Machine learning in trading: theory, models, practice and algo-trading - page 2978

 
https://youtu.be/_j5HP2wrBPE
I don't understand much of it, but it looks good.
 
What is it, if not a fantasy on minimal, questioned in the code
 

Igor Ashmanov on the Alice voice assistant

About two fundamentally different approaches to AI programmes.

Игорь Ашманов о голосовом помощнике «Алиса»
Игорь Ашманов о голосовом помощнике «Алиса»
  • 2017.11.10
  • www.youtube.com
Мы просто спросили Игоря Ашманова об «Алисе» от Яндекса, а узнали об истории голосовых помощников, перспективах развития нейронных сетей и разобрались в том,...
 

Finally got a good guide on casual.

It answers the question why ML is only for prediction, but not for causality.

ML is notoriously bad at this inverse causality type of problem. They require us to answer "what if" questions, which economists call counterfactuals. What would happen if I used another price instead of this price I'm currently asking for my merchandise? What would happen if I do a low sugar one instead of this low-fat diet I'm in? If you work in a bank, giving credit, you will have to figure out how changing the customer line changes your revenue. Or, if you work in the local government, you might be asked to figure out how to make the schooling system better. Should you give tablets to every kid because the era of digital knowledge tells you to? Or should you build an old-fashioned library?

At the heart of these questions, there is a causal inquiry we wish to know the answer to. Causal questions permeate everyday problems, like figuring out how to make sales go up. Still, they also play an essential role in dilemmas that are very personal and dear to us: do I have to go to an expensive school to be successful in life (does education cause earnings)? Does immigration lower my chances of getting a job (does immigration causes unemployment to go up)? Does money transfer to the poor lower the crime rate? It doesn't matter the field you are in. It is very likely you had or will have to answer some type of causal question. Unfortunately for ML, we can't rely on correlation-type predictions to tackle them.

 
Maxim Dmitrievsky #:

Finally got a good casual guide.

It answers the question why MO is only for prediction, but not for causality search.

Well, if an algorithm cannot predict a sample, then it has no value, so there is no point in explaining it.

So being able to predict correctly is the first condition for touching the data at all.

Isn't that right?
 
Maxim Dmitrievsky #:

Finally got a good casual guide.

Most of all it looks like an "unnecessary" matstat)

Basically, it is suggested to see what methods are used, for example, in evidence-based medicine and try to apply them to your task.

 
mytarmailS #:
Well, if an algorithm cannot predict a sample, then it has no value, so there is no point in explaining it.

So being able to predict correctly is the first condition for touching the data at all.

Isn't that right?
You know, like causality tests. You can isolate the sets that work, as I understand it.
It's interesting to make an automaton that does everything itself. Maybe I could use something from there.
 
Aleksey Nikolayev #:

Most like an "unnecessary" matstat)

Basically, the suggestion is to see what methods are used, for example, in evidence-based medicine and try to apply them to your task.

I'm not familiar with this science yet either, but it's getting more and more praise on the internets.
There's stuff about meta lerners and dual machine learning. It's remotely similar to what I've been doing and I liked it.
I don't think it's worse than predicting muving lstm-cnn network :)
 

As one of the virints of the fitness function.

The task for AMO is to train in such a way that the forecast built on the equity of AMO trade was as good as possible.



I do not want a beautiful curve of equity on history, but I want to get a confident forecast in future trading...

Forecast with confidence intervals, the same statistical test...


I used two algorithms for forecasting, auto arima and holt.

Here you can see the area where the forecast "guarantees growth" of equity.



 

What's the point?

It would be interesting to develop some statistical criterion for the validity of the obtained balance curve/model error/..... and to teach AMO under this criterion.