Econometrics: one step ahead forecast - page 87
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
Yes.
Yes. For example, to make a prediction one hour ahead, the previous 22 hours are sufficient.
To the next one, Yusuf, to the next one.
What is there to write it for? Does EViews have HP but no Moving Average? And enlighten people as to what R^2 is there...
..... of history since the beginning of the year, is 36 hours, and for last year it was 15 hours, how do you explain this fact? ....
Я об остатке от индикатора.
I'm also talking about the residue - the difference between MA and kotira. What, colleague, there's no regular mash-up in EViews to replace your beloved HP?
I'm also talking about the residue - the difference between MA and kotira. What, colleague, there's no regular mash-up in EViews to replace your beloved HP?
I am interested in taking into account in the model, not in the result.
You may use it as a model, but in real life you should use it in the account, whatever way you want))
You analyse the system's returns. The system has a forecast - MO. In fact, you get pluses, minuses in each trade. The deviation of the results of transactions from the MO is the residue, or prediction error. You can take Sko for example.
There is no sense to analyze the indicator residues, because it does not forecast anything by itself.
Why, I have explained. I can do it again.
Let's take a quotient. While there is a trend in it, we cannot say anything about the statistics - the trend will overrun all statistics.
We select the trend by smoothing НР, take some bars НР (4) and add to it the difference between the kotir and smoothing.
We look at the residue of this regression. We can see that we again see the trend (ACF). Once again, like above, but for the residue of the first regression.
We look at the residue of the new, extended regression. We see that there is no ACF. Let's rejoice and look for ARCH - it is some properties of initial quotient that were not visible at the beginning. That's it.
Result: we have two smoothing + two different residues after smoothing. If we add them up we get the initial quote, not a pip loss. In addition we model ARCH which is not expressed in pips, but was the initial quotes.
I have written it many times. You can see in the formula.
Well, in tests you will take into account in the model, but in real life on the account, no matter how you turn))
Analyze the system returns. The system has a prediction - MO. You get pluses, minuses in each trade. The deviation of transaction results from the MO is the residue or forecast error. You can take Sko for example.
There is no sense to analyze the indicator residues, because it does not forecast anything by itself.
I've been following the thread for 87 pages now, and I still don't understand why faa1947 needs these conversions. But I think I understand Slava. Why the hell do you need to analyse indicator/system residuals? There is a system (in this case a linear trend), its m.o. is either greater than zero or less. If the value of M.O. is insignificant with respect to S.Q.O., then M.O. is not significant. Why bother with this and analyze the residues from the system? What does it give? I must have really skipped a lot of lectures at university.