Diversification Through Correlation

 | Jan 08, 2014 | 10:00 AM EST
  • Comment
  • Print Print
  • Print
Stock quotes in this article:








As we start 2014, I am going to touch on a few portfolio construction topics over the next couple weeks. Most investors tend to focus on the individual stock ideas, or perhaps investment themes or macro issues. These are all important but do not cover what you must do to build a robust portfolio. I probably spend as much time shaping my portfolio characteristics as I do on each name, because the wrong characteristics can completely offset all the hard work you do in stock-picking.

I want to touch today on the age-old concept of diversification, and how investors don't always fully analyze what it means to be truly diversified. Obviously, owning multiple stocks eliminates so-called idiosyncratic risk. Most investors try to further diversify by owning names across multiple industries or sectors. Fair enough, but what really matters is how stocks behave in relation to each other. If two companies in completely different industries both move to a common economic factor, then not understanding that can completely defuse the diversification you intended to achieve.

Institutional managers have incredibly sophisticated tools to measure and react to common factors (BARRA being one famous and pioneering tool), but average investors must work smart with different tools. I measure the diversification of my portfolio by looking at the correlations between price movements in my stocks.

Although this does not explicitly explain why two names move in unison or not, I can look at the correlation and see whether my diversification effort has teeth. (At the extreme, if I own 40 stocks that all have a 1 correlation, they will move in unison and I might as well own just one stock!) By looking at the correlations amongst my holdings, I can selectively construct the portfolio to reduce the correlations and build better diversity.

The table below shows the current holdings in my long-only Growth at a Reasonable Price (GARP) strategy. The matrix measures the correlation of the price movement between each name over the past six months. Highlighted names are particularly valuable as diversifiers, as they have correlations below 50%. My average correlation is 70% -- not bad -- and 21% are below 50% with 41% below 75%. There is probably some room to improve here, but all in all it's not too bad.

Disciplined Growth Price Correlation

The especially uncorrelated names are Marathon Petroleum (MPC), MeadJohnson Nutrition (MJN), RedHat Software (RHT), and the Home Depot (HD). Off the top of my head I can't really explain why these names would be so universally uncorrelated with the rest of the portfolio, but that is the value of the exercise. I did not have to come up with a theory or test a lot of data. I can simply see that I am getting diversification value from those names. That is the value of a proxy!

As I mentioned the first time I wrote about this exercise, I am aware that many readers won't have the statistical tools to do this sort of analysis (I use FactSet). However, your financial advisor or stock broker most certainly does, and this would be a good exercise to ask them to do for you.

Columnist Conversations

volatility is quite low here, and we could see some downsides here in the short term. ...



News Breaks

Powered by


Except as otherwise indicated, quotes are delayed. Quotes delayed at least 20 minutes for all exchanges. Market Data provided by Interactive Data. Company fundamental data provided by Morningstar. Earnings and ratings provided by Zacks. Mutual fund data provided by Valueline. ETF data provided by Lipper. Powered and implemented by Interactive Data Managed Solutions.

TheStreet Ratings updates stock ratings daily. However, if no rating change occurs, the data on this page does not update. The data does update after 90 days if no rating change occurs within that time period.

IDC calculates the Market Cap for the basic symbol to include common shares only. Year-to-date mutual fund returns are calculated on a monthly basis by Value Line and posted mid-month.