[R-SIG-Finance] fPortfolio plotting of efficient frontiers

[Brian G. Peterson]

No worries.  Unfortunately now I can't really be of much assistance.

I suspect that you may have changed the portfolioSpec, and changing the option 
again doesn't change it back, but I don't really know without  code to examine.

I'd suggest using one of the standard data sets, like 'edhec' or anything from 
fEcoFin, and building a little test script.  The list software will allow you 
to attach PDF's of the output, and your test script will let Diethelm or others 
more knowledgeable in fPortfolio replicate the problem.

Regards,

     - Brian

Charles Ward wrote:
> I am sorry Brian - my query was unclear.
> I plotted the frontier on two different graphs. In both cases, it was 
> the same calculation - only the options in the frontierPlot were changed.
> I am not quite clear what the units are in the default plot of the 
> Kendall estimator but the x-axis shows units of 150, 200 and 250 with 
> the settings                            risk=c("Sig") 
> If I then repeat the function with the option set to         risk= 
> c("Cov")  
> I would have expected the horizontal scale to show the standard 
> deviation where the units would be in the region of 0.008 to 0.011. In 
> fact, the graph appears unchanged with the same units of 150, 200 and 250.
> So my question really was about the option    risk = ... in the 
> frontierPlot function.
> What does it do?
> Charles
> 
> 
> On Mon, Mar 1, 2010 at 11:11 AM, Brian G. Peterson <brian at braverock.com 
> <mailto:brian at braverock.com>> wrote:
> 
>     Charles Ward wrote:
> 
>         But if I calculate the frontier using the Kendall method (stored
>         in f9) and
>         plot two frontiers using frontierPlot,
>         they appear on the same horizontal risk-scale. Shouldn't the two
>         scales be
>         different?
> 
> 
>     Charles,
> 
>     The two efficient frontiers would not be comparable if plotted on
>     different axes on the same chart.
> 
>     The (typical) reason for plotting multiple efficient frontiers is to
>     see how the unconstrained (or long only) Markowitz mean-variance
>     frontier 'shrinks' with different constraints or risk measures.  By
>     definition, the area under these more-restricted efficient frontiers
>     will be smaller than the unconstrained or long-only case.
> 
>     A modification to this approach that we use often it to set the
>     'risk' axis to the risk measure you really care about, such as CVaR,
>     and then plot the unconstrained or long-only efficient frontier
>     on/against *that* axis.
> 
>     Regards,
> 
>        - Brian
> 
>     -- 
>     Brian G. Peterson
>     http://braverock.com/brian/
>     Ph: 773-459-4973
>     IM: bgpbraverock
> 
> 


-- 
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock