Create efficient frontier in PortfolioAnalytics without an xts object

Question

Is there a way to create an efficient frontier in the PortfolioAnalytics package without specifying an xts object of asset returns? Instead I'd like to supply the vector of expected returns and the covariance matrix.

Answer 1

There are two ways. First you can supply a list containing containing your matrices with the structure shown below and then call optimize.portfolio including this list as an argument.

# num_assets is the number of assets in the portfolio
  momentargs <- list()
  momentargs$mu <- matrix(0, nrow=num_assets, ncol=1 )
  momentargs$sigma <- matrix(0, nrow=num_assets, ncol=num_assets)
  momentargs$m3 <- matrix(0, nrow=num_assets, ncol=num_assets^2)
  momentargs$m4 <- matrix(0, nrow=num_assets, ncol=num_assets^3)

  optimize.portfolio(R, portfolio, momentargs=momentargs, ...)

Alternatively, you can supply your own function to calculate the moments. A simple example reproducing some of the PortfolioAnalytics options is shown below.

 set.portfolio.moments.user=function(R, portfolio, user_moments=NULL, user_method=c(returns, input, two_moment)) {
 #  
 #  Sets portfolio moments to user specified values
 #
 #  R               asset returns as in PortfoloAnalytics
 #  portfolio       a portfolio object as in PortfolioAnalytics
 #  user_moments    a list of four matices containing user-specified
 #                  values for the first four return moments
 #  user_method     user-specified method for computing moment matrices
 #                  defaults to calculation used by PortfolioAnalytics "sample" method
 #                  which uses PerformanceAnalytics functions to computer the higher-order moments

 if( !hasArg(user_method) | is.null(user_method)) user_method <- "returns" 
  tmpR <- R
  switch( user_method,  returns = { 
     momentargs <- list()
     momentargs$mu  <-  matrix(as.vector(apply(tmpR,2, "mean")), ncol = 1)
     momentargs$sigma  <-  cov(tmpR)
     momentargs$m3  <-  PerformanceAnalytics:::M3.MM(tmpR)
     momentargs$m4  <-  PerformanceAnalytics:::M4.MM(tmpR)
  }, input = {
     momentargs <- user_moments
  }, two_moment = {
     momentargs <- list()
     momentargs$mu <- matrix(as.vector(apply(tmpR,2, "mean")), ncol = 1)
     momentargs$sigma <- cov(tmpR)
     momentargs$m3 <- matrix(0, nrow=ncol(R), ncol=ncol(R)^2)
     momentargs$m4 <- matrix(0, nrow=ncol(R), ncol=ncol(R)^3)
   } )

   return(momentargs)
 }

You would then call PortfolioAnalytics with

 optimize.portfolio(R, portfolio, momentFUN = "set.portfolio.moments.user", ...)

as an example.