Regression equation - page 9
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
but you can do a multivariate polynomial regression... is it worse than linear regression? i don't know, there is only one check - if prediction accuracy increases or prediction time increases with the same accuracy, then yes, it's better... I don't know what's the point of checking it but you don't just have to understand how to do it but you have to explain everything to the tool ...
You can do any regression you want. What do you mean, worse or better?! Have you even read my bike? You have to understand what tool to apply to what! I wrote my bullshit being ignorant of the concepts of regression and close to it. I had to get clearly what I wanted. I did. Then I found out that it could be subsumed under the definition of multivariate linear regression.
And if you stick a polynomial, or any other, in there, I just don't get the point.
Trying to predict the behaviour of a financial instrument through any regression of other financial instruments is certainly the way to go. But it is so trivial... that I am sure it is a dead end. And implementing any multivariate regression is not a problem. You just have to understand why you're doing it, not just a desire to try something you haven't tried yet.
alsu:
it'll be better, but it'll load the computer, too:)
It won't load anything. Numerical methods rule.
P.S. Man, read your writing. Sounds so cool, all sorts of terms, etc.. Thought the same when I read the others without understanding the terms. And in fact they are discussing such elementary school level stuff... Just read a little about the basics of regression analysis, how the problem is set. And you will understand that the discussion is about the simplest ideas at a level above MA.
....
And if you put a polynomial, or any other polynomial, then I just don't get the point.
Trying to predict the behavior of a financial instrument through any regression of other financial instruments is certainly a way. But it is so trivial... that I'm sure it is a dead end. And implementing any multivariate regression is not a problem. You just have to understand why you're doing it, rather than just wanting to try something you haven't tried yet.
If you don't see the point (you don't understand what you're doing and why you're doing it), then everything will be stupid and pointless. You are absolutely right.
The absolute analogue of this action. take 2 machines, optimized class parameters... ...I optimised parameters for a demo, not better than a real one... and then I scratched my head, let's optimise three machines, OK, I optimised them, went to the real ... bam, alligator all the way... learned about RSI ... optimization... bam... neural network... bam...
and so on and so forth, because people don't use their heads, they act like robots...
Z.I. alsu can think. you can tell by the questions. he asks the right questions. If he gets the answer, well done, he's thinking with his head and not with his computer, maybe he'll get it right...
If you don't understand what you're doing and why you're doing it, everything will be dumb and pointless. You are absolutely right.
Not to a higher level, but literally at the level of a wrecker. I wrote here once that linear regression is the l.c. of two different most common dunks. People didn't immediately believe it. Here's the topic.
Don't insult regression analysis with the simplest cases.
On a related note, it is the methods of regression estimation other than OLS that got me thinking. I can only see the application of regression so far in the area of optimal portfolio construction. And in predicting the profitability of the optimal portfolio. But by no means BP of financial instruments.
A relatively optimal portfolio has already been openly proposed with all the details...
Yes, not a level above, but literally at the level of the wrecker. I wrote here once that linear regression is the l.c. of two different most common dummies. People didn't immediately believe it. Here's the topic.
Alexei, you know very well I've been there. In that thread. It's just that the man is asking the right questions.
So, we have a time series containing N samples. At this stage it is not important what exactly is meant by samples - ticks, OHLC or something else. The important thing seems to be the answer to the question about the optimal length of the training sample n which is not equal to N, the optimal number of adjustable parameters k<=n (degree of polynomial)
1. I answered that the polynomial degree is at most 3. I don't use regression, this number is given by other considerations (stochastic diff. levels)
2. optimum number of adjustable parameters is zero
3. Optimal sample length I do not know, I can not yet calculate, it seems to me it depends on at least two variables, the time of day and ACF. but that's how ...
What is meant by 'optimal'?
I have only been able to apply this word to the construction of the best portfolio on the window.
Performed simulations on EURUSD-GBPUSD pairs
A linear regression equation was used.
You are welcome to interpret it or suggest your own variants of the experiment.
The file is in the attachment.
I have written an example of a multivariate linear regression. Algorithm of getting ready system of linear equations (you can solve it with Gauss) is given in function GetLinearMatrix:
The Mathcad file itself is also attached.