Do you know how to make canals? - page 12
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
A channel is good in that you can see the direction of the market, but the boundaries of the channel may not match expectations.
But it all hinges on the principle by which the channel is constructed.
Maybe Alexey will show us something interesting?
My friend is awake, she is hilarious from the comments)))) Friends, write more)))
A friend woke up, laughing at the comments)))) (Dudes, write some more)))
What is there to write about, Alexey? I have reread most of the topics on the forum - people are in a complete anabiosis. As soon as a thought pops up, the greatest of men like SanSanych, Yusuf, and Automat pounce on it, totally dumbfounded in their ignorance, and nipping a man in the bud. These grown up kids should, for order, be reprimanded and permanently blacklisted. That would be the right thing to do.
Happy holidays to you too and coming out of anabiosis and schizophrenia
Happy holidays to you too and coming out of anabiosis and schizophrenia
Thanks, bro.
I have seen off a girl-friend, a Yakut, very beautiful, she gave me a broom in summer. Said a shaman knitted it for good luck )))))))))))
I said, 'Well, if you want, you can sleep on it.)
Ladies and Guys, I have learned so much about Yakutia)))
We know from statistics that stationary processes can be predicted, but non-stationary ones are extremely poorly predicted. This is exactly the problem. Non-stationarity has rendered useless mountains of mathematics extremely effective in other fields.
GARCH ideology:
Does your idea go this way?
There used to be an anecdote about mathematicians. Mathematicians have composed an algorithm for hammering a nail. The mathematician is asked, "How do you hammer a nail that's already half hammered?" The mathematician answers: "Pull it out and proceed according to the algorithm already worked out."
A rough outline of the path to follow. An unsteady process is reduced to a stationary one by repeatedly (sometimes once) differentiating it (taking the differences). Then the obtained series is forecasted and restored by integration, obtaining a forecast of the initial series. The exchange processes become non-stationary because of abrupt and unpredictable jumps in which it seems to me, even after multiple differentiation the aspect of heterogeneity will be pierced, which will also produce a large prediction error when approaching these points, which will increase during multiple integration leveling the usefulness of the forecast. That's how I see it in general terms, but it may not be true.
In any case, it seems to me that the solution to the problem of forecasting nonstationary series should follow the path of creating good models of these very series.
There used to be an anecdote about mathematicians. The mathematicians have an algorithm for hammering a nail. The mathematician is asked: "How do you hammer a nail that is already half hammered?" The mathematician answers: "Pull it out and proceed according to the algorithm already worked out."
A rough outline of the path to follow. An unsteady process is reduced to a stationary one by repeatedly (sometimes once) differentiating it (taking the differences). Then the obtained series is forecasted and restored by integration, obtaining a forecast of the initial series. The exchange processes become non-stationary because of abrupt and unpredictable jumps in which it seems to me, even after multiple differentiation the aspect of heterogeneity will be pierced, which will also produce a large prediction error when approaching these points, which will increase during multiple integration negating the usefulness of the forecast. That's how I see it in general terms, but maybe it's not true.
Anyway, it seems to me that the solution of the problem of non-stationary series forecasting should follow the way of creation of good models of these very series.
Alexey, read at your leisure
https://cran.r-project.org/web/packages/PSF/vignettes/PSF_vignette.html + attached file
There used to be an anecdote about mathematicians. The mathematicians have an algorithm for hammering a nail. The mathematician is asked: "How do you hammer a nail that's already half hammered?" The mathematician answers: "Pull it out and proceed according to the algorithm already worked out."
A rough outline of the path to follow. An unsteady process is reduced to a stationary one by repeatedly (sometimes once) differentiating it (taking the differences). Then the obtained series is forecasted and restored by integration, obtaining a forecast of the initial series. The exchange processes become non-stationary because of abrupt and unpredictable jumps in which it seems to me, even after multiple differentiation the aspect of heterogeneity will be pierced, which will also produce a large prediction error when approaching these points, which will increase during multiple integration leveling the usefulness of the forecast. That's how I see it in general terms, but it may not be true.
In any case, it seems to me that the solution to the problem of predicting non-stationary series should follow the path of creating good models of these very series.
You are absolutely right, but you have only described part of the path. There is a continuation that solves the disadvantages you mentioned, but there are new ones that have also been solved, and then there are new ones that have not been solved today. There are no 100% models for non-stationary processes today.
Let's not forget about pattern trading, as they call it in TA, and in mathematics it is classification. There are other ideas there, but also other difficulties, for which there is no complete solution today.
Judging by your profile you are quite capable of GARCH. Take R, it contains the rugarch package. Concentrate and in half a year you will get rid of many naive ideas and you will have the tool. You will be in the mainstream, you will receive many publications from reputable companies in reputable magazines. Moreover, maybe you will find a currency pair that you can predict with 95% confidence interval. But that is as luck would have it. But 75% is easy.
You are absolutely right, but you have only described part of the journey. There is a continuation that solves the shortcomings you have mentioned, but there are new ones that have also been solved, and then new ones that have not been solved today. There are no 100% models for non-stationary processes today.
Let's not forget about trading by patterns, as they call it in TA, and in mathematics it is classification. There are other ideas there, but also other difficulties, for which there is no complete solution today.
Judging by your profile you are quite capable of GARCH. Take R, it contains the rugarch package. Concentrate and in half a year many naive ideas will disappear and you will have the tool. You will be in the mainstream, you will receive many publications from reputable companies in reputable magazines. Moreover, maybe you will find a currency pair that you can predict with 95% confidence interval. But that is as luck would have it. But with 75%, it's easy.
You have already moved on GARCH. Maybe you may write an article on mql5 for us, I think many traders may be interested in it. Like: 1) Introduction - basic principles; 2) Stages of development; 3) Modern developments (there is enough of review + literature). I'm not yet friends with R. MATLAB is my love.
About the article - not ready yet, it turned out to be a lot of black work.
I used to be familiar with matlab. Toolbox "Econometrics". Paid, alien to financial markets, alien classification of instruments. Not even close to R for about three years.
Alexey, read at your leisure
https://cran.r-project.org/web/packages/PSF/vignettes/PSF_vignette.html + attached file