Matstat Econometrics Matan - page 22

New comment
 
Aleksey Nikolayev:

Just a quick look at how the parameters (coefficients and variance of the residuals) of this very linear relationship change over time. Probably, we can only talk about the fact of sliding if the correlation and variance are approximately constant, and the shift fluctuates smoothly around some mean value. Accordingly, one can try to use the parameters of this fluctuation to construct a TC)

This is all true. The question is what exactly to take as a slippage between the two rows. For example, there is the traditional view that the length of the perpendicular to the regression line. But it seems to me that this is not the right way to do it. For it gives a spread not in relation to the previous values, but in relation to their midpoint. It loses such a substance as "asymmetry" of the sliding, which is what I would like to feel.
 
secret:
This is all true. The question is what exactly to take as the slippage between the two series. For example, there is the traditional view that the length of the perpendicular to the regression line. But I don't think that's quite the right way to do it. For it gives a spread not in relation to the previous values, but in relation to their midpoint. It loses such a substance as "asymmetry" of the sliding, which is what I would like to feel.

I don't even know if you can think of a perpendicular as a vector of two components) They are of course proportional to length, but with different coefficients.

But I guess I just didn't get the point. Maybe it's about keeping track of a possible violation of the linear relationship condition (model decoupling) at all times? If there is always a certainty that the relationship is preserved and unchanged, then any measure of discontinuity should (ideally) be expressed in terms of perpendicular length and regression coefficients.

 
Алексей Тарабанов:
I wonder what happens if the ei errors are white noise with Alexei Nikolaev's distribution.
Oh, Sanya has turned up, where has the poor guy been?
 
secret:

Therefore, it is necessary to study the regression residuals structure. In fact, this is what half of econometrics is about)

 
Who took the messages?
 
Maxim Dmitrievsky:
For quite objective reasons. A stationary portfolio only works in the moment, on new data everything breaks down without the right skills.
However, the fact that it does break down is 100%.
Isn't that a pattern? .....
 
Wrong toolkit.
 
MetaQuotes, have you even noticed that your forum has lost messages?
 

With MS and TV, you have to be careful sometimes. It can sometimes show a pattern where there is none at all.

 
CHINGIZ MUSTAFAEV:

With MS and TV, you have to be careful sometimes. It can sometimes show a pattern where there is no pattern at all.

Don't worry about MS and TV - the effect of false correlation has long been studied, there are appropriate tests and validation algorithms.

1...151617181920212223242526272829...38
New comment