Rolling vrtest?

Question

I am trying to run the commands on the 'vrtest' package over rolling windows. I can run them perfectly on one set of data but I am having difficultly on using the rollapply function from the 'zoo' package. Say for instance my data is a dataframe x, where the variable is y and I want to roll over 10 observations at time. I have 30 observations so I should have 21 outputs for JR1, JR2 and JS1. So for instance, if I am trying to run a rolling joint wright test, the furthest I have got it;

JWroll <- rollapply(x$y, 10, function(x){
    x<- Joint.Wright(x$y, kvec)}

I know I need to define the 'kvec' somewhere within the brackets but I haven't got it to work yet. The data is;

10
10 
8
5
10
2 
8
2
10
2
8
8
10
10 
8
5
10
8 
8
12
10
7
8
7
8
1 
8
5
10
2 

Any advice or help with this would be greatly appreciated!

Answer 1

The function in rollapply must return a scalar or a plain vector, not a list. Try this:

# input data
kvec <- c(2, 4, 6)
x <- c(10, 10, 8, 5, 10, 2, 8, 2, 10, 2, 8, 8, 10, 10, 8, 5, 10, 8, 
8, 12, 10, 7, 8, 7, 8, 1, 8, 5, 10, 2)

library(zoo)
library(vrtest)

# the -1 removes the holding period from the output; unlist makes it a plain vector
rollapply(x, 10, function(x) unlist(Joint.Wright(x, kvec)[-1]) )

giving:

            JR1       JR2      JS1
 [1,] 1.5864093 1.5907354 3.042555
 [2,] 2.1701906 2.1396224 3.042555
 [3,] 2.5298221 2.4731367 3.042555
 [4,] 2.1701906 2.1405101 3.042555
 [5,] 1.1831651 1.1973584 3.042555
 [6,] 0.9238698 0.9238698 3.042555
 [7,] 1.0155461 1.0177885 3.042555
 [8,] 1.4546477 1.4897304 3.042555
 [9,] 1.2983318 1.2741745 3.042555
[10,] 1.1148843 1.1287839 3.042555
[11,] 1.2347644 1.1613745 3.042555
[12,] 1.0711498 1.0541897 3.042555
[13,] 1.1704861 1.1552450 3.042555
[14,] 1.1895470 1.1769137 3.042555
[15,] 0.9343208 0.9388534 3.042555
[16,] 0.9531326 0.9467291 3.042555
[17,] 0.8966971 0.8695012 3.042555
[18,] 0.7543456 0.7304734 3.042555
[19,] 0.8354976 0.7837317 3.042555
[20,] 1.0690090 0.9367724 3.042555
[21,] 2.5407608 2.3289587 3.042555