Machine learning in trading: theory, models, practice and algo-trading - page 2813

 
Maxim Dmitrievsky #:
Agent states, or actions. I suggest you spend a couple of months reading books to understand what you have written about, and come to the same conclusions ) without the reaction of the environment to the agent's actions, there is nothing to optimise, it is done in one pass.

There are environment states, agent states, matrices of transitions (policies) of the agent from state to state, taking into account changes in the environment. Your environment is static, it does not change due to the agent's actions. That is, you only need to define the matrix of agent's actions in a static environment, i.e. targets. The marking of targets is done in one pass.
I'm still dumb about describing the state of the row manually.))))))
 
Valeriy Yastremskiy #:
I'm still stumped on describing the state of the row manually.))))))
Futile
 
Maxim Dmitrievsky #:
Futile.
No argument there, but fascinating)))))
 
Valeriy Yastremskiy #:
No argument there, but fascinating)))))
There are 2 states there - shifting the mean increments up or down
 
No one's seen my set date?
 
Valeriy Yastremskiy #:
I'm still stumped on describing the state of a series manually.))))))

I recently came across a video explaining the Markovian approach to state transition.
I'm not saying that these states should be used.
It just seemed that you can apply this concept to any states you think necessary.
Maybe it will give you some other ideas.



Maxim don't make fun of me for being a Hindu again ))
I haven't met any others )

 
Roman #:

I recently came across a video explaining the Markovian approach to state transition.
I'm not saying that these particular states should be used.
It just seemed that you can apply this concept to any states you see fit.
Maybe it will give you some other ideas.



Maxim don't make fun of being a Hindu again ))
I haven't come across any other )

You can find articles for time series segmentation too. You can replace clustering with clustering. It probably makes sense to train different models for each of the states, because there will be different characteristics. Basically, it is the shift of mean increments, at the change of which the models break down.
 
Maxim Dmitrievsky #:
It can be used for segmentation. You can replace clustering.
You can't, it's different.

Hmm predicts what cluster you are in now, Clustering shows what cluster you were in, post facto.
Simply put.
 
mytarmailS #:
You can't, it's different.

Hmm predicts what cluster you are in now, Clustering shows what cluster you were in, post factum.
Simply put.
And if you think about it.
 
Maxim Dmitrievsky #:
On second thought
mytarmailS #:
You can't, it's different...

Hmm predicts what cluster you are in now, Clustering shows what cluster you were in, post factum.
Simply put.

I've thought, I've tried, I've experimented, I've written code...