ITA Wealth Management

I would call the following a very conservative portfolio. We will follow Lydon and Wasik’s lead and call it a conservative portfolio.  Below are the holdings and the target percentages.

• SPY = 10%
• RSP = 10%
• EFA = 10%
• TLT = 10%
• SHY = 20%
• IEF = 20%
• TIP = 20%

The conservative portfolio yielded the following projections.

• Return = 4.73%
• Standard Deviation = 5.6%
• Portfolio Autocorrelation = 60.86%
• Diversification Metric = 36%
• R^2 = 76.2%

Readers likely noticed that in no portfolio did an asset target percentage dip below 5%.  That is a common minimum as anything lower has little impact on the portfolio.  Avoid target allocations lower than 5%.

In all five portfolios, the PA value is quite high.  This is due to the short (three-year) period used for the analysis.  Three to four years is the recommend period by the developer of the Quantext Portfolio Planner (QPP) software.

When setting up a portfolio, one of my goals is to see the projected return be similar in value to the standard deviation.  In addition, I want to see the DM value exceed 60%.

Lowell Herr

Photograph:  Villefranche, France - photo by Kenneth Appel

A few days ago I ran a Quantext Portfolio Planner (QPP) test on a rather standard portfolio, looked at the projected returns over the next year for each asset classes, and then made adjustments to see how the Returns, Standard Deviation, Portfolio Autocorrelation (PA), and Diversification Metric (DM) would respond to my changes.  Below is a data table show the ETFs I use, the different allocations, and the QPP results.  Changing the allocations improved the projected results with exception of Portfolio Autocorrelation.  Ideally, we would like that number to move closer to zero, while staying positive.  At least that is my understanding of PA based on reading the definition on the QPP web site.

Portfolio Comparisons

My preference is to invest a minimum of 5% in each asset class, but this did not provide the change needed.

The third portfolio in this series of five is the moderate portfolio.  Again, these sample investments come out of the Lydon and Wasik book.  Below is the ticker list and target percentages.

• SPY = 30%
• VO = 10%
• IWM = 10%
• EFA = 15%
• EEM = 15%
• TLT = 10%
• SHY = 10%

Running an analysis on the moderate portfolio we have the following values.

• Return = 8.39%
• SD = 14.24%
• PA = 50.27%
• DM = 26%
• R^2 = 95.4%

Note the return decrease as we move toward more conservative portfolios.

Lowell Herr

Photograph: Leaving Villefranche, France

For Premium Clients, this is a rerun of the Barclay “portfolio” vs. a balanced Vanguard “portfolio.”  As you will recall I selected nine asset classes for this study.  When I ran an analysis on the Barclay portfolio, using iShares, I came up with the following results.

• Projected Return = 10.13%
• Standard Deviation = 18.65%
• Portfolio Autocorrelation = 34.9%
• Diversification Metric = 11%
• R^2 = 93.4%

With some tactical asset allocation tweaking, it would be possible to move the return over 11% and lower the SD.  It will be very difficult to increase the DM as 11% is quite low for a portfolio diversified over these nine asset classes. We could increase the number of asset classes and see out low correlation assets.

The Vanguard portfolio does better than the Barclay portfolio on several fronts.  Here is the data.

• Projected Return = 10.82%  (Not much difference)
• Standard Deviation = 14.04%  (Lower, but not as much as preferred)
• Portfolio Autocorrelation = 39% (Essentially the same)
• Diversification Metric = 39% (Big difference and more to our standards for this type of portfolio)
• R^2 = 74.1% (The lower number is preferred as it indicates the portfolio is not following the S&P 500 as closely. R^2 will merit further study later this fall.)

Using the power of the software, we can improve all the above values if we alter the asset allocation percentages.  For this study I used 11% for each asset class and added the remaining point to VEU.  Additional study is what is planned for the new test portfolio later this fall.

Lowell Herr

The second portfolio, as defined by Lydon and Wasik, is moderately aggressive.  The holdings and target percentages are listed below.

• SPY = 20%
• VO =15%
• IWM = 15%
• RSP = 10%
• EFA = 10%
• EEM = 10%
• EEB = 5%
• ILF = 5%
• TLT = 10%

Analyzing these holdings using a three-year time duration, as before with the aggressive portfolio, note the following values.

• Return = 9.77%
  
• SD = 16.6%
• PA = 45.2%
• DM = 21%
• R^2 = 94.6%

The standard deviation (SD) is too high for the return we are collecting from this portfolio.  Further, the diversification metric (DM) is too low.

Lowell Herr

Photograph: Nice, France

The first of five sample portfolios found in Tom Lydon and John F. Wasik’s book, iMoney: Profitable ETF Strategies for Every Investor is an aggressive portfolio.  Below are the holdings and target percentages.

• SPY = 10%
• VO = 15%
• IWM = 15%
• QQQQ = 10%
• EFA = 20%
• EEM = 10%
• EEB = 5%
• ILF = 10%
• TLT = 5%

Running an analysis on this portfolio we have the following.

• Projected Return (using three-year data) = 10.58%
• Standard Deviation (SD) = 18.6%
• Portfolio Autocorrelation (PA) = 44.3%
• Diversification Metric (DM) = 20%
• R^2 = 92.3%

We can improve these values with a different asset mix.

Lowell Herr

Photograph: Eze, France

If you are building a portfolio from scratch, are you better served using Vanguard or Barclay (iShares) ETFs?  That is the question I posed to myself now that I have access to the Quantext Portfolio Planner. What I did was to populate a portfolio using the “Big Six” ETFs plus developed international, emerging markets, and REITs.  For Barclay or iShares I used the following ETFs.

