Machine learning in trading: theory, models, practice and algo-trading - page 2892
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Generally, there are penalties for unreasonableness.
I tend to believe that what people use to make their decisions works in the market, the only question is the amount of money at a moment in time of a group of people with a specific abstraction in their head regarding the vision of the market. I would even add meteorological data and the movement of space bodies - if I could get such data into the model.
The validity of the same approach can vary depending on time, instrument, timeframe, etc. So it is important to be able to check the validity on a case-by-case basis. Ambiguous methods do NOT provide such a check. It is always possible to take advantage of the uncertainty of the method and deceive yourself and others with the presence of a good option, which is selected after the trade is completed.
If there is no purpose to deceive oneself or others, then only unambiguously algorithmic methods are needed. You can test them and have an informed opinion about them. The specifics are not as important as the principle possibility of unambiguous formalisation and testing.
Unformalised approaches are a breeding ground for false gurus "who have learned the laws of the market".
Maybe we'd better try to reproduce the methodology of calculations of the Central Bank of the Russian Federation?
In my opinion, it's not the best time for such research).
In any case, all Central Banks are in one way or another copying the Fed, whose activities are more stable and better documented.
Because I think it's nonsense and I want you to realise that too
Well, I'm not expressing certainty, I'm talking about the possibility of research in this direction, and there the result can be different, perhaps there will be related branches in the ideas, which will just give the result.
So not "did", but could not do, otherwise the result would be. Nobody here tried to apply MO to this problem :)
Yes, I can work with one idea for a long time until I get results. Maybe it would be good to know five more languages, but when it comes to pulling out algorithms, no one can help me. Vaughn, even in C++ nobody can interpret methods of CatBoost quantisation even for reasonable money with sources.
By the way, the problem of a private researcher is absence of systematic analysis of obtained results - I judge by myself.
Why nonsense, during interventions maybe one algorithm works better and during sales another one). At 5 minutes.
And maybe there is no algorithm that works equally well in both situations.That's my reasoning.
The validity of the same approach may vary depending on the time, instrument, timeframe, etc. Therefore, it is important to be able to check the validity on a case-by-case basis. Ambiguous methods do NOT provide the opportunity for such verification. It is always possible to take advantage of the uncertainty of the method and deceive yourself and others with the presence of a good option, which is selected after the trade is completed.
I think that there can be deception, but not systematic - we are talking about institutional participants, where the decision is made by more than one person. Of course, everything is backed up by statistics for the required period of time ;)
It's like scoring in banks to evaluate a borrower - use the MO, but describe the logic in words.
If there is no purpose to deceive yourself or others, you need only unambiguously algorithmic methods. You can test them and have an informed opinion about them. Specifics are not so important as the principal possibility of unambiguous formalisation and testing.
Unformalised approaches are a breeding ground for false gurus "who have learned the laws of the market".
So I agree that the system approach is better, it's just that it is used to study a hodgepodge of different approaches - hence the effect of random wandering, I think....
In my opinion, it's not the best time for such research)
In any case, all Central Banks in one way or another copy the Fed, whose activities are more stable and better documented.
Too bad. Because I don't understand their formulas - I thought you could help me understand them.
That's a shame. Cos I don't understand their formulae - I thought you could help me understand.
Well, if you want to discuss a specific document with very specific formulas, then post it. I think it would be of interest to everyone.
Probably depends on the curriculum, here is the first institute that offers such knowledge in 81 hours.
And here are the questions for the exam for an applicant for a qualified investor (FFMS 1.0) from the Central Bank of the Russian Federation Topic 8.1. Technical analysis (p. 135).
Example of the question in the figure
Well, if you want to discuss a specific document with quite specific formulas, then post it. I think it would be of interest to everyone.
Well, let's say here are two formulas - at least I do not understand, here the window increases as the year progresses?
And in general, it would be better to understand how to calculate correctly by points.
It's called a learning programme from.... Then there's guessing like this)
Can you justify it? Seemed pretty good to me for a general understanding of the issue.
Well, let's say here are the two formulas - at the very least I don't understand, here the window increases as the year progresses?
And in general, it would be better to understand how to calculate the correct points here.
Whole document
Yes, for FNERT the window is incremented and equal to the month number, it seems obvious. RNERM is counted immediately for a month.
It seems to be simple, the P sign means the product.