[R-SIG-Finance] portfolio theory in terms of partial covariance

[aschmid1]

Hopefully the new framework may be of interest to this community:
http://ssrn.com/abstract=2436478

Here is the abstract:

It is found that partial correlations between the major US equity ETFs 
conditioned on the state of economy (mimicked by S&P 500) differ 
dramatically from their Pearson’s correlations. The mean-variance 
portfolio theory is reformulated in terms of partial covariance. 
Performance of various two-asset portfolios formed by the US equity ETFs 
is analyzed. It is found that on average, the best long-only mean 
partial covariance portfolios outperformed the best long-only mean 
covariance portfolios in 2007-2013.

Comments will be greatly appreciated. Also I wonder if there is an 
R-based software package for portfolio management that allows for easy 
replacement of Pearson covariances/correlations with their partial 
values.

Thanks! Alec