Machine learning in trading: theory, models, practice and algo-trading - page 1516

New comment
 
Biqvi:

Good afternoon.

There is a pattern that I see with my eyes, but I cannot code it due to its complexity and not quite a clear understanding of what I see. The core is understood and formalized, but there are details that are caught by "lateral vision", are considered, but are not transmitted to the consciousness. The pattern has been successfully traded with my hands for many years.


I would like to teach the neuronku to see it (more precisely the part that set-up 1) to solve two problems:

1) to take out what she saw, to understand "what she was hooked on" in the set-up and through this better understand what I see and what exactly the graph characteristics are important.

2) To transfer the trade to it, or (a minimum variant) hang a bell.

The question to the pro, tell me if the problem is set correctly, and where to move to solve it?

P.S. I take it easy to the fact, that it will have to spend a year or two for a decision and even calmer to collaboration with the pro.

Now I visually analyzed the chart for 5 years, TF D1

I did not see any such patterns.

Please give an example from a real chart

 
Dear participants of this branch, is there computer with Windows 7 64x - very necessary to check run the latest console version of CatBoost - enough just to download and run the executable. I have stopped working the last two versions - the developers claim that they should work, but like they have all machines on Windows 10 and there everything works.
 
Aleksey Vyazmikin:
Dear participants of this branch, is there computer with Windows 7 64x - very necessary to check the performance of the latest console version of CatBoost - just download and run the executable. I have stopped working the last two versions - the developers claim that they should work, but like they have all machines on Windows 10 and everything works there.
there is an old beech with a semerka 64 proffesional, let me check the link
 
Ivan Nehrishnyi:
There is an old beech with a 64 Professional

Here is a link to download

 
Aleksey Vyazmikin:

Here is a link to download

ok downloading, if there are complicated problems, then tell me...

1. After downloading, protection scolds

2. 0xc0000005 error at startup

maybe the file is corrupted, my file size is 148 395 520 catboost-0.15.2.exe
 
Ivan Nehrishnyi:

ok downloading, if there are complicated things there, then tell me...

1. after downloading the protection scolds

2. 0xc0000005 error at startup

maybe the file is damaged, I have the size 148 395 520 catboost-0.15.2.exe

Thank you - yes the error when working with Windows 7 - I have the same problem - if you have the opportunity to write here to confirm the error.

 
Kesha Rutov:

Neural network (MLP) and other classifiers (random forest, SVM, kNN etc.) are needed to automatically to look for such and much more nontrivial patterns, For your problem, a simple convolution (sliding scalar product) will do, it's programmed from scratch in an hour, and with ready-made tools for minutes, you don't need year.

But I can disappoint you in advance that the probability of success is close to zero, because all such simple structures are found without problems by automata, and if you were able to trade hands in the plus, it means that besides the pattern you relied on a number of auxiliary conditions, which are probably "obvious" to you, but nevertheless significantly affect the result. Remember the tale about the "axe soup"? It's the same for many candlestick shapes of manual traders, a simple pattern, but before that the trader looks at all the news, all the markets, listens to gossip and trades or not his simple pattern)))

The task is as follows.

1) I have a known pattern. I want to put a set of data: section 2 (the resultant shot after the set-up) and N bars before it, so that it finds something of its own in these N bars. That is, yes, the same automatic search, but in pre-selected areas where I know it is there, and not on the entire chart.

2) To look inside the neuronics and understand what it has found.

The pattern is complex, the picture was given just to understand that there is area1 and area2.

 

I have some theoretical-philosophical questions almost on the subject.

I would like to hear more or less detailed answers, not born in a hurry.

 

Question 01.

The following indicators have only one external input parameter

  • Exponential Smoothed - period/coefficient.
  • PriceChannel (highest and lowest prices on the interval) - size of the interval.
  • ZigZag - minimum knee size.

I have chosen these indicators because of the minimum of external input parameters.

Is it possible to reproduce their algorithms with the help of MO methods. I.e. take any history, run the indicators with any parameters and feed them to MO. Is it possible to obtain the corresponding indicator algorithm in the output?

 

Question 02.

Let the TS with four input parameters perform thousands of transactions on some interval. Add as a filter two input parameters with a low number of possible variants. At the output we have a straight up graph with about one thousand trades. And all of them are more or less evenly distributed over the entire testing area.


What is the reason of the high probability of losing 5% of the initial interval on OOS? A huge interval and only six inputs gave an upward straight line on a really large number of trades. And what came out was a bullshit fit.

Does this mean that there are actually more than six input parameters? It is a sort of a reference to the first question: aren't algorithms that are simple to us really complex by their nature?

1...150915101511151215131514151515161517151815191520152115221523...3590
New comment