Suppose, I have fitted my data to a multivariate DCC model and do the forecasting to get mean returns. Below is my reproducible code.
# load libraries
library(rugarch)
library(rmgarch)
library(FinTS)
library(tseries)
library(fPortfolio)
data(dji30retw)
for (i in 1:2)
{ Dat.Initial = dji30retw[, 1:8, drop = FALSE]
Dat <- Dat.Initial[1:(1000+(i-1)), ]
#Fitting the data
uspec = ugarchspec(mean.model = list(armaOrder = c(0,0)), variance.model = list(garchOrder = c(1,1), model = "sGARCH"), distribution.model = "norm")
spec1 = dccspec(uspec = multispec( replicate(8, uspec)), dccOrder = c(1,1), distribution = "mvnorm")
fit1 <- list()
fit1[[i]] = dccfit(spec1, data = Dat, out.sample = 120, fit.control = list(eval.se=T))
#Out of sample forecasting
dcc.focast <- list()
dcc.focast[[i]]=dccforecast(fit1[[i]], n.ahead = 1, n.roll = 0)
#print(dcc.focast[[i]])
#Mean returns
mean.focast <- list()
mean.focast[[i]] = fitted(dcc.focast[[i]] )
print(mean.focast[[i]])
#Var-Cov
covmat.focast[[i]]= rcov(dcc.focast[[i]])
print(covmat.focast[[i]])
}
R has given me the output in this form:-
, , 2004-01-30
AA AXP BA BAC C CAT CVX DD
T+1 0.002903173 0.003462776 0.00295735 0.003485212 0.004442807 0.002986181 0.002069151 0.002367464
, , 2004-02-06
AA AXP BA BAC C CAT CVX DD
T+1 0.002933206 0.003511235 0.00300817 0.003510005 0.004436862 0.0029913 0.00205337 0.002382313
Is there any way we can save the output as one object/xts so it will look something like this?
AA AXP .......... DD
30-01-2004 0.002903173 0.003462776 ........... 0.002367464
06-02-2004 0.002933206 0.003511235 ........... 0.002382313
And another question, why is it if I type only 'mean.focast' , it gives me the following output?
[[1]]
NULL
[[2]]
, , 2004-02-06
AA AXP BA BAC C CAT CVX DD
T+1 0.002933206 0.003511235 0.00300817 0.003510005 0.004436862 0.0029913 0.00205337 0.002382313
Where is the value for [[1]]?
Let's see. Your "troubles" are from this part of the code
#Mean returns
mean.focast <- list()
mean.focast[[i]] = fitted(dcc.focast[[i]] )
print(mean.focast[[i]])
You ask to print the output for each iteration (which is why you get the output you get), but note that you reinstantiate the list mean.focast in each step of the iteration. That is why you get NULL for the first list element, because you've made a new list and are only filling in element 2 (for iteration 2). The mean.focast <- list() should be moved outside the loop.
As for the mean.focast result. You can combine the list elements into a single data frame in tons of ways. Here's one using reshape2.
library(reshape2)
res <- dcast(melt(mean.focast), Var3 ~ Var2)
res
Var3 AA AXP BA BAC C
1 2004-01-30 0.002903173 0.003462776 0.00295735 0.003485212 0.004442807
2 2004-02-06 0.002933206 0.003511235 0.00300817 0.003510005 0.004436862
CAT CVX DD
1 0.002986181 0.002069151 0.002367464
2 0.002991300 0.002053370 0.002382313
So your final code becomes
mean.focast <- list()
for (i in 1:2)
{ Dat.Initial = dji30retw[, 1:8, drop = FALSE]
Dat <- Dat.Initial[1:(1000+(i-1)), ]
#Fitting the data
uspec = ugarchspec(mean.model = list(armaOrder = c(0,0)), variance.model = list(garchOrder = c(1,1), model = "sGARCH"), distribution.model = "norm")
spec1 = dccspec(uspec = multispec( replicate(8, uspec)), dccOrder = c(1,1), distribution = "mvnorm")
fit1 <- list()
fit1[[i]] = dccfit(spec1, data = Dat, out.sample = 120, fit.control = list(eval.se=T))
#Out of sample forecasting
dcc.focast <- list()
dcc.focast[[i]]=dccforecast(fit1[[i]], n.ahead = 1, n.roll = 0)
#print(dcc.focast[[i]])
#Mean returns
mean.focast[[i]] = fitted(dcc.focast[[i]] )
print(mean.focast[[i]])}
res <- dcast(melt(mean.focast), Var3 ~ Var2)
res