Many widely used portfolio allocation techniques (maximum Sharpe Ratio, minimum volatility, risk parity, etc.) rely on the assumption that correlations are stable over time.  This could not be farther from the truth.  The following graph plots historical trailing 252-day correlations between high yield bonds and the S&P 500 and between high yield bonds and fixed income.

correlation

A few observations from this graph:

  1. Correlations are unstable from an absolute perspective.  High yield’s correlation with fixed income ranges from -0.5 to 0.7 and its correlation with the S&P 500 ranges from -0.1 to 0.8.
  2. Correlations are unstable from a relative perspective.  Sometimes high yield behaves more like bonds and sometimes it behaves more like stocks.
  3. Correlation regimes can change with great velocity.  For example, in early April 2013 the trailing correlation between high yield and fixed income was approximately -0.4.  By August 2013, this figure was above 0.2, driven by highly correlated (0.6) returns during this time period.

Put more succinctly, getting correlation estimates right is hard.  This is crucial for investors and advisors because every assumption made in the portfolio construction process is a potential weak link that can lead to risk management failures when markets stop behaving nicely.

Going into 2007, it may have been tempting to replace some more traditional fixed income exposure with high yield.  After all, high yield had historically provided a nice income stream without having to take a lot of additional equity risk  (~0.1 correlation to S&P 500 from 1995 to 2006).  This trade didn’t work out well from 2007-09, when high yield’s correlation to equities skyrocketed.  For example, a 60/40 S&P/traditional fixed income portfolio lost 19% from January 2007 to March 2009 compared to a loss of 26% for a 60/20/20 S&P/traditional fixed income/high yield portfolio.  That 7% difference amounts to more than $125,000 today on an initial $1,000,000 account.

Pretend for a moment that you go to buy insurance for your home and your agent says that he has written a random number on your policy documents between one and ten.  He then asks you to pick your own number.  He writes your number next to his and closes the file without letting you see his number.  If you ever have a claim, the insurance company will only cover your losses if you happened to pick the right number.  Would you ever buy this type of policy?  Of course you wouldn’t.  Right?  Maybe you wouldn’t when it comes to your home.  But this is the exact type of risk management that is used in client portfolios where advisors and other fiduciaries rely exclusively on strategic allocations that are to a great extent driven by correlation estimates.

One of the reasons Newfound focuses on momentum-based strategies is that we can build momentum models that make very few assumptions.  This eliminates the need to estimate inherently difficult to estimate parameters and as a result can produce more robust investment solutions.  Pairing momentum-based, tactical strategies with independent investment approaches (i.e. value-based investing) along with a little bit of common sense (i.e. setting asset class bands based on client situation) can deliver a more repeatable portfolio construction process with a lot less room for error.

Justin is a Managing Director and Portfolio Manager at Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Justin is responsible for portfolio management, investment research, strategy development, and communication of the firm's views to clients.

Justin is a frequent speaker on industry panels and is a contributor to ETF Trends.

Prior to Newfound, Justin worked for J.P. Morgan and Deutsche Bank. At J.P. Morgan, he structured and syndicated ABS transactions while also managing risk on a proprietary ABS portfolio. At Deutsche Bank, Justin spent time on the event‐driven, high‐yield debt, and mortgage derivative trading desks.

Justin holds a Master of Science in Computational Finance and a Master of Business Administration from Carnegie Mellon University as a well as a BBA in Mathematics and Finance from the University of Notre Dame.