Re: a movjng average median

["Eric Zivot" <ezivot@u.washington.edu>]

Something like this is particularly easy using the S-PLUS command aggregateSeries. To illustrate, suppose you have a time series object ret.ts containing daily returns on an asset. To compute the median over a window of length say 24 days incremented by one day use

 

roll.median = aggregateSeries(ret.ts,moving=24,adj=1,FUN=median)

 

The arguments moving=24, adj=1 and FUN=median tell the splus function aggregateSeries to evaluate the median function on a moving window of length 24 and to adjust the output positions to the end of each window.

 

I discuss several extensions of calculations like this in chapter 9 of my book Modeling Financial Time Series with Splus. ez

-----Original Message-----
From: s-news-owner@lists.biostat.wustl.edu [mailto:s-news-owner@lists.biostat.wustl.edu]On Behalf Of Leeds, Mark
Sent: Monday, October 20, 2003 1:24 PM
To: s-news@wubios.wustl.edu
Subject: [S] a movjng average median

does anyone have or know of an easy way to calculate

the "windowLength" day moving median given a vector of numbers.

 

one can use filter to calculate the "windowLength" day moving average

as shown below but i can't think of a good way to

do the median.

 

"backwardsmovingaverage<-function(series,windowLength)

{

return(  c(filter( series, sides=1, filter=rep(1/windownlength,windowlength))))

}

 

thanks.

 

                                            mark