Tuesday, March 10, 2015

Dynamically Weighting Uncorrelated Investment Methodologies

In our Trend Aggregation approach we believe in using a number of different, uncorrelated methodologies together.  Some methodologies, like counter trend, perform better in a choppy market. Others, like momentum, perform best in a trending market.  The million dollar question that comes up when you use different methodologies that work best in different markets is how to weight them.  We have experimented with all sorts of risk parity approaches but nothing seems to add any value over and above equal weighting.  We have always been fascinated by the idea that perhaps a timing mechanism could be developed that would dynamically weight different methodologies based on the probability that they would outperform.

Value and momentum are two facts that have been proven in a number of studies to beat the market over time.  They also tend to be uncorrelated with each other so a portfolio that is split 50/50 between value and momentum strategies has crushed the S&P 500 over time.  Could you improve on a 50/50 split by timing when it is best to be in value and when it is best to be in momentum?  I think you can, but the guys over at Alpha Architect just published a study showing one approach----using valuation spreads----doesn't add any value.


That doesn't mean that it can't be done, just that perhaps this is not the way to do it.  This is an area we will continue to do work on as I believe that finding a way to dynamically weight uncorrelated methodologies has the potential to add a ton of value.

No comments:

Post a Comment