The ‘No Direction’ Portfolio: S&P 500 Outperformance With Lower Volatility

One of the most important decisions that an investor can make is the degree to which his or her portfolio is coupled to the broader market. This decision is quite distinct from risk tolerance. Whether or not you want an aggressive (high return / high risk) portfolio or a more conservative (lower return / lower risk), it is a good idea to be proactive in deciding how much you want your portfolio to be a bet on the direction of the market as a whole. Bonds help investors to temper portfolio sensitivity to the broader market, but only at the expense of returns. The correlation of your portfolio to domestic markets can be managed to a limited extent by investing in foreign market indices, but these are also quite highly correlated to the S&P500 [pdf].

If you want to build a portfolio that can generate returns consistent with a high degree of equity exposure but have concerns about the direction of the overall market, a promising strategy is to assemble a series of stocks that have low Beta values and have relatively low correlation to one another.

This approach allows you to have a portfolio with solid returns but which is largely de-coupled from the performance of the S&P500. Put into statistical terms, this means that you have a portfolio that has low Beta and low R-squared (for definitions see Investopedia.com). After writing the article above, one of the users of Quantext Portfolio Planner, our portfolio management software, developed a portfolio designed to achieve these objectives and he asked my opinion of it. The portfolio was interesting enough that it merited an article. I am calling his portfolio the ‘No Direction’ portfolio because its performance does not depend on the direction that the S&P500 goes.

This portfolio varies somewhat from the one that was sent to me. In particular, the original portfolio used TIP, the TIPS ETF, rather than the bond fund and also included another REIT. I have used VBIIX and COY because they achieved many of the same goals but had longer data histories and I wanted to look at this portfolio in a range of historical market conditions. Aside from these substitutions and some changes to the allocations, this is very similar to the portfolio that I received. This portfolio is invested in a range of companies with well-known brands and spans many different industries. If you want to reduce your exposure to swings in the broader market, many of these stocks are all-stars. The investor who designed this basic portfolio went to considerable effort to find stocks that exhibited low correlation to each other and to the broader market, while still generating solid returns. The historical performance of this portfolio has been quite attractive:

In the trailing three-, five-, and ten-year periods, this portfolio has out-performed the S&P500 with a lower level of volatility (i.e. risk) than the market as a whole. This is not, in and of itself, a terribly notable achievement. What makes this portfolio more interesting is that it has exhibited very low Beta and low R-squared for all of these periods:

This portfolio has achieved its returns without tracking the market very closely. To give a sense of how interesting this portfolio feature is, consider the Beta and R-squared values for a series of index ETF’s:

SPY is designed to track the S&P500 perfectly, so Beta and R-squared should both be 100%. TIP, an inflation-protected bond fund, is expected to have very low Beta and very low R-squared. This is characteristic of bonds. ICF, a real estate focused ETF, has a fairly high Beta but a low R-squared. The trailing three years have been a bit odd for real estate. As real estate has become a hot asset class, its Beta and R-squared have increased dramatically. The trailing five year Beta and R-squared for ICF are considerably lower at 48% and 16% respectively.

Building a portfolio with low Beta, low R-squared, and a return that is considerably above bonds on an absolute and risk-adjusted basis is easy if you are willing to look back and optimize to the past. A portfolio with 55% in VBIIX and the remaining fraction equally allocated between ICF and IDU has accomplished the same high returns with low Beta and low R-squared over the trailing three and five year periods (with the same level of risk) as the ‘No Direction’ portfolio, but this is a result of the fact that real estate and utilities have seen enormous rallies over this period. Such a portfolio is too concentrated in terms of style. The ‘No Direction’ portfolio shown at the start of this article is much more diverse in terms of sector allocations. The recent past is not a good guide to choosing a highly concentrated portfolio because the hot sectors tend to change in time. This is where forward-looking Monte Carlo is useful and Quantext Portfolio Planner [QPP] uses this approach to generate outlooks for the future performance of a portfolio. When I ran the ‘No Direction’ portfolio through QPP, it predicted future average annual return of 11-12% per year and standard deviation of around 12% per year. This makes this portfolio quite attractive in terms of the risk/return balance. For a general reference on this issue, see this article [pdf]. By contrast, our 55% VBIIX / 22.5% ICF / 22.5% IDU portfolio is predicted to generate an average annual return of 6.7-7.8% per year, with a standard deviation of 8.3-10.7% per year. The QPP simulation results confirm what one would expect. The ‘No Direction’ portfolio is likely to provide a higher future return because it is not simply a raw bet on a couple of sectors that have out-performed during a recent period.

