Econometrics: one step ahead forecast - page 94
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
Don't get sidetracked.
Highlight the deterministic component. What about the residual? We check again for the deterministic component. The cause is old. Get to the noise. Get the noise without the deterministic component, we can reason.
For non-stationary series, this deterministic component will change a lot over time. Adaptive methods must be applied to predict
For non-stationary series this deterministic component will change a lot in course of time. For forecasting, adaptive methods should be applied
If you look at the topic first, you will see details of the following idea and practical calculations for it.
Idea:
1. fit the model to the sample
2. forecast 1 step ahead
3. once the fact arrives, fit again on a shifted sample
4 forecast 1 step ahead again.
There was a table to the right of the model parameters (lags) and you can see that the model parameters change almost every time.
If you look at the topic first, you will see details of the following idea and practical calculations for it.
Idea:
1. fit the model to the sample
2. forecast 1 step ahead
3. When the fact arrives, fitting again on a shifted sample.
4 again forecast 1 step ahead.
There was a table to the right of the model parameters (lags) and you can see that the model parameters change almost every time.
So that's what it is - unsteadiness, damn it to the ground)))))
Nothing good can come out of it. All this adaptation to pseudo-stationarity is just shamanism. This series will still be non-stationary no matter how you rape it. Very strongly non-stationary or strongly non-stationary or weakly or strongly non-stationary - there are no reliable forecasting methods. All mathematical statistics are developed on the basis of stationarity and ergodicity hypothesis.
So that's what it is - non-stationarity, damn it to the ground)))))
Nothing good can come of it. All this reduction to pseudo-stationarity is shamanism. This series will still be non-stationary, no matter how hard you force it. Very strongly non-stationary or strongly non-stationary or weakly or strongly non-stationary - there are no reliable forecasting methods. All mathematical statistics are based on the hypothesis of stationarity and ergodicity.
There are totals for the profit factor greater than 1, but that's little consolation, as there were worse.
Don't want to get discouraged. Maybe a bad smoothing, maybe the wrong regressors? Or a wrong interpretation of the forecast? We need to look into it. That's what I opened a branch for.
Mathemat:
I wonder what ergodicity has to do with it? Explain it for me, Demi: perhaps I'm used to understanding it in some other way.For ergodic processes, both the expectation, variance and autocorrelation function calculated from one realization will be the same for any other realization.
(Partial citation.)
If a series is non-stationary, therefore it is non ergodic.
And you still intend to stick with a model that is highly unsustainable in the main parameters?
Demi: Тогда и только тогда, когда стохастический процесс является эргодическим
Well, firstly, these are all general words which I am already familiar with.
Secondly, looking at the quotation process as a set of realisations was not the point. The realisation is one, full stop. At least in econometrics.
Third, and most importantly: how can it be checked, ergodicity, if we can't make other possible realizations? (If we could do it, we could make an ideal tester, which would be able to check any TC for robustness one hundred percent, as we would have, roughly speaking, as much input data as needed, i.e. infinity).
faa1947: I'm not sticking with the model, I'm sticking with the metod.
OK, the question is the same, but to the method.
OK, same question, but to the method.