Liz Cheval and Negative Correlation

Discussions about Money Management and Risk Control.
Post Reply
CyTrend
Roundtable Fellow
Roundtable Fellow
Posts: 57
Joined: Mon Oct 25, 2010 5:28 pm

Liz Cheval and Negative Correlation

Post by CyTrend » Wed Apr 13, 2011 7:29 pm

Hey guys,
I came across this very interesting link from a presentation by Elizabeth Cheval (original turtle) talking about the value of CTAs as an alternative asset and most importantly giving a very intuitive explanation of linera correlation. However i have a question. Somewhere in PART II she shows an asset (Ugly Asset A) and says "it has no long-term return, although i must mention at this point it has a positive average monthly return, the positive monthly return is erased over time by the high volatility the effects of compounding and the timing of the big drawdown" but then she still goes on to prove that because it has a perfect negative correlation of -1 with the other asset in the portfolio it 's still an amazing addition. I am not sure i understand how the asset can have "no long-term return" but have a "positive average monthly return" (she specifies at the beginning that for an asset to improve the portfolio in a mean-variance sense it must have positive returns and negative correlation) If anyone has seen this presentation before or is interested to take a look please share your thoughts.
http://www.emccta.com/correlation/

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Wed Apr 13, 2011 8:05 pm

You may be interested in (this) old thread as well, particularly Figures 5 and 6 of the first post.

As correlation falls, Desirability (aka "Goodness") (aka "Gain to Pain ratios") rises. You get sizable benefits at correlation = +0.1 and even greater benefits at correlation = -0.2.

Although it would be very pleasant indeed, Liz's assumed correlation of -1.0 is not a prerequisite to happiness.

rgd
Roundtable Knight
Roundtable Knight
Posts: 396
Joined: Thu Aug 10, 2006 5:21 pm

Post by rgd » Wed Apr 13, 2011 10:49 pm

favorite quote from Winton presentation:

High sharpe strategies are hard to come by and have a limited shelf life.

Low sharpe strategies tend to be far more abundant.

Therefore, the goal is to find as many uncorrelated, low sharpe strategies as possible and blend them together.



Of course, all this is made more simple by having 100 researchers in lab coats at your beck and call.

Eventhorizon
Roundtable Knight
Roundtable Knight
Posts: 229
Joined: Thu Jul 08, 2010 2:36 pm
Location: Boulder, CO
Contact:

Post by Eventhorizon » Thu Apr 14, 2011 10:22 am

CyTrend:
I am not sure i understand how the asset can have "no long-term return" but have a "positive average monthly return"
It's the difference between an arithmetic average rate of return and geometric average rate of return: consider monthly returns of 1.000% and -0.995% alternating. The arithmetic average return is +0.0025% per period. The geometric average rate of return is -0.0025%.

Qualitatively it arises because you are making gains on smaller sums and losses on larger sums. It's a variation of: "you need 11.11% in gains to make up for 10% in losses". You can see this effect dramatically in index tracking etf's, especially 2x and 3x leveraged ones: annual returns do not appear to track because the compounding is done daily. The arithmetic average is the return you "expect" to see selecting a return period at random.

Hope this helps.

EDIT: Corrected grossly incorrect example arithmetic! Note to self: more COFFEE!

Post Reply