Or alternatively you could concatanate all Jan contracts, Feb contracts, etc... this tends to deal with jump issues and also caters for those securities which have seasonal tendencies e.g. agricultural, energies, etc. My 2c, R.
On 2/16/05, davidr@rhotrading.com <davidr@rhotrading.com> wrote:
Even if this were available, I wouldn't use it.
This method will give you jumps when the contracts roll
(for almost all products.)
I would roll my own, blending from the old to the new over some period to
smooth out the jumps. This results in more realistic volatilities and
correlations.
At least, that's what I've been doing for 20 years.
David Reiner
-----Original Message-----
From: Omar Lakkis [mailto:uofiowa@gmail.com]
Sent: Wednesday, February 16, 2005 9:06 AM
To: r-sig-finance@stat.math.ethz.ch
Subject: [R-sig-finance] nearby
In Splus there is a function called "nearby" (description below). Is
there an equivalent utility in R?
Futures Nearby Creation Function
DESCRIPTION:
Create a nearby series from a multivariate time series representing
multiple futures contracts.
USAGE:
nearby(x, rule, contracts)
REQUIRED ARGUMENTS:
<b>x </b>
multivariate time series containing futures prices for the contracts.
<b>rule</b>
rule giving the roll date (last date to use each contract) from one
futures contract to the next, relative to the beginning of the
contract month. This can be given as a time span or relative time
object, or a character string which can be coerced into a relative
time object.
<b>contracts</b>
the contract months for the columns of the time series, given as
character strings of the form "F95", or "F1995", where the first
letter is the month code and the remainder is the year, and where
two−digit years are converted to four−digit years using
options("time.century"). These should be in ascending order.
VALUE:
a single−column time series containing the specified nearby series.
DETAILS:
The series is generated with the following steps. First, the contract
months are converted to dates of the first of the month, e.g. Z98
becomes December 1, 1998. Then the roll rule is added to each date,
to calculate the last day each column of the data set should be used.
The time series positions and the end dates are then passed to the cut
function to determine which column is needed for each output date,
and these data values are subscripted from x and returned. NA will be
the result for positions that have no valid column data.
SEE ALSO:
timeRelative, timeSeries
EXAMPLES:
# make a 5−column time series
x <− timeSeries(data=matrix(1:750, ncol=5),
pos=timeDate("1/1/1995", format="%a %02m/%02d/%Y")+1:150)
# make a nearby series, assuming this are Feb−June 1995 contracts (ignori
# the fact that there should be NAs in the matrix), rolling on the 3rd
# Friday of the month before the contract month
nearby(x, timeRelative("−1mth −1day +3fri"),
paste(c("G", "H", "J", "K", "M"), 1995, sep=""))
_______________________________________________ R-sig-finance@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance _______________________________________________ R-sig-finance@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance