Machine learning in trading: theory, models, practice and algo-trading - page 3229

New comment
 
fxsaber #:

I wonder if Kaggle would have done well.....

Finances, of course, could be an incentive....

 
fxsaber #:

If it's not enough, then what do you look for on MO when you feed in many years?

Well, I forecast every bar. Some people even predict 1 trade a day if some conditions are met, like you. It's different for everyone.

 
fxsaber #:

Theoretical question.


Condition.

  1. Someone has invented a very simple TS.
  2. Realised what should be done with quotes, so that on any Sample after optimization of the TS the best passes would show themselves perfectly on OOS.
  3. I generated a random sequence of quotes and added the result of point 2 to it.

As a result, there is a certain history, which definitely has a regularity. This history can be of any required length.


Question.

Will a super-advanced (someday in the future) MO technique with infinite computational power find a TC (data from item 3) as in item 1. or with properties as in item 2?

If we are discussing MO, we need to separate the flies from the cutlets:

1. there is a classification model that predicts the next step(s) with some error

2. there is an Expert Advisor that uses the model according to item 1 and trades on the account with some profit/loss.


I am able to create classification models according to point 1 that give less than 20% error. This classification error is about the same on the following four files:

1. a training file obtained by randomly sampling from the original file

2. a testing file obtained as an addition to a random sample of the training file

3. a validation file obtained as an addition to a random sample of the training and testing file

4. An additional regular file with the usual sequence of prices.

Here on the site, I've laid out many times the theoretical justifications for the stability of this result.

And on quotes, on prices, it is theoretically impossible to create an Expert Advisor that trades profitably, because financial series are NOT stationary, i.e. any history does not matter, with or without your tricks. The initial price series CANNOT be used to build a profitable system. And not only price series (quotes), but also any varieties of mashes.

 
Renat Fatkhullin #:
We are planning to launch another championship aimed at promoting neural networks:
1) provide a single MQL5 robot template with downloadable model.onnx
2) within 5 months, participants will upload their modules in the form of model.onnx
3) every day they will be automatically run on the history from 2023.01.01 to the current day on 4 major exchange rates
4) daily ranking of participants will be published
5) at the end of the preliminary period of accumulation of participants within 5 months, the real trading period will start within 1 month.
6) based on the results of work during 1 month winners will be determined
7) the prize fund from our company 30 000 dollars will be divided into three winners as 15 000, 10 000 and 5 000 dollars
8) we guarantee that at the end of the championship all model files will be erased to preserve the intellectual property of the developers

The purpose of the championship is solely to stimulate the development of machine learning in trading. Programmes only as a single unchangeable MQL5 template + model.onnx

The idea is good.

However, onnx is for model conversion, but how to generate a date for it? Or will there be ready-made predictors in the MQL5 template?

 
Renat Fatkhullin #:
The purpose of the championship is solely to stimulate the development of machine learning in trading.

100% of MO traders said YEEEEEEEEEEEEEEEEEEEEssss!!!!

Renat Fatkhullin #:
Programmes only as a single unchangeable MQL5 template + model.onnx

99.9% of MO traders said - oh shit ((( what the hell is onnx ))

 
Renat Fatkhullin #:
1) provide a single MQL5 robot template with downloadable model.onnx

Will there be any requirements for the complexity of models? Or are the models allowed to be so simple that it is possible to participate even with a "MACD" model?

 
fxsaber #:
is to find a solution so that we can
.

IO tools have not yet been able to do this on the fly.

And it won't.

I've been watching the results in this thread.

Every strategy is a pips.

I wonder how a programme with a 1 year deal will behave?

 
Andrey Dik #:

Will there be any requirements for the complexity of the models? Or are the models allowed to be so simple that it is possible to participate even with a "MACD" model?

Scalpers will be prohibited by the rules.

The goal is clearly stated - to stimulate the development of ML models for trading, and not to give an opportunity to make money in the old way, to fill up trivial scalpers under the guise of models, etc.

 
Renat Fatkhullin #:

Scalpers will be banned by rule.

Kitchen.

 
fxsaber #:

Kitchen.

MetaQuotes-Demo.

We have been running championships since 2006 and we know very well that any championship under ideal conditions (e.g. demo server) degenerates into a useless venture when scalpers are allowed. Other reasonable strategies are immediately at the bottom.

That's why we have always removed scalpers in order to let participants compete in more or less calm conditions.

The goal is not to give another scalper easy money. The goal is to stimulate more active development of serious machine learning models and their application in MQL5.

I recommend to look at the world not from the point of view of yourself and your personal optimisation strategies. I pointed out the function of championship optimisation right away.


History of Automated Trading Champions hips.

1...322232233224322532263227322832293230323132323233323432353236...3699
New comment