutorial material on learning AI (for students), is now laid out in the framework of Python.

StarCraft2 bots are written not by game developers, but by enthusiasts like you and me, of course their level of knowledge of the theme and programming is much higher than that of the majority of people on this forum.

The most important thing for me to remember is that I want to be able to write a scalping strategy for the beginning of the day and then spend 5 to 30 minutes on analysis, and that's not the most critical thing that can happen.

And not mediately about the game itself, Python's performance is more than enough to play StarCraft2 against a human player.
 
Maxim Dmitrievsky:

You came up with your own nonsense and now you're a little jealous of it.

It's okay, it's okay... everyone has to go through this

 
govich:

It's okay, it's okay... Everybody has to go through this.

What? Please don't drag everyone with you. ))

i wrote above that python is easier to use for matrices and iO, less stress, more output

 
Maxim Dmitrievsky:

What? Please don't drag everyone with you... ))

I wrote above that it's easier to work with python matrices and MO, less stress, more output

It's ok, it's cool, I'm not judging, on the contrary, encouraging, envying, it just made me laugh:

Maxim Dmitrievsky:

As you wrote yourself, through a neural network you can get any indicator (conditionally). It remains a matter of habit to switch to a more powerful language and just download the price history

It's a wonderful feeling, like on New Year's Eve waiting for presents under the tree, it seems that neural networks can do everything, but in Python you only need to load the library via pip and write a couple of lines, it's awesome!

 
govich:

it's a wonderful feeling, like on the night before the new year, waiting for presents under the tree, it seems that neural networks can do everything, and on python it is enough to load the library via pip and write a couple of lines, it's super!

that wasn't the point, it's understandable that it doesn't cancel out the need to be a geek

about adaptivity... It's been 4 years since your thread, but try

http://setosa.io/conditional/

as opposed to the freqqensionist... ugh, frequency, approach, allows for adaptivity. The alignment of distributions will have some lag, but if the frequency of events is sufficient, the alignment will be pretty fast under the current sentiment

Conditional probability explained visually
  • setosa.io
P(|) = {{ pOfBGivenA | number:3}} {{ pOfBGivenA * 100 | number:1}}% If we have a ball and we know it hit the shelf, there's a {{ pOfBGivenA * 100 | number:1}}% chance it also hit the shelf. P(|) = {{ pOfAGivenB | number:3}} {{...
 
Maxim Dmitrievsky:

That wasn't the point, it's clear that it doesn't cancel out the need to be a geek.

about adaptability... It's been 4 years since your thread, but try

http://setosa.io/conditional/

as opposed to the freqqensionist... ugh, frequency, approach, allows for adaptability. The alignment of distributions will have some lag, but if the frequency of events is sufficient, the alignment will be pretty fast under the current sentiment

When I looked at the last life the question was not about adaptation, but about the grail. An inductor itself changing its parameters is useless in itself, if it doesn't predict the future, but I was green at that time.

 
govich:

That was in a past life, the question was not really about the adaptation, but about the grail, in itself an index that somehow changes its parameters, no one needs in hell, if it does not predict the future, but I was green then

and now what color are you? what did you learn?

 
Farkhat Guzairov:

Python is considered here as a ready-made tool for NS/MO, almost all of the tutorial material on learning AI (for students), is now presented in the framework of Python.

Bots for StarCraft2 are written not by game developers, but by enthusiasts like you and me, of course their level of knowledge of the topic and programming is much higher than the majority of people on this forum.

I don't think that it's really that much of a problem if you're not going to write a scalper in intraday trading strategy, spending 5-30 minutes on analytics is not that critical that can't be.

Well and not mediately about the game itself, Python's performance is more than enough to play StarCraft2 against a human player.

Forgot to add that games use RL which I'm trying to apply to trading

11:50 :))


 
Maxim Dmitrievsky:

Forgot to add that games use RL, which I try to apply to trading

11:50 :))

this is cool! I looked it up, there are even packages in the r-ka already

 

Guys! If anyone has a month of free time and itch to program, I can tell you about the way of earning from 5 - 15% a month guaranteed, or even more, it is not a racket and not a joke.

Requirements:

1) the ability to create a GUI

2) working with the web api

3) Possibly parsing

Please, write only zainterisovannye, and not like: you tell me, and I'll think, and then start smacking, hu ... kat, you can safely go right away.