Global Giants and Diversifiers To Supercharge a Portfolio

In our new book Yes, You Can Supercharge Your Portfolio, Ben Stein and I describe a series of investing stages. We hypothesize that most investors do not begin with an overarching theoretical framework for their holdings, but rather own what we term a “Glom” (for agglomerative) portfolio, made up of scraps of whatever looked good at the time they made the initial purchases. The problem with this “cats and dogs” approach is that there is no overall sense of what it means: the likely risks and returns are unknown, as is its suitability to meet the investor’s short- , intermediate-, and long-term goals. There is no match between assets and liabilities, or knowledge of whether the assets are invested in an efficient manner.

We recommend that investors in this position could “supercharge” their holdings by moving toward what we call a “Glob” portfolio – a great global glob of securities held in some conventional, common-sense allocation. This is essentially the John Bogle move: using index funds to pursue a low-expense, highly diversified set of holdings. This captures the free lunch of low expenses and a market-wide degree of diversification to improve the risk/return portfolio efficiency in harnessing the returns from global capital markets. For many investors, we think this would be a giant step forward.

A third stage involves “supercharging” the globally-indexed portfolios. This involves the use of portfolio management tools as opposed to the seat-of-the-pants approaches outlined above. Here we employed Geoff Considine’s Quantext Portfolio Planner Monte Carlo simulator [QPP] to address the remaining deficiencies in the “Glob” portfolios. In the first place, we used targeted assets – ETFs and/or individual stocks – that had low correlations with the rest of the global portfolios to boost the risk-adjusted returns. An example of how this worked during the recent market correction is shown in a previous article.

Then we recommend adding bonds to control portfolio volatility, and tuning the overall allocation so that it meets the investor’s goals. For most, the big long-term goal here is successfully preparing for retirement.

The advantage of the supercharged “Glob” portfolio is that the core of the portfolio is familiar, anchored in major market indexes. The tracking error is not severe – investors holding this type of portfolio will not be shocked to discover large discrepancies between their holdings and the daily headlines in The Wall Street Journal. The projection is for the returns to be slightly better and the risks to be slightly lower – small but meaningful differences that will compound into a large advantage over time.

Finally, the book comes full circle – to a portfolio composed entirely of individual stocks, this time not picked haphazardly but with each security selected with an eye toward its contribution to the portfolio as a whole. We do not recommend this for most retail investors, because individual stock selection calls for skills that may be beyond their expertise, and because the tracking error may be considerably greater than with the “Glob” portfolios above. For example, if this portfolio were to go down when the indexes are up, amateur investors might become discouraged and sell when the market is low, leaving them worse off than before.

While we did not recommend any particular portfolio, but we did sketch one methodology how such a portfolio might be assembled, following a fairly conservative approach (pp. 140-148).

We began by looking at global-dominating companies – giants with over $10 billion in market capitalization whose stock prices had exhibited low volatility (standard deviation) in recent years. Next, we weeded through the list to pick just one company from each industry group: for purposes of this illustration, we didn’t want to own four oil companies, four pharmaceutical companies, etc. – not that there’s anything wrong with that. We came up with a core list of ten companies.

Next, we sought to diversify these global giants by adding ten more companies that had a low-correlation to them and that appeared to provide a valuable diversifying function when included with them. That is, these companies tended to either boost the expected returns of the resulting portfolio, lower the expected standard deviation, or both. Many of these companies also were low in volatility themselves, and some of them had performance histories that were largely independent (low R-squared) of the stock market as a whole. The final list is shown below.

click all images to enlarge

We allocated 80 percent (8 percent/each) to the ten global giants and 20 percent (2 percent/each) to the ten diversifiers. We used the Quantext Portfolio Planner to gather five years of historical data (from 12/31/2001 through 12/31/2006) and project future risks and returns for the resulting portfolio. While the individual company names taken one-by-one did little to get our hearts racing, taken as a group they seemed to display admirable characteristics – making it an almost textbook example of Modern Portfolio Theory in practice, as shown in the following screenshot from QPP.

Note that the expected return (12.4 percent) is even higher than the expected standard deviation (10.3 percent) – a feat seldom achieved in markets with normal levels of volatility.

A further look at the correlation table shows this to be a highly diversified slate, with the highest correlation of any security to the portfolio as a whole at registering at 0.55 and most falling well below that. This is a portfolio with a high degree of internal diversification.