Barclay

• IVE
• IJJ
• IJS
• IVW
• IJK
• IJT
• EFA
• EEM
• ICF

I invested 11% in each asset class with exception of international.  For that asset class, I invested 12%.  The time of examination was from 10/8/03 through 10/8/2008 or a five-year period.  Here are the results.

• Return = 10.90%
• Standard Deviation = 20.18%
• Diversification Metric = 12%  (See definition below)
• Portfolio Autocorrelation = 26.51%  (See definition below)

Vanguard

From the Vanguard ETF garden I plucked identical asset classes.  These are the ETFs I selected.

• VTV
• VOE
• VBR
• VUG
• VOT
• VBK
• VEU
• VWO
• VNQ

Again, I invested 11% in each asset class with exception of VEU, the international asset class.  I invested 12% in VEU just as I invested 12% in the international iShare, EFA.  Here are the Vanguard results.

• Return = 12.32%
• Standard Deviation = 9.11%
• Diversification Metric = 61%
• Portfolio Autocorrelation = 7.91%

It is obvious Vanguard is the better choice based on this analysis.  The return is greater with lower SD.  The Diversification Metric is much higher and the Portfolio Autocorrelation is closer to zero.

What is portfolio autocorrelation?

QRP and QPP both calculate an historical statistic called portfolio autocorrelation.  This is a recent feature and is not yet included in the user manual / textbook.  Portfolio autocorrelation is the correlation in portfolio returns from one month to the next.  If it is positive then high returns tend to be followed by high returns and vice versa.  If portfolio autocorrelation is negative, then the portfolio returns tend to be ‘mean reverting’ which means that very high return months tend to be followed by returns closer to the mean–the portfolio tends to damp out periods of very high or very low returns.  Portfolio theory generally assumes that autocorrelation is zero–the random walk.  QPP and QRP model the market as though autocorrelation is zero, and the metric shown is for historical performance.  If you have a portfolio that shows a lot of positive autocorrelation, this is a flag–this means that big swings get amplified.  These effects are widely debated, but there is evidence that they can be meaningful:

What does the Diversification Metric (DM) mean?

In our latest release of the software, we have added a new analytical function that accounts for non-market correlation between portfolio components.  This is important because many asset classes have correlation to one another beyond what can be captured by Beta.  This is a major challenge for many portfolios, but especially those with concentrations in a sector.  This problem is described in a recent article. (Go to QPP site to find active link.) Our software generates a statistic that measures how effectively the non-market component of returns actually diversify one another.  In the best possible case, the non-market component of returns would be totally uncorrelated with one another.  In the worst case, they would be highly correlated.  The diversification metric (DM) measures how un-correlated the non-market returns are across the portfolio.  Higher values of DM mean that the non-market component of returns shows low correlation across the portfolio.  Higher DM means that your are getting more real diversification out of your portfolio.

I’ve participated in a few discussions as to whether it is better to diversify the portfolio by using Sectors or Asset Classes.  It has always been my contention that spreading ones wealth over numerous asset classes is preferable to using sectors.  This morning I set up a portfolio built around ten sectors.  Here are the ETFs I used in this portfolio.

• IYK
• IYC
• IYE
• IYF
• IYH
• IYJ
• IYM
• IYW
• IYZ
• IDU

In each of the 10 iShares, an investment of 10% was allocated.  The test was to check the projected annual return, SD, Portfolio Autocorrelation, and Diversification Metric and compare them with the same metrics from a portfolio built around asset allocation.  It is not even a contest.

1) The projected return for ETFs using asset allocation is nearly 2% better when compared with sector diversification.
2) The standard deviation is more than double if one goes the sector route.
3) The Portfolio Autocorrelation is almost four times as high with sectors.  This is “Red Flag” country.
4) The Diversification Metric is about 1/3 with sectors.  That does not surprise me as that has been my argument against diversifying via sectors for a long time.  This only provides data for something I thought to be true.

To improve the sector diversification, I added an international ETF as well as an emerging market ETF.  The results did not change significantly.  In fact they got a little worse in three of the four metrics.

Until there is compelling evidence, I plan to stick with diversification via asset classes and let others use sectors.

Premium Content subscription available for $6.99 per month.

Lowell Herr

Photograph: Weaver in China

Next month I plan to examine the software, Quantext Portfolio Planner in greater detail.  Over the next few weeks you will see some very brief portfolio analysis using this software.  There are many articles on the Internet related to this software and the whole idea of reversion-to-the-mean (RTM). I suggest interested investors read up on this approach to portfolio management.

The goal is to design a portfolio around this forward looking approach to populating and managing investments.  Stay in touch for more information.  The development of the portfolio will likely take place over on the Premium Content side of the blog.

Lowell Herr

Photograph:  Science Museum in Florence, Italy.  This is where one will find many pieces of apparatus used and designed by Galileo.  Also, his middle finger is on display, likely saluting his past foes.

Premium Content subscription is available for $6.99 per month.

Here is an example of an analysis for the Passive Portfolio, an account that has been in operation for nearly eight years.  The portfolio holds the following investments with the respective target percentages.

• IVE = 14%
• IJJ = 15%
• VBR = 13%
• IVW = 13%
• VOT = 13%
• IJT = 5%
• EFA = 15%
• ICF = 5%
• EEM = 7%

The projected return for such a portfolio is 11.03% and the risk is 16.44%.  I consider this risk to be too high and unacceptable.  This issue is, what does one do about it?  That is now part of the analysis.  What ETFs should be replaced to maintain a return greater than 11% while reducing the standard deviation?

Lowell Herr

Premium Content subscriptions available for $6.99 per month.