Skip to content

aggregate an xts by factors

3 messages · Aaron Goldenberg, Joshua Ulrich

Original
Aaron Goldenberg
# 
This seems like a simple issue but I cannot get it to work for some reason.
I have a time series object that has daily returns of several stocks that I
would like to aggregate by qualitative factor.  In my toy example, for each
day, I would like to have two entries, one for the sum of the returns of my
two computer stocks and another for my financial stock. Then I would like
to calculate the cumulative sum of the returns for each factor. What am I
not doing correctly?

library(quantmod)
getSymbols("AAPL", src="yahoo", from="2015-07-01")
AAPL <- dailyReturn(AAPL$AAPL.Adjusted)
AAPL <- cbind(data.frame(AAPL), sector="Computer")
AAPL <- as.xts(AAPL)
getSymbols("GOOG", src="yahoo", from="2015-07-01")
GOOG <- dailyReturn(GOOG$GOOG.Adjusted)
GOOG <- cbind(data.frame(GOOG), sector="Computer")
GOOG <- as.xts(GOOG)
getSymbols("GS", src="yahoo", from="2015-07-01")
GS <- dailyReturn(GS$GS.Adjusted)
GS <- cbind(data.frame(GS), sector="Financial")
GS <- as.xts(GS)
combined <- rbind(AAPL, GOOG, GS)
#combined <- period.sum(combined$daily.returns, endpoints(combined,
on="days"))
combined <- aggregate(combined, by=combined$sector, sum)
Aaron Goldenberg
# 
Alternatively, I could have two separate time series, one for each factor.

On Mon, Aug 3, 2015 at 7:48 PM, Aaron Goldenberg <aaron at quantrisktrading.com

  
  


  


      

    

    

      
      

        


  

        

      Joshua Ulrich
    

    

      
      #
    

  


  

      
On Mon, Aug 3, 2015 at 6:51 PM, Aaron Goldenberg

        
<aaron at quantrisktrading.com> wrote:

            

      
      
      
    

        
The problem is that you're trying to mix types in the xts data.  xts
extends zoo, which is simply a matrix with an index attribute, and you
can't mix types in a matrix.

There are many different ways to do this.  Here's one way that should
be easily extensible to more sectors.

library(quantmod)
Computer <- new.env()
Financial <- new.env()
getSymbols("AAPL;GOOG", from="2015-07-01", env=Computer)
getSymbols("GS", from="2015-07-01", env=Financial)

sectorReturnSum <- function(env) {
  adj <- do.call(cbind, eapply(env, Ad))
  ret <- ROC(adj, type="discrete")
  xts(rowSums(ret, na.rm=TRUE), index(ret))
}
combined <- merge(Computer = sectorReturnSum(Computer),
  Financial = sectorReturnSum(Financial))