aligning dates from different sources

I've downloaded some FRED data (ICSA) using quantmod and I'm trying to
align it with weekly GSPC data. I'm running into trouble because the ICSA
data seems to come out on Saturdays and weekly closes of GSPC are usually
on Friday. I tried a simple fix of subtracting 1 from the ICSA dates but I
still run into problems around holidays. ?For example, the market was
closed on April 6, 2012 so I can't align the data for that week.

Any suggestions how I get the dates lined up so I can do some further
analysis ? ?I know I could use a brute force method of doing a simple cbind
(combined <-cbind(sp,ICSA[,1])) ?instead of merge but wondering if there's
a better way that takes holidays into account.

require(quantmod)
getSymbols('ICSA', src='FRED')
getSymbols('^GSPC', from='1967-01-01', to=Sys.Date())
index(ICSA) <-index(ICSA)-1
sp <-to.weekly(GSPC)[,6]
names(sp) <-'GSPC'
combined <-merge(ICSA,sp)
tail(combined, 15)

indexFormat(ICSA) <- "%a"
head(ICSA)

index(ICSA) <- as.Date(index(ICSA))-1
head(ICSA)

indexFormat(ICSA) <- NULL  # use the default
head(ICSA)

On Sat, Jun 30, 2012 at 1:21 PM, Eric Thungstom
<eric.thungstrom at gmail.com> wrote:

I've downloaded some FRED data (ICSA) using quantmod and I'm trying to
align it with weekly GSPC data. I'm running into trouble because the ICSA
data seems to come out on Saturdays and weekly closes of GSPC are usually
on Friday. I tried a simple fix of subtracting 1 from the ICSA dates but I
still run into problems around holidays.  For example, the market was
closed on April 6, 2012 so I can't align the data for that week.

Any suggestions how I get the dates lined up so I can do some further
analysis ?  I know I could use a brute force method of doing a simple cbind
(combined <-cbind(sp,ICSA[,1]))  instead of merge but wondering if there's
a better way that takes holidays into account.

cbind.xts simply calls merge.xts, so I'm not sure why you would expect
different results.

You could merge the two raw series, use na.locf to fill in the missing
values, then use apply.weekly to get the last value for each week:

x <- merge(ICSA,Ad(GSPC))
x <- na.locf(x)
y <- apply.weekly(x, last)

require(quantmod)
getSymbols('ICSA', src='FRED')
getSymbols('^GSPC', from='1967-01-01', to=Sys.Date())
index(ICSA) <-index(ICSA)-1
sp <-to.weekly(GSPC)[,6]
names(sp) <-'GSPC'
combined <-merge(ICSA,sp)
tail(combined, 15)

<snip>

Best,
--
Joshua Ulrich  |  FOSS Trading: www.fosstrading.com