[R-SIG-Finance] Diversification Comments...

[BBands]

On 8/22/07, Marc E Levitt <marcl at svquant.com> wrote:
> Consider the question you are looking to answer before we dismiss diversification and how much diversification is useful. There are many papers that explore diversification in terms of terminal wealth standard deviation which is quite important for those in the asset-liability management world, e.g. pension funds, where a short-fall at any given instance is perhaps most important. Most conclusions in this area is that more diversification even with high correlation coefficients (of course not 1.0) between managers (instruments) pays off depending on implementation costs. I looked at this in terms of trend following CTAs and found that diversification continues to pay off when looking at terminal wealth vs time series standard deviations well beyond what the standard rules of thumb tell you.

Terminal wealth means nothing in the presence of large drawdowns.

See Ralph Vince's new book for some consideration of the above idea.

http://www.amazon.com/Handbook-Portfolio-Mathematics-Formulas-Allocation/dp/0471757683/

"Risk is the probability of being ruined."

"Ruin is touching or penetrating some lower barrier on your equity."

When correlations converge on one and that lower barrier is touched
long-term goals mean nothing as investing stops and liquidation
starts. Many examples of this in the last two weeks news.

Also of interest, the 23 August ISI Quantitative Research piece
entitled "Market Direction and Sector Correlation".

    jab
-- 
John Bollinger, CFA, CMT
www.BollingerBands.com

If you advance far enough, you arrive at the beginning.