[R-SIG-Finance] semi variance

[Brian G. Peterson]

On Thursday 28 December 2006 09:36, BBands wrote:
> On 12/27/06, Brian G. Peterson <brian at braverock.com> wrote:
> > We implemented semideviation and downside deviation here:
> >
> > https://stat.ethz.ch/pipermail/r-sig-finance/2006q4/001170.html
>
> Yes, I saw that package. Very useful. Thank you.
>
> > But perhaps you could be a bit more specific
> > about what you're looking for?
>
> OK. I first tackled this general area 10 years ago when I created
> www.EquityTrader.com, which presents positive and negative alphas and
> betas, a very popular feature. While doing some recent portfolio work
> I revisited my "semi" alpha and betas and considered some related
> ideas. The "semis" seem to have become popular, so I thought I'd ask
> this august group if anybody was doing anything "interesting" with
> them.

Kris an I both discussed use of semi-variance in optimization in other 
replies to your query.  I've found these approaches to be somewhat useful 
as part of the portfolio optimization problem.

In general, I've gotten more "signal" from higher moments (skew, kurtosis) 
than from the broader area of lower partial moments.  That said, depth 
and length of drawdowns (and upside runs) does seem to have some 
information content for some instruments.

Regards,

   - Brian