Econometrics: why co-integration is needed - page 5

New comment
 

And how quickly do you determine that the difference between the values of the two functions is stationary - and god forbid - just random?

 
faa1947:
Very simple. We take two series and calculate the correlation using a formula. You always get a number and you never get no number. ....
In the absence of impregnated sleepers it will not be a tram (c)
 
Mathemat:
faa, OK, I'll figure it out myself.

there is nothing to understand: given that the correlation range is [-1;1], anything outside this range is a false correlation!!!!!

;)))))))))

 
tara:

And how quickly do you determine that the difference between the values of the two functions is stationary - and god forbid - just random?

One at a time. It's called the unit root test.
 
paukas:
In the absence of impregnated sleepers it will not be a tram (c)

I would say that correlations are usually false. You have to prove that they are not false. And so trams, sleepers, horses, people .....

 

Correlation is not a direct indicator of the relationship. At +-1 there is a functional linear relationship. At 0, there is no linear relationship. That's it.

 
2 faa. the difference between two highly correlated instruments can be non-stationary and autocorrelated. Example: AUDUSD NZDUSD D1. sorry for the typos... Friday...
 
alexeymosc:
2 faa. the difference between two highly correlated instruments can be non-stationary and autocorrelated. Example: AUDUSD NZDUSD D1.
The cointegration is stationary. I've opened a topic with the question: what do we need from it? So far I've found that if the quote is cointegrated with the profit, the TS can be trusted. What else?
 
tara:

Correlation is not a direct indicator of the relationship. At +-1 there is a functional linear relationship. At 0, there is no linear relationship. That's it.

No, that's not it. I'm not going to teach you.
 

San Sanych!

Can't be trusted, but can be trusted with that probability :)

123456789101112...28
New comment