Machine learning in trading: theory, models, practice and algo-trading - page 1785

 

https://writings.stephenwolfram.com/2020/04/finally-we-may-have-a-path-to-the-fundamental-theory-of-physics-and-its-beautiful/

Nice article. There is even one for trading. Simple rules give rise to complex and often without the ability to predict behavior, and there is often no reversal either. By the behavior of a system it is often impossible to model the rules of its behavior.

Finally We May Have a Path to the Fundamental Theory of Physics… and It’s Beautiful—Stephen Wolfram Writings
Finally We May Have a Path to the Fundamental Theory of Physics… and It’s Beautiful—Stephen Wolfram Writings
  • 2020.04.14
  • writings.stephenwolfram.com
It’s unexpected, surprising—and for me incredibly exciting. To be fair, at some level I’ve been working towards this for nearly 50 years. But it’s just in the last few months that it’s finally come together. And it’s much more wonderful, and beautiful, than I’d ever imagined. In many ways it’s the ultimate question in natural science: How does...
 
mytarmailS:

thanks

Here is one of them, for example.

The usual trend algorithm, no parameters, no tweaks , there is only one sweep that smoothes one signal line

I think this robot has a very good effect, but I want to reiterate that it has no parameters and no optimizations, only adaptivity and spectral analysis. You can't do that with ordinary indicators.

If you want to implement such a robot you should write in MT4 the principal components method and the calculation of the Pearson correlation coefficient.

If I wanted to use R, I would need two lines of code, because everything is already there.

If you want to solve this problem you need to understand what you are coding. I do not know how to calculate the "method of principal components", if you describe everything step by step, you can do it step by step, but even the wikopedia will not help me. An alternative solution may be to create a separate branch with this problem, where people will help to decipher incomprehensible phrases and formulas, and the result will be open.

And in general there is everything for MT4 andMT5
 
Aleksey Vyazmikin:

To solve a problem like this, you need to understand what you're coding. I do not know how to calculate the "method of principal components", if you describe everything step by step, it can be done step by step, but otherwise even wikopedia will not help me. An alternative solution may be to create a separate branch with this problem, where people will help to decipher incomprehensible phrases and formulas, and the result will be open.

There are all for MT4 and MT5

It would be nice to have a branch for meaningful discussion of matstat/theorist methods. I am not sure that this is feasible.

 
Aleksey Nikolayev:

It would be nice to have a branch for a meaningful discussion of matstat/theorist methods. I'm not sure if this is feasible.

As sad as it is, I agree.
 
Aleksey Nikolayev:

It would be nice to have a branch for meaningful discussion of matstat/theorist methods. I am not sure that this is feasible.

If there will be a concrete examples, and the branch will be led by a responsible person, giving answers to all questions of the level of the training course, it may be useful. Difficult questions will already be discussed separately, with opinions of different participants.

It would be good to start such a branch with examples of Forex trading or IO, and more and more people will follow.

 
Valeriy Yastremskiy:

https://writings.stephenwolfram.com/2020/04/finally-we-may-have-a-path-to-the-fundamental-theory-of-physics-and-its-beautiful/

Nice article. There is even one for trading. Simple rules give rise to complex and often without the ability to predict behavior, and there is often no reversal either. By the behavior of a system, it is often impossible to model the rules of its behavior.

https://ru.wikipedia.org/wiki/%D0%90%D0%B4%D0%B2%D0%B0%D0%B9%D1%82%D0%B0-%D0%B2%D0%B5%D0%B4%D0%B0%D0%BD%D1%82%D0%B0

 
)))) You know how to distract / entertain. The simple rules of motivating traders' actions create a complex system of price behavior)))) I'm going in circles, the perpetual motion machine does not work yet )))))
 
Valeriy Yastremskiy:

https://writings.stephenwolfram.com/2020/04/finally-we-may-have-a-path-to-the-fundamental-theory-of-physics-and-its-beautiful/

Nice article. There is even one for trading. Simple rules give rise to complex and often without the ability to predict behavior, and there is often no reversal either. By the behavior of a system you often can't model the rules of its behavior.

This article is a bomb, I understood nothing, but I read it with my mouth open... Thank you!


I even got an idea, if a lot of random rules will still converge to a common structure, just in different ways, then I can draw an analogy with the algorithm Random Forest, its creators have conducted many tests and it turned out that it doesn't matter either the sequence of rule formation or breakdown, just a simple random can always get similar results ...


So I think what if we take for example a 5 min/hour chart and take it as a certain big pattern - "BP", and within it generate a sample using different sliding windows (of course with normalization by scale)

Then train a Forest on this sample, i.e. generate a bunch of random rules inside BP

Then scale BP to sample mastab and predict BP by its internal rules which were generated earlier...

Take into account fractality and check if all this mutual nesting works

Interesting...

 
Valeriy Yastremskiy:
))) you know how to distract / entertain. The simple rules of motivating traders' actions create a complicated system of price behavior)))) I'm going around in circles, the perpetual motion machine does not work yet )))))

I suggest not to rack our brains, but to think of systems that once worked / are currently working. Based on many years of experience of different traders.

Such as: volatility breakout, return to the mean (no matter what), patterns (cycles), arbitrage.

All these TS are event-driven, i.e. they are triggered when an event occurs. They do not approximate everything, like in MO.

MO without clear rules for trading is a real dumb-headed dummy.
 
Maxim Dmitrievsky:

I suggest not racking your brains, but thinking about systems that once worked/worked at the moment. Based on many years of experience of different traders.

Such as: breakout volatility, return to the average (no matter what), patterns (cycles), arbitrage

All these TS are event-driven, i.e. they are triggered when an event occurs. And not approximating everything, like in MO.

IO without clear rules in trading - is a real dumb-headed dummy

I agree) Many years of experience only no answers.

I agree, I don't like arbitrage.

I would add overbought/oversold synergies to events, but not a fact.

Met a MOSHKA MA and how I started to mUch it))))

So far, I'm bogged down by BP's characteristics. Events of volatility, reversion to the mean, cycles work only with certain characteristics of the series. If they change, they do not work anymore. What those characteristics are, I don't know. I am looking for the answer. One thing I understand is that the characteristics must be obtained from different scales of the series by the same simple methods. At the edges of the scales, month and tick there may be other methods. There should be logical / event levels of characteristics, to understand the state and its changes. But as soon as I take rates besides increments, for example, I get confused. It turns out too complicated.

Of course, I can just take different simple TSs, see where they stop / start working and compare characteristics, but so far everything is complicated, and it is not clear what to look at.