Custom expected returns in the Portfolio Analytics package

Question

I have trouble incorporating custom expected returns in Portfolio Analytics package. Usually expected returns are some professional expectations / views or calculated separately from fundamental indicators. Portfolio Analytics allow to create custom moments function to calculate moments from past returns, but I don't understand how to incorporate already calculated returns to optimization problem. Any help is appreciated and here is small example dataset:

#Download package and sample returns
library(PortfolioAnalytics) 
library(PerformanceAnalytics)
data(edhec)
returns <- tail(edhec[,1:4], 10)

#Example expected return xts that I'm usually working with. Calculated separately.
N <- 10
M <- 4
views <- as.xts(data.frame(matrix(rnorm(N*M,mean=0,sd=0.05), N, M)), order.by = index(returns))
colnames(views) <- colnames(returns)

Lets create basic portfolio with some objectives.

pf <- portfolio.spec(assets = colnames(returns))
pf <- add.constraint(portfolio = pf, type = "full_investment")
pf <- add.constraint(portfolio = pf, type = "long_only")
pf <- add.objective(portfolio = pf, type = "return", name = "mean")
pf <- add.objective(portfolio = pf, type = "risk", name = "StdDev")

Now I would like to optimize portfolio pf at each period and take account views (expected returns for that period) but I'm running out of ideas at this point.

Answer 1

I realise now, after setting the bounty, that the questions has already been answered here. I'll summarise as best as I can understand it.

When you call optimize.portfolio, there is an optional parameter momentFUN, which defines the moments of your portfolio. One of its arguments is momentargs, which you can pass through in optimize.portfolio.

First, you need to choose a set of expected returns. I'll assume the last entry in your views time series:

my.expected.returns = views["2009-08-31"] 

You'll also need your own covariance matrix. I'll compute it from your returns:

my.covariance.matrix = cov(returns)

Finally, you'll need to define momentargs, which is a list consisting of mu (your expected returns), sigma (your covariance matrix), and third and fourth moments (which we'll set to zero):

num_assets = ncol(current.view)
momentargs = list()
momentargs$mu = my.expected.returns
momentargs$sigma = my.covariance.matrix
momentargs$m3 = matrix(0, nrow = num_assets, ncol = num_assets ^ 2)
momentargs$m4 = matrix(0, nrow = num_assets, ncol = num_assets ^ 3)

Now you're ready to optimize your portfolio:

o = optimize.portfolio(R = returns, portfolio = pf, momentargs = momentargs)