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
How do you think this formula is derived? It's the old-fashioned way. I didn't know it before either.
I wonder how you use a pen and paper?
How do you think this formula is derived? Exactly by this "grandfather's" method. I didn't know it before either.
I wonder how you use a pen and paper?
If you mean what they gave me to calculate the RMS, the established practice in this thread does not allow me to post the result here. I will answer the letter :). The address was in principle present in some of my codes on this forum.
It seems to me that the sect (I like your definition) has gathered a very smart team. We understand each other at the level of ideas. And everyone knows how to write in MQL. Is it worthwhile, given these circumstances, to post the code here?
Vladislav suggested we avoid any actions that would be useful for freeloaders. And we all agreed.
Let's, if you have a strong desire to share your work, do it in a targeted way by email.
One more thing. Something I don't understand is this: СКО2/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5
As I see it, in your notations, RMS2/3[N]=(D[2N/3])^0.5
Or, if you try to represent it as a difference:
СКО2/3[N]=({S[N]-S[последняя треть]}/{2N/3})^0.5
No doubt about it. Except for error variance D(E) and error spread R(E).
That's just the way I am. I'm glad there are more people who also use pen and paper.
And I am doubly pleased that you also support the established practice in this thread.
And I am doubly pleased that you also support the established practice in this thread.
Likewise :)
Here is a picture of one of the variants of probability summation. It also shows Murray levels, the latter are drawn in 1 due to the lack of buffers.
http://kursovye-diplomy.narod.ru/ERO_CKO.rar
I think I found the reason why I can't upload a picture to MQL4.com, it seems to be a browser bug (Opera9). I use Explorer to check if the text and the empty file are ok, but when I loading the file, it took 60 sec and the message "BOLT you young man" appears and I got a message that time for one operation should not exceed 60 seconds.
1.As far as I understood, one of the criteria for channel selection is the smallest variance of regression error, which does not seem quite correct to me. That is, in my opinion, channels should be compared, for example, by determination coefficient or number of distinguishable gradations of response, and to take the channel with greater coefficient.
2.Even if the variance of the regression error is taken as the basis, how is the smallest one calculated? As far as I understand, as the error variance is a random value, you can use the chi-squared confidence intervals to determine the class, the group of the smallest ones which will be statistically indistinguishable from each other. And how can we select from this group what we need?
3.Again, the question about the 2\3 bracket is about the accuracy of the 2\3 number. Why not say 5\8 or some other number. How significant would be the deviations from this number. I remember that Vladislav talked about approximation of 2\3 sample. Maybe he has some criteria for choosing accuracy?
4.Since we have to compare sko2\3 and sko of the regression sample, and these are again random variables, we again have to deal with borderline cases when we cannot say definitely sko2\3 is more less or equal to sko. What to do with this group of channels?
5. Once again the question I asked. We can talk about adequacy of the regression model built by ANC (according to Bulashev) when the following points are answered:
The distribution of the regression error is close to normal
The expectation of the regression error is close to 0
The variance of the error is constant
The errors are independent, autocorrelation is close to 0
Since, although I may be wrong, I noticed that no one checks the channels for these conditions, I would like to know why, under what assumptions, or just to reduce the number of calculations?
Thanks in advance for the answers
Respectfully.
The choice of channel selection criteria is your own creativity. In general, any strategy is based on the model and logic. Vladislav has shared the model. I have left the logic to each person to come up with his own. And criteria - the basic element in making decisions with logic. Create.
Vladislav takes the worst version of this class.
The choice of accuracy of the bracket is determined by the statistical accuracy of its definition. You yourself said that it is a random variable.
Does it really matter? No matter how many channels you get, you can only use one (I mean from a class of close ones). And that one is borderline, i.e. the worst of the acceptable ones. Since the decision to be made is probabilistic anyway, an error of an enumeration does not affect anything. It is like a borderline between good and evil - everyone agrees that they are different poles, but everyone draws the borderline himself. :-)
Since, although I may be mistaken, I noticed that nobody checks channels for these conditions, I would like to know why, on what assumptions, or just to reduce the number of calculations?
If you are interested in it as a scientist, make a research and define whether these conditions are fulfilled or not. However, I think that this attempt will fail already at the stage of determining the nature of the error distribution. The market will not allow you to enjoy the law of large numbers. Channels emerge and collapse as soon as most have realised that a trend has emerged.
If you are interested in a working model, then take all this as an axiom, implement this model programmatically and the market itself will show you whether your set of axioms is fair or not.