Machine learning in trading: theory, models, practice and algo-trading - page 1502
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
This is for Asaulenko and other radio amateurs
))))
This Max is an objective reality
1) The market has a wave structure, right?
2) Waves have a height (amplitude), maybe 10 points and maybe 210 points.
3) The waves have a different width (period or frequency), one has 10 bars and the other 210, will the indicator with fixed parameters be useful to win back these waves?
4) Market waves are never repeated by their parameters, right?
So what kind of moron would I need to be to look for patterns in indicators with fixed parameters? Probably a round one, I have been like that for many years (?)
The solution: to teach the network to find the right parameters for the indicator at any moment of time using the "market parameters" - AFR
Only then we can look for patterns
Or who wants to argue about it?
Way out: you need to teach the network at each moment of time to find the right parameters to the indicator using "market parameters" - AFR
Hmmm, interesting, I know someone who just chose this way...
I wonder, did you do it yourself or did you read it somewhere?
2) a sum of hormones can describe a function of any complexity
Correction - it is possible to describe any complexity, BUT if it is periodic. I.e. it repeats after some time 1 in 1. But nothing repeats 1 in 1 in the market.
))))
This Max is an objective reality.
1) the market has a wave structure so? so!
A wave structure, but all known waves have a periodic structure or the addition of several harmonics, but no one was able to decompose the price using Fourier or wavelets on the history and then reproduce it with sufficient accuracy in the future... It means the market does not have a wave structure )))
there is also a story that wave analysis does not work on forex, it works exclusively on stocks, no on indices, no.... on history! ))))
There are many legends and interesting readings on the Internet, physicists even quote Gann, this has all become a religion, and religion, as you know, appears due to a lack of information - alas, people are like that )))
Hmmm, interesting, I know a man who just chose this path...
I wonder, did you come up with it yourself or did you learn it from somewhere else?
You could say 50/50.
Adaptive filters (filters that change their parameters depending on the characteristics of the signal) are used for decades, in the market I think we must also use similar filters, but conventional AF will not work because besides the characteristics of the signal you also need to take into account the pp levels. Here is a filter that takes all this into account I think will work, but I do not understand how to implement it(( so far
Correction - any complexity can be described, BUT as long as it is periodic. I.e. it repeats after some time 1 in 1. But nothing repeats 1 to 1 in the market.
Yes, I agree, you can PCA.
wave structure, but all known waves have a periodic structure or the addition of several harmonics, but no one has been able to achieve a decomposition of prices using Fourier or wavelets on history and then reproduce them with sufficient accuracy in the future... It means that the market does not have a wave structure )))
We do not need to predict waves and the future, we just need to use spectral analysis to get the objective characteristics of waves that are present in the market at the moment - do you understand the difference, Igorek?
After all, all your indicators are built according to the price, to these waves and the indicator does not really care whether you believe in these waves or not, you may measure the characteristics by a zigzag, it makes no difference and the main thing is that it works and is objective.
But if you need to separate the dominating waves from the noisy data and measure their characteristics you have to use spectral analysis, the whole world does it, I don't know why))) Maybe you know how to do it better?)
We do not need to predict waves and the future, we just need to use spectral analysis to take objective characteristics of waves that are present in the market at the moment - do you understand the difference, Igor?
After all, all your indicators are built according to the price, to these waves and the indicator does not really care whether you believe in these waves or not, you may measure the characteristics by a zigzag, it makes no difference and the main thing is that it works and is objective.
But if you need to isolate the dominant waves from the noisy data and measure their characteristics you have to use spectral analysis, the whole world does it, I do not know why))) Maybe you know how to do it better?)
search the forum, first thing I found:
https://www.mql5.com/ru/code/1276
https://www.mql5.com/ru/articles/185
i have already checked these codes, the article is very good, and the author is one of the few whose articles i remember
there are some good spectrum analyzers in english kodobase.
If i look at the statistics, i can't foresee the time of validity of this article, it will help to analyze the real time situation.
SZZY: there is one of few trading guru advices in the internet that really works, it sounds like this: having found your TS, do not look for others, and constantly improve your own - something in it there, capable indicator TS or examples that Maxim published in his articles, in principle enough to trade with a positive expectation, but the main thing to count the risks: Max Gunther "Axioms of the stock speculator": "Auxiliary axiom number 3. Decide in advance what profit you want to make on a trade, and once you have it, close the position immediately."
a forum search, the first thing I found:
https://www.mql5.com/ru/code/1276
https://www.mql5.com/ru/articles/185
Thanks, but I do not need it, I can do it myself, I did not even want to talk about spectral analysis in order to reduce the amount of information is not necessary.
I asked a specific question to which I did not know the answer, we discussed everything we could, but not the question(((.
a forum search, the first thing I found:
https://www.mql5.com/ru/code/1276
https://www.mql5.com/ru/articles/185
I have already checked these codes, the article is very good, and the author is one of the few whose articles I remember
there are some good spectrum analyzers in english kodobase, i used to find
If i look at the statistics, i can't foresee the time of validity of this article, it will help to analyze the real time situation.
SZZY: there is one of few trading guru advices in the internet that really works, it sounds like this: having found your TS, do not look for others, and constantly improve your own - something in it there, capable indicator TS or examples that Maxim published in his articles, in principle enough for trading with positive expectation, but the main thing to count the risks: Max Gunther "Axioms of the stock speculator": "Auxiliary axiom number 3. Decide in advance what profit you want to make on a trade, and once you have it, close the position immediately.
The output will be 8 amplitudes of different frequencies - which can be fed to the forest/nS. In this case, some information will be lost.
Or you can just feed the last 60 bars. In this case the information will not be lost.
I asked a specific question to which I do not know the answer, we have discussed everything for nothing, but not the question(((
Your question is not formalized and does not correspond at all to the information that the bars on the charthttps://www.mql5.com/ru/forum/86386/page1499#comment_12044235 can give
1. The period of bars in question is important, but to each his own, I consider a period equal from the beginning to the end of the day, from the beginning of the month to the end of the month, but not 24 bars or 31 bars (month), do you have a cutoff?
2. unfortunately we only have bars available, how do you draw the width, height... is this likely a graphical analysis?
3. the ideal deal depends on the time of holding the position in the market, if trading on the H1, then the ideal deal is from the high bar to low, but no more than 1 hour, imho, but if the M1, then such an ideal TS will lose on the spread, OK, so again ZigZag? ....
4. To be honest, MA does not characterize the market, draw any 2 lines and trade by touching or crossing them by price, The effect will be the same as from the MA, there is some illusion that the MA follows the price and describes something, but imho, MA simply follows the price, any algorithm that moves the 2 lines (which I mentioned) will be on profitability, and the MA, but at different points in time than the MA (ie, then the MA in profits, and not MA in profit.i.e. the MA is on the profit, the lines are on the profit)
as for your questions, there are no definite answers, although I may be wrong, you'll just have to wait for someone who knows how to answer your question
I took a look at the spectrometer code. The spectrum is built based on the last 60 bars.
The output will be 8 amplitudes of different frequencies - which can be fed to the forest/NS. In this case, some information will be lost.
Or you can just feed the last 60 bars. In this case the information will not be lost.
But you can just pick up standard RSI, Stochastic, in my opinion it's easier and the effect will be about the same, imho
I read the articlehttps://habr.com/ru/post/197606/ a long time ago, it's the same case why these spectra and Fourier transforms don't work, but I don't like mathematics at all lately, except that I'm used to doing school work in the evenings )))