Historically, QPP shows that the portfolio has a beta of 0.44 and an R-squared of 0.58. It looks to be much less volatile that the stock market as a whole and is not just another closet index portfolio. QPP makes a “value at risk” calculation, and over a three month period it projects that this portfolio has a 1-in-20 chance of losing 5 percent of its value – a remarkably small expected loss for an all-equity set of holdings. This suggests that we are banking on portfolio effects to come through for us during bad times.

This last projection became of singular importance when, no sooner was our book released, than world stock markets entered a major correction. For the three months from 10/31/2007 through 1/31/2008, the S&P 500 (SPY) fell 10.6 percent, foreign stocks (EFA) declined 13.8 percent, and emerging market stocks (EEM) lost 17.1 percent of their value. How did our all-equity, all-individual-stock portfolio fare during this period? Since indexes, with their diversification into thousands of individual positions, are often regarded as lying at an extreme point of risk-adjusted returns, had our portfolio – with its much smaller set of holdings – really held up, or had centripetal forces torn it apart as all asset classed autocorrelated into a tailspin during those months?

Our portfolio of individual stocks was down 3.4 percent from 10/31/2007 through 1/31/2008. It performed much better than the indexes, and in fact had behaved in line with the expectations engendered by the QPP’s calculation of a possible loss in the 5 percent range. This is all the more remarkable when one considers that the portfolio had a market-like weighting toward financials, with Berkshire Hathaway, Wells Fargo, and about half of General Electric (GE Credit) falling in the worst-performing market sector over the period. By way of comparison, QPP showed a value at risk for the S&P 500 (SPY) for a 10 percent loss over a 90-day period – just slightly under what the S&P 500 actually experience during this correction.

In the short run, our experience with a didactic portfolio of individual stocks held up well during a sharp market correction. It remains to be seen how it will perform over longer periods, during bull markets, and how it will compare with simple market indexes.

Comments

[granger]

Myabe the makings of a semi-actively managed ETF. I would rather on an ETF designed like that, then a market blanket one.

Good stuff, keep us posted. I thorughly enjoyed it.

[Geoff Considine]

Note also that the EXPECTED return for this portfolio was higher than the historical trailing return. Reversion to the mean was set to act in favor of this portfolio. This is an attribute that I like in a portfolio and was not true of a number of broad indexes at the time.

[Roger Beep]

Hi Phil;

I thought that this was another well written article. In addition to the "inputs" or holdings portion of the portfolio I would like to see a chart of the "outputs" to include the portfolio’s projected return using QPP's forward looking analysis using perhaps the 1st, 5th 10th, 20th and 50th (Median) values from p. 6 of QPP and the actual monthly return. I invite you to examine a standardized Excel chart template for this purpose which is resourced at harderchartingtemplates.pbwiki.com / and listed as SCCFB_mcpm_v7_1.xls. This chart covers a ten year period for each month. See the "LineFree" chart. It appears that BlueLine and LineFree have been switched. It should read "BlueLine." This would allow for comparisons using a standardized charting metric.

[Phil DeMuth]

Mike & Geoff: Thank you for your comments.

Roger: Thank you for your helpful suggestion. I tend to take a rather unsophisticated view of these things. I look at where the estimated median returns are, and then I look at the 1st percentile value-at-risk for an educated guess as to what I might experience in a really terrible year, short of World War III. Then, if I can't do the time, I don't do the crime, and I make adjustments. As for the upside, I figure it will take care of itself. As Peter Bernstein says, I can't control the returns, but I can control (at least to some extent) the risk. This is part of the message of the book as well. We see people devoting endless attention to returns but insufficient attention to risk.

[Roger Beep]

Hi Phi;

I appreciate you thoughts regarding taking a rather unsophisticated view of things, the 1st percentile risk, estimated median returns and endless attention to returns. What I really liked about the flow of the Supercharge book was the emphasis on risk and with examples showing us what to avoid up front and then moving on to core holding portfolios using indexes and ending up with an all stock portfolio that was developed with QPP.

I could live with the "unsophisticated view" that you mentioned if I or perhaps you had developed the Individual Stock Portfolio that appears in the Supercharge book for my personal use. And on a practical level I think you are correct. However, as a purchaser of the Supercharge book and QPP I would like to see more on the predictive validity of QPP using the example portfolios.

