[R-SIG-Finance] frontierMarkowitz problem: Tangential POrtfolio

Brian G. Peterson brian at braverock.com
Sat May 19 13:08:28 CEST 2007


Christian Grupp wrote:
> Hi,
> 
> I am playing around with frontierMarkowitz
> since I am interested in the tangetial portfolio.
> Unfortenately frontierMarkowitz does not work properly with my own data.
> i scan them with read.table,
> make them a time series with ts.
> 
> The result looks fine, assetsStats shows results, so i think that my 
> data are in the right format...
> 
> t.weights
> [1] NA
> 
> Rm
> [1] NA
> 
> Sm
> [1] NA
> 
> All other output looks fine so far.
> 
> Does anyone has an idea, where the problem is?

You haven't really given us enough information to give you any feedback.

If your constraints are too restrictive, it is quite possible to fail to 
generate *any* portfolio that will meet your constraints.

Have you tried using portfolio.optim or even solve.QP directly?

Can you produce a portfolio using fewer assets?

Have you used debug() in R to examine the intermediary results inside 
the function?

etc...

mean-variance optimization is very sensitive to inputs and constraints. 
  Without information on those inputs and constraints, this list can't 
give you anything more than general guidance.

Regards,

   - Brian



More information about the R-SIG-Finance mailing list