Econometrics: let's discuss the CU balance sheet. - page 17

 
Demi:

how the stationarity of a series with a single realisation is determined - a sufficiently long series is chopped into pieces, the MO is determined and it is compared. For a stationary series it should not change by more than 3 - 5 %.

Let's put out the "scatter". can we have a link to your definition of stationarity. I have not come across one. I use exactly one, different from yours. The mathematician here once proclaimed more variation, but your definition is just news, so link please.
 
Avals:

(back to square one)) What is quality trading? Obviously a high level of return/risk. The risk is actually the variance of those residuals. So if the variance is infinite/undetermined like Cauchy's, how can income/risk satisfy?
Risk is by no means the variance of the residuals.
 
faa1947:

Let's put the "discrepancy" out of the way. can we have a link to your definition of stationarity. I haven't come across one. I use exactly one, different from yours. The mathematician here once declared more variation, but your definition is just news, so a link please.


This is an applied definition of stationarity, because the same MO on the whole sample or all realizations is an abstraction that happens very rarely in life.

Well, have a look at the article - it' s in the text:

"For stationary random processes, the mathematical expectation is a constant. For ergodic processes, both mathematical expectation and variance and autocorrelation function calculated for one realization will be the same for any other realization. So to verify ergodicity, it is sufficient to calculate the variance for three to five realizations of equal length and compare them with each other. If the difference between them is 3-5%, then the process is ergodic and the length of the realization is sufficient for the calculation of its characteristics. If the discrepancy is greater than 10%, then either the process is non-stationary or too short realizations are used" (C)

 
Demi:


This is an applied definition of stationarity, because the same MO for the whole sample or all realisations is an abstraction which happens very rarely in life.

Well, have a look at the article - it' s in the text:

"For stationary random processes, the mathematical expectation is a constant. For ergodic processes, both mathematical expectation and variance and autocorrelation function calculated for one realization will be the same for any other realization. So to verify ergodicity, it is sufficient to calculate the variance for three to five realizations of equal length and compare them with each other. If the difference between them is 3-5%, then the process is ergodic and the length of the realization is sufficient for the calculation of its characteristics. If the difference is more than 10%, then either the process is non-stationary, or too short realisations are used." (C)

The quote has "variance" and you don't. That is what all my questions to you are about. You don't have to divide the whole into two parts to use them separately. Throughout the above I have only used them together and only using them together makes sense in this thread.
 
faa1947:
The quote has "variance" and you don't. That is what all my questions to you are about. You don't have to divide the whole into two parts to use them separately. Throughout the above I have only used them together and only using them together makes sense in this thread.


I didn't quite get it - the discussion was about stationarity. Stationarity is the constancy of the MO.
 
Demi:


14

credit?

No credit.

In the picture of the automaton there is an analytical line with the formula y=a+bx. And the location of points on this line is predetermined by this formula.

The expectation is a characteristic of a random variable and has nothing to do with analytical, deterministic predetermined sets of points.

If we look at the straight line in this graph as a realization of NE, then we have to subtract the deterministic component, and the remainder will have mo and variance (dispersion). If this is done, then mo=0 and variance = 0, which confirms that we are dealing with a deterministic set of points.

Stationarity is a characteristic of random variables and has nothing to do with deterministic variables.

I use the definition of stationarity: mo=constant and variance=constant. Always both. You can google and refine this worker-peasant definition, but the meaning remains. Your definition doesn't exist at all.

 


Avals:

well back to square one)) What is quality trading? Obviously a high level of return/risk. Risk is actually the variance of those residuals. So if the variance is infinite/undetermined like Cauchy's, how can income/risk be satisfying?


avtomat:

risk -- it is by no means a variance of the residuals.

And besides, one has to understand that any distribution is characterised by its parameters, not its name ;) and therefore to say that since it is a Cauchy distribution, then drain the water is a misunderstanding of the essence of the phenomenon.... It is possible to drain water at any distribution, if its parameters prove to be draining, whether it is Cauchy or normal, or any other.

here's Cauchy -- with different parameters

 
avtomat:

Besides, it is necessary to understand that any distribution is characterized not by its name, but by its parameters ;) and therefore it is a misunderstanding of the essence of the phenomenon to say that if the distribution is Cauchy, then drain the water -.... It is possible to drain water at any distribution, if its parameters prove to be draining, be it Cauchy or normal, or any other.

Here's Cauchy -- with different parameters.


There you go again with the AFC.

Well what does Coshi have to do with the topic and cotiers in general. We have, here in the marketplace, the next NE value beyond the right edge of the graph is predetermined in some interval, i.e. there is both a mathematical expectation and a variance. Well, what a Cauchy. Also drew, thank God it's not density, otherwise people would be confused with normal....

 

This is how the structure of the universe was imagined in the Middle Ages.

.

At the insistence of the Church and the scholastics, observations of nature were replaced by the study of Aristotle's works. The following case is typical: a monk, having seen sunspots through a telescope, decided to show them to his ecclesiastical superior. But he refused to look, saying: "In vain, my son, I have read Aristotle's works from the beginning to the end many times and I can assure you that I have not found anything like that in him anywhere. Go and take it easy. Be assured that what you mistake for sunspots is only a defect in your glasses, or your eyes."

Thus, in isolation from life, from nature, was the study of the world around us in the Middle Ages.

.

reminds you of anything? In modern parlance - with "nobels" ....

 
avtomat:

This is how the structure of the universe was imagined in the Middle Ages.

.

At the insistence of the Church and the scholastics, observations of nature were replaced by the study of Aristotle's works. The following case is typical: a monk, having seen sunspots through a telescope, decided to show them to his ecclesiastical superior. But he refused to look, saying: "In vain, my son, I have read Aristotle's works from the beginning to the end many times and I can assure you that I have not found anything like that in him anywhere. Go and take it easy. Be assured that what you mistake for sunspots is only a defect in your glasses, or your eyes."

Thus, in isolation from life, from nature, was the study of the world around us in the Middle Ages.

.

Ring any bells? In modern parlance, with "nobels" .....


I have given advice many times, from the heart. Look at the state-space models - I think they're all derived from TAU, but stripped down.

For the rest, just don't bother - ridiculous.

Reason: