Dear group,
I have started working with the fPortfolio package and I am amazed by the
possibilities and the great work that has been done. However, I was
wondering, whether the timeSeries timestamps are used when it comes to
optimizing a portfolio. It seems to me, that the return and variance are not
annualized, but taken directly from the data (and have to be annualized
manually by the user). Is that the case? Does anybody have experience with
annualizing "efficient portfolios" (I don't want to mess up results in the
end after so much great work has been done...).
Just a short example of what I mean:
Taking the DIJ datasets from the fEcofin package:
library(fPortfolio)
library(fEcofin)
data <- as.timeSeries(DowJones30[,1:5])
data.ret <- 100* returns(data)
minrisk <- minvariancePortfolio(data.ret,portfolioSpec(),"short")
minrisk
>>>>result:
Portfolio Weights:
AA AXP T BA
0.2513 0.1786 0.2799 0.2902
Covariance Risk Budgets:
AA AXP T BA
0.2513 0.1786 0.2799 0.2902
Target Return and Risks:
mean mu Cov Sigma CVaR
0.0448 0.0448 1.2447 1.2447 2.8584
VaR
1.9407
Description:
Tue Jan 18 00:23:25 2011 by user:
>>>>end result
If we are just multiplying the weights with the covariance matrix, we do get
the following result for sigma:
weights <- c(0.2513, 0.1786, 0.2799, 0.2902)
covariance <- cov(data.ret)
sigma <- (weights %*% covariance %*% weights) ^(1/2)
We do get 1.2447 for sigma, which is exactly the value we got from our
minvariance function. However, as the dataset reveals, the series starts in
1990 and ends in 2001... The data has therefore not been normalized for a
specified time window - which means that we either have to carry out the
calculations manually or choose our "input data time frame" accordingly ( A
sigma of 1.2447 is of course indicating that the data is most likely not for
a one-year time horizon... :-) )
Thank you very much in advance!
Keep up the good work!
Lui
[[alternative HTML version deleted]]
_______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should
go.