At this point, our discussion has gotten a bit abstract. Let’s get back to reality. The reason why most investors might want a ‘no direction’ portfolio (i.e. low Beta and low R-squared) is to temper the impact of bear markets. Portfolio managers with a quantitative bent talk about ‘stress testing’ a portfolio. ‘Stress testing’ a portfolio means seeing how it would have held up in market conditions with big declines in the broader market. This can be particularly important because correlations between some assets and asset classes have been observed to increase when things get bad. In this case, we can examine a three-year bear market period. Consider the performance of Vanguard’s S&P500 index fund, VFINX from 1/1/2000 to 12/31/2002:

VFINX averaged -12.6% in return, with almost 19% in volatility. This was an unpleasant period for the broader market. By contrast, the ‘No Direction’ portfolio fared very well during this period (R^2 is another notation for R-squared):

Further, the metrics of correlation to the broader market are also consistent with the other market periods analyzed. Beta and R-squared are comparable to what we have seen in the most recent 3-, 5-, and 10-year periods. This kind of result provides some confidence that the statistical properties of this portfolio can provide protection when you really need it: during an extended bear market.

The concept of the ‘No Direction’ portfolio is that it does not depend on whether the broader market goes up or down. In quantitative finance, this is often referred to as a ‘market neutral’ strategy. While hedge funds develop complex market neutral portfolio strategies, a portfolio like the one shown here is a low-cost alternative. As always, I note that a ‘model’ portfolio used to illustrate a point does not mean that I am endorsing it for anyone. This is, however, an interesting portfolio. This portfolio gains foreign exposure (and lower Beta and R-squared) partly due to the presence of companies that sell a lot of their wares overseas in the local currency (AFL, PG, and BUD are great examples). There are also a lot of consumer goods companies in this portfolio. These consumer goods firms with strong global brands provide some protection from swings in the value of U.S. currency. If you want protection from market swings, these are all valuable portfolio considerations. Selling meat products may not be a very ‘cool’ business, but Hormel (HRL) has famously outperformed Motorola (MOT) since the start of 1990, when it went public, and with far less volatility. Many basic consumer goods firms are essentially de-coupled from the broader U.S. market. The portfolio has a couple of bank stocks (BBT and BAC) and even some telecom exposure, albeit in Canada (BCE). Ultimately, of course, the proof of the ‘No Direction’ portfolio is in the fact that it weathered the millennium bear market with flying colors and has shown that its performance is also de-coupled from the broader market in terms of Beta and R-squared over a variety of timescales.

The gentleman who sent me this portfolio is approaching retirement and thus wishes to protect himself from big market swings, realizing at the same time that he needs to harness equity markets to generate the growth that he needs to support a long retirement. This is a very important issue, with many advisors realizing that older investors may need to be more aggressive than previously thought [pdf].

The type of portfolio strategy laid out in this article may be a good way to go—with the allocation to bonds being determined by his total risk appetite.

Disclosure: the author owns JNJ.

Comments

[Joe Sheehan:]

Would this hold up in other bear markets? I believe the analysis is thorough, but isn't it always possible to just find a portfolio that would have worked in the most recent bear market? I would interested in what sort this would have done in 1980, 1987, and 1971 markets. I doubt an exact portfolio would work, but a similar one could possibly be constructed.

Also, what is it (besides strong brands and sector diversification) that explains why this portfolio is so strong? There has got to be some other fundamentals at work here. Maybe there isn't... i dunno... but I'd be curious as to what other aspects of a stock screen these stocks would all qualify for?

2007 Jan 12 05:23 PM

[Geoff Consi...:]

Hi Joe:

First, you are of course correct that just because this portfolio works in one bear market, it may not work in others. My point is that the R-squared and Beta over the last three and five years are also very low--this portfolio continues to be de-coupled from the broader market. This shows that the bear market performance is not some odd fluke. Further, the general themes here are sectors which have inelastic demand--i.e. demand for their products is not driven by the moves in the broader market. People need beer, and meat products, and electricity--regardles... of what the S&P500 does. Further, there is some behavioral finance here. There is an area of behavioral finance called heuristics that studies the short-cuts that people make in reaching decisions. One of the standard heuristics is called 'attribute substitution' and is more commonly referred to in the form of 'Choosing by Liking':

www.behaviouralfinance.../

In the current context, this means that people tend to think that an exiciting company will be a good investment while a boring company will be a bad investment. This is a well known investor bias. Nobody goes to dinner parties and talks about how they invested in Hormel because of its exciting business model--at least nobody I know:) The point is that there are some great investments available in boring businesses and behavioral finance explains why the value of these firms is not already'priced in'. This is my thinking on this issue--I'd love to hear from others.

2007 Jan 12 06:37 PM

[Geoff Consi...:]

Errata:

WM is Washington Mutual, not Waste Management, Inc., as listed above. The analysis used WM. Sorry.

2007 Jan 12 06:26 PM

[M Yates:]

How has this portfolio held up since Oct 2007? That would be a real test, especially with B of A and Wash Mut in it.

2008 Sep 03 05:24 PM