I'm using the quantmod package in R to pull historical data from yahoo finance.
To use the package I create a list of tickers like so:
symbols <- c("MSFT", "ORCL", "AAPL", "FB")
To get the historical data I call the quantlib method getSymbols:
try(getSymbols(sym, src="yahoo"))
This creates variables in my environment called MSFT, ORCL, APPL and FB.
To calculate a correlation between MSFT and ORCL with closing prices I can use
cor(Cl(MSFT), Cl(ORCL))
To get:
ORCL.Close
MSFT.Close 0.6597159
How can I make this generic so that i can pull say 20 symbols and run the correlation between them?
I don't know how to refer to a variable given a string. ie I have the string "MFST", how do I refer to the variable MSFT ?
I've got the string "MFST" and "ORCL" how can I use that to write cor(Cl(MSFT), Cl(ORCL))
The easiest way is to store the data in an environment, and then loop over all the objects in the environment with eapply:
library(quantmod)
dataEnv <- new.env()
getSymbols("MSFT;ORCL;AAPL;FB", env=dataEnv)
# Extract the close column from all objects,
# then merge all close columns into one object
cl <- do.call(merge, eapply(dataEnv, Cl))
# note the warning in the Details section of ?cor
# regarding the use of "pairwise.complete.obs"
cor(cl, use="pairwise.complete.obs")
# AAPL.Close ORCL.Close MSFT.Close FB.Close
# AAPL.Close 1.0000000 0.6467283 0.2528689 -0.5997337
# ORCL.Close 0.6467283 1.0000000 0.6597159 0.8211939
# MSFT.Close 0.2528689 0.6597159 1.0000000 0.8979836
# FB.Close -0.5997337 0.8211939 0.8979836 1.0000000