Econometrics: one step ahead forecast - page 129

 
Debugger:


Exactly.

Accordingly, each of these assets has its own frequency phase and amplitude. I stress there are two (2)

Currency is a division of one into the other. And you cannot simply predict the division of one into the other, there will almost always (66% of the time) be an error

There is no USD currency. It is a mathematical abstraction, dimensionless, it is an index that contains the impact of 6 currencies on this dimensionless value. The applicable model: EURUSD = f(dx), where dx is just the USD index
 
faa1947:

About HP. See.

(1) the question is not answered

(2) what is the correlation of first error lags?

 
Debugger:

Yes, only the indices should give a quote when one is divided by the other.


So to predict one currency pair you need to analyse two synthetic instruments consisting of several currency pairs each? And that would be easier and more accurate?
 
Farnsworth:

(1) the question has not been answered

Thought I'd answered. Clarify.

(2) What is the correlation of the first lags of the error?


 
faa1947:

Please see EURUSD increments by 1/DX increments

The result is even worse


no, no, no, no, no, no. The result is very correct!!!!!

You gave fake data before, now it's correct. Absolutely, the model will never work :o)))

My advice to you - look for a normal transformation that brings the quote to at least some stationarity. There is no other way. Well... Apart from dancing with tambourines and stuff

 
Debugger:

Yes, only the indices should give a quote when one is divided by the other.

There has been a thread about this. It's not that simple, especially if the index trades on its own
 
faa1947:

Thought I'd answered. Clarify.

(2) what is the correlation of the first lags of the error?


(1) I repeat, what is your problem statement for filtering?

(2) I see, some polynome, which will not work in a non-stationary environment, it will not work in a stationary one. Well, there is no such technology.

QUESTION:

is your HP redrawable???

 
faa1947:
There is no USD currency. It is a mathematical abstraction, dimensionless, it is an index that contains the influence of 6 currencies on this dimensionless value. The applicable model: EURUSD = f(dx), where dx is just the USD index


Right.

Tell that to those who are paid in USD and use it to pay for goods and services.

 
Farnsworth:

no, no, no, no, no, no. The result is very correct!!!!!

You were giving fake data before, now it's correct. Absolutely, the model will never work :o)))

My advice to you - look for a normal transformation that brings the quote to at least some stationarity. There is no other way. Well... other than dancing with tambourines and stuff.

What am I doing?

You suggest converting to stationary in the first step. I do it in the second step and I can do it in the nth step.

 
Debugger:
I did not say anything about easier.
If dividing one index by the other produces a quote, then the forecast should be flawless.


Where is this conclusion coming from? A perfect forecast is accurate to 1???? How did you come to this conclusion? Is there a model or is it your opinion?