How to use output of acf2ar() function (matrix) for linear prediction

[gogof26]

This is my code. I am using acf.mrw from the mrw(multiplicative random walk) package:

library(mrw)
require(graphics)

R = 149.49;
lambda = 0.334;
n = 60
h = 0:n


ACF <-acf.mrw( q=1 , h ,lambda ,R ,type="cov", sigma=0.9) # "covariance"
plot(h,ACF,type="l")
x <- acf2AR(ACF)

plot(1:length(x),x,type="l")


The acf2AR() function returns a matrix but I dont know how to proceed from there. Any suggestions?

I would really appreciate any kind of thought on this subject!

Thanks

 

[Dason]

This should be in the R subforum. I'll move it for you...

Also you haven't described what you're trying to do and why it returning a matrix is causing you problems.

 

[gogof26]

Thanks for moving the post!

I am trying to compute forecasts for the MRW (multiplicative random walk) model. My supervisor told me to use the acf.mrw function to compute
the auto-covariance function and then use acf2ar() in order to obtain the linear prediction filter.

I dont know how to proceed and interpret the output of acf2ar() as it is a matrix n x n(in my case 60x 60). There are two similar papers: A multifractal approach towards inference in finance(Lovsletten and Rypdal, 2012) and a related paper by Skung and Yu (2009). I cant figure out in what sense the acf2ar() output is related
to the linear prediciton filter for the model.