Econometrics: one step ahead forecast - page 134

New comment
 

What are you fighting about?

The wording? Well name the series = deterministic + stochastic components.

to Farnsworth:

help a comrade - the models have poor predictive power. I expressed my opinion - non-stationarity of the series + non-ergodicity of the series. Maybe you know how to bring the series to a stationary form?

 
Farnsworth:

uses MESA, which has nothing to do with a recognised econometric model

I didn't mean that. The spectrogram is calculated on their basis, but this is a trivial matter and many similar calculations can be done.

Do you think cyclicality is proved that way? "You see the obvious clustering."

How?

 
Demi:

About wording? Well call the series = deterministic + stochastic components.

He is against that formulation.

I made my point - non-stationarity of the series + non-ergodicity of the series.

What is ergodicity in your understanding?

 
faa1947:

About wording? Well call the series = deterministic + stochastic components.

He is against that formulation.

I expressed my opinion - non-stationarity of the series + non-ergodicity of the series.

What is ergodicity in your understanding?

This formulation is universal and classical.

I have already written about nonergodicity - too lazy to repeat it. And to hell with it, with non-ergodicity - non-stationarity must be defeated.

 
Demi:

This formulation is universal and classic.

I have already written about non-ergodicity - too lazy to repeat it. And to hell with it, with non-ergodicity - non-stationarity should be defeated.

In my model, I try to do this by separating out the deterministic component step by step. At the end I get a residual, which sometimes, by chance, is stationary. Usually this residual is not stationary. In this case I model it with GARCH. As a result, I model as much as I can. It's not certain that the residual after the GARCH model is stationary. But while the range of this residue is less than a pip and this is another sign to finish modelling.
 
faa1947:

I didn't mean it that way. The spectrogram is calculated on their base, but this is a trivial matter and many similar calculations can be made.

It is not calculated on 'their basis'. MESA is an assumption about some BP properties, a model of that series, and an evaluation of the spectrum by the entropy maximisation criterion. But you look at the spectrum obtained by the MESA method and draw some strange "econometric" conclusions. Thank you that at least you don't calculate the amount of goods on the spectrum.

How?

Not really, it's not trivial when it comes to rigorous proof, a lot of very specific statistics. But what do you need it for, you prove the absence of cycles within a day, and then prove their presence. I cannot keep up with your discoveries. :о)

 
Farnsworth:

It is not calculated on "their basis". MESA is an assumption about some properties of BP, a model of that series, and an entropy maximization criterion estimate of the spectrum. But you look at the spectrum obtained by the MESA method and draw some strange "econometric" conclusions. Thank you for not calculating the amount of goods from the spectrum.

Realistically, it's not trivial when it comes to rigorous proof, a lot of statistics are very specific. But why would you want to do that, you proved the absence of cycles within 24 hours and then proved their presence. I cannot keep up with your discoveries. :о)

You proved the absence of cycles within 24 hours and then proved their presence. I can't keep up with your findings.

Once again, look at the sample size: 80000 vs 236. On 80000 there are no trends, mo=constant and variance is most likely equal to constant. On 236 it is not, because the outliers of the spectrum.

 
faa1947:

You proved the absence of cycles within twenty-four hours and then proved their presence. I can't keep up with your findings.

Once again, look at the sample size: 80000 vs 236. On 80000 there are no trends, mo=constant and variance is most likely equal to constant. On 236 it is not, because the outliers of the spectrum.

faa, I've been messing around with MESA a lot, first with this really great program (a few years ago), then I implemented it in matcadec (one version of the calculation). In many ways I can guess what else you have to see. Is that clearer?
 
Demi:

... ...unsteadiness has to be conquered.

It's akin to the "make a dog out of a cat" problem.

It does not need to be defeated, because it exists as a given. You have to work with it as a given. This requires other methods.

The wonderful tool "shovel" is good for digging the ground, but practically unsuitable for efficient scooping of water...

 
avtomat:

it is akin to the "make a dog out of a cat" task.

It does not have to be defeated, because it is there as a given. You have to work with it as a given. This requires other methods.

The wonderful tool "shovel" is good for digging the ground, but practically unsuitable for efficient scooping of water...

Once again a clever man from a neighbouring power has appeared. You can't say anything of substance without saying something of substance.
1...127128129130131132133134135136137138139
New comment