That is the reason for advancing a standardized graphical assessment. For example, the standard deviation (S.D.) levels from page 6 of QPP can be visually represented by the frequency envelope about the median using the Excel chart. I believe this can be varied by the user to reflect various S.D. levels. As I recall an X2 (read as “times 2”) frequency envelope is a ± 0.66 S.D. This all done graphically without having to resort to complicated statistical procedures. Most end users would much prefer a picture (chart) of these matters as opposed to confirmatory statistical data or summative comments about declines and gains. The upshot is one would be able to graphically see how QPP tracks to actual returns over time. In other words we would have a graphical picture of QPP’s predictive validity given the portfolio examples in the Supercharge book.

[Phil DeMuth]

Hi Roger -- You'll get no argument from me here, I love charts. Perhaps Geoff will put this idea in the hopper for QPP version 5.0. In the mean time, if you have QPP, you can enter the tickers and date ranges and do the analysis yourself.

Imagine that the 25%ile rank showed a loss of 1%. Then, a year later, the portfolio is down 1%. Is that a successful prediction? Perhaps. It would take a lot of samples to confirm -- you are likely to run out of lifetime before you accumulate enough data points, and by then it will be too late anyway.

So it's really the tails that are of interest, and why the past few months are so relevant to examine here. Since the whole thing is based on the currently unfashionable but nevertheless highly relevant normal distribution curve, all the information (Mean - 2SDs, for example) is really contained in the tail values. Unless I'm mistaken, the shape of the chart in every case would be a standard curve.

The issue of how to graphically present the risk/return tradeoffs in the most compelling way is one I often think about. For clients, I might present it as median expected return vs. 1 percentile loss at different equity allocations. I think this is more communicative than just mean vs. standard deviation.

[Roger Beep]

Hi Phil;

I might try it. I have a few concerns about undertaking this project. One concern I have is determine the actual monthly return on the Global Giants and Diversifiers Individual Stock portfolio. If I personally held that portfolio the brokerage firm would tell me what the monthly return was.

Let me step you thought how I might determine the monthly return and then you may wish to comment. Since this portfolio represents several years of investing experience and a few generations of portfolio development as per the Supercharge book I am going to assume $100K in total assets -- in reality it should be more but for the sake of modeling I will stipulate $100K.

Using Yahoo Finance I think I can determine the closing price of a stock for any particular trading day. I will assume I bought all of the holdings on 12/31/2006. I am unclear how I could from a retrospective point of view determine the monthly return on the portfolio, including the dividends.

Additionally, I would like to automate future monthly return data so I did not have to be scheduled in front of my computer to get that month's return data. From where I stand there are some practical considerations that I would need to figure out and for that reason I may not be able to undertake this project at least right away. For example, I anticipate having to write a macro for Excel to call the data that I would need. I am not that skilled at Excel but for a guy that self-taught himself computing in the old DOS days I feel like I could probably take it on. This could be a great learning experience for me after I clear the deck on a couple of other intense projects I have currently under taken.

Also, I appreciate your observations about the extremes, the tail values and what they portend for a porfolio's risk vs. return and the median expected return vs. 1st percentile loss.

[Phil DeMuth]

Hi Roger --

Yes, you can calculate the monthly returns based on Yahoo! historical quotes or you can enter a transaction portfolio into Morningstar and they will keep track of it for you. You would still need to use Yahoo! to get the historical prices when you set up the portfolio, but from then on Morningstar would have it on autopilot for you. Good luck with it.

[Vig Oren]

Hi, since when Morningstar's Portfolio Manager, in the "Transaction" mode, updates account's (ticker's) number of shares without entering it manually, all the time?

[Vig Oren]

BTW, M* Portfolio Manager now shows 3 years standard deviation of a portfolio. It also shows Expected Annual Return in the Asset Allocator (AA) tool. When I entered Geoff's "modified David Swensen's" allocation into these M* tools I got the three years SD at 11.45, same as Geoff got from the QPP. However, the expected annual return at the AA tool shows 7.5% vs. Geoff's at about 10% for the next three years. It would be interesting to wait and see which is right. M* AA also shows a possible 3 month loss of 10.3%, for this allocation (from today).