The problem seems to be that the tangency portfolio that fPortfolio calculates is not correct. It cuts through the efficient frontier, much like described here http://r.789695.n4.nabble.com/fPortfolio-and-tangency-portfolio-for-two-assets-td4677700.html Is there a general problem with tangencyPortfolio() in fPortfolio? On Wed, May 14, 2014 at 1:36 PM, Albert Darenberg
<albert.darenberg at gmail.com> wrote:
I'm trying to recreate the Matlab 2013 result for the tagency portfolio in fPortfolio. However, the resulting weights are not the same. I use the standard parameters in both Matlab and fPortfolio. The Markowitz efficient frontier is correct in both software suites. To validate this, I compared the minimum variance portfolio and a few random portfolios with a given variance on the efficient frontier. In Matlab I use "[PortRisk, PortReturn, PortWts] = portopt(ExpReturn, ExpCovariance, 50)" to calculate the efficient frontier and "portalloc (PortRisk, PortReturn, PortWts, RisklessRate, BorrowRate, RiskAversion)" for the tangency portfolio. Is there a catch that I'm missing when recreating results with Matlab and fPortfolio?