A great book on testing and optimisation - page 18
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
To be honest - I don't really give a shit. I didn't want to get involved. But it just came to me. No Real? - No. So write whatever you want. I don't care.....))))
I haven't been on the forum for 5 days, and everything's the same here (my advice is don't visit the forum for 3-4 days, the forum will remain the same, but your attitude will change).
FOXXXi, now it's your turn. LeoV (thanks, Lenya!) has removed this letter.
What a queue, you're all crazy here. People!!! I asked you for something here, maybe in the privacy of someone, call me back, or maybe someone gave me money, and there was a secret correspondence, call me back!!!
Gee, it turns out there's talk of cointegration.
Two points:
1. The "physical", substantive part of the theory of the idea of cointegration is very shaky. Essentially, the situation is like finding a purely mathematical relationship between some digits. Often, as a result, it is bullshit, but it gives bread and butter to the analysts from banks and funds.
2) A limited set of objects of study. The limits of this set are not formalized. Hence, false stationarity.
Actually, the theoretical basis is not so shaky. If the data is empirically derived for a stationary process, the theory can work + some error.
If you substitute economic indicators into the equations, that's obvious nonsense.
And then, the theory is very outdated, i.e. again, the basis was developed in pre-computer times + on it the American Nobel laureates put on nonsense in post-computer times.
And so, in general, the problem is trivial. The point is that the multiple linear regression formula is a typical one-layer perceptron. The final result boils down to finding a model with minimum RMS between the empirics and the model over the set of points. Accordingly, the typical problem is to find an extremum. Using genetic algorithm to find all the equation coefficients is like two fingers on the asphalt. And no Nobel Prize is needed, no strain of wiggling your brain, while the result will be ready in a few minutes on a usual PC. You don't even need to have an idea about all sorts of autoregressions and autocorrelations, as genetics will take into account all corrections when searching for an extremum.
Another thing is that modern economists do not need such a trivial approach. Because it is accessible to anyone for check and it will be impossible to cover up pseudo-scientific machinations by any awards of the inventor of dynamite.
To be honest - I don't really give a shit. I didn't want to get involved. But it just came to me. No Real? - No. So write whatever you want. I don't care.....))))
You see, Mathemat learned what cointegration is, something else useful others will learn. You understand my position, but you continue to be a fool. I was not mistaken about you, I felt it, but I will not write here who you really are.
In fact, the theoretical basis is not so shaky. If the data are empirically derived for a stationary process, then the theory can work + some margin of error.
:) on yes, if on a stationary process, then yes, it may work. and it is not surprising.
Let's not argue till hoarseness, in my opinion - there's no "physics" there, in yours - it kind of is.
If economic indicators are substituted into the equations, that's obvious nonsense.
Yep.
And then the theory is a little outdated, i.e. again the basis was developed in pre-computer times + American Nobel Prize winners have stuck on it with nonsense in post-computer time.
But in general, the problem is trivial. The point is that the multiple linear regression formula is a typical one-layer perceptron. The final result is reduced to finding a model on the set of points which has minimum RMS between the empiricism and the model. Correspondingly, it is a typical task of searching for an extremum. Using genetic algorithm to find all the equation coefficients is like two fingers on the asphalt. And no Nobel Prize is needed, no strain of wiggling your brain, while the result will be ready in a few minutes on a usual PC. One may not even have an idea about all sorts of autoregressions and autocorrelations, since genetics will take into account all corrections while searching for an extremum.
There is nothing to argue about.
I should just note that the more I look at the economics Nobel, the more I find features of the peace Nobel in it.
My position is one - no reals? - No. Just words and screens? - Yes. Then "take a walk, Vasya!" ....))))
My position is one - no reala? - No. Just words and screens? - Yes. Then "take a walk, Vasya!" ....))))
You'll soon be repeating that like a spell, you're screwed, get over it.
You'll be repeating that like a spell, you're screwed, get over it.
Yeah, I know, that's what everyone says when there's zero on the real thing.....))))
Yeah, I get it, everyone says that when there's zero on the real.....))))
Here's the link that he asked me for the software. Now apologize to me here, for that shit you posted here, heroes do not do that, I told you that everything will be cleared up.