Building a Momentum Based ETF Portfolio

There has been some excellent work written recently about the merits of a momentum based asset class rotation system. Two good starting points are Mebane Faber and an article on CXOAG. The system is simple: Rank a list of securities on the average of the trailing 3, 6, and 12 month total returns. Purchase the "top" of the list and rebalance monthly. I put "top" in paranthesis because there are different approaches one could take and evidence seems to show that the smaller your "top", the more volatility you can expect.

Putting this in practical application we can use one of the portfolios I detailed earlier this month. For simplicity, we can look at the basic portfolio:

	Symbol	Allocation
	VTI (US Stock)	20.00%
Basic	VEU (Intl Stock)	20.00%
Portfolio	BND (Bond)	20.00%
	VNQ (REIT)	20.00%
	DBC (Commodity)	20.00%

Now, lets look at the 3, 6, and 12 month returns to determine the highest average using the 6/30/09 closing price:

	June 30th Close	03/31/09	3 Month Return
VTI (US Stock)	46.27	39.58	16.90%
VEU (Intl Stock)	35.91	28.12	27.70%
BND (Bond)	77.69	77.15	0.70%
VNQ (REIT)	31.01	24.28	27.72%
DBC (Commodity)	22.62	20	13.10%
	June 30th Close	12/31/08	6 Month Return
VTI (US Stock)	46.27	44.74	3.42%
VEU (Intl Stock)	35.91	32.32	11.11%
BND (Bond)	77.69	78.97	-1.62%
VNQ (REIT)	31.01	36.45	-14.92%
DBC (Commodity)	22.62	21.19	6.75%
	June 30th Close	06/30/08	12 Month Return
VTI (US Stock)	46.27	64.09	-27.80%
VEU (Intl Stock)	35.91	52.79	-31.98%
BND (Bond)	77.69	76.2	1.96%
VNQ (REIT)	31.01	58.23	-46.75%
DBC (Commodity)	22.62	44.82	-49.53%
	Aggregate Return	Rank
VTI (US Stock)	-7.48%	3
VEU (Intl Stock)	6.83%	1
BND (Bond)	1.04%	2
VNQ (REIT)	-33.95%	5
DBC (Commodity)	-29.68%	4

Thus, the best performer (and buy) on 6/30/09 would be VEU. And second best would be the total bond market, BND.

To add an additional twist, both the CXOAG and Faber studied the impact of only buying the security IF it was above its 10 month simple moving average. If we were to add this factor, we get the following:

	Average Return	Above 10 Month SMA on 6/30/09
VTI (US Stock)	-7.48%	NO
VEU (Intl Stock)	6.83%	YES
BND (Bond)	1.04%	YES
VNQ (REIT)	-33.95%	NO
DBC (Commodity)	-29.68%	NO

Thus, at the start of July, VEU had the top ranking and maintained a Buy signal even when incorporating the 10 month SMA qualifier. To diversify further, expand the total number of securities you analyze and also the number of securities you purchase. For this portfolio, also purchasing BND would give some additional diversification for July and potentially help limit drawdowns.

Beginning on the first of every month, I will be updating these results and other portfolio strategies on my blog, Scott's Investments.

Disclosure: none

Comments

[richjoy403:]

Thanks Scott -- I'm sure many of us are hoping you will stick with this type of analysis for the long term.

By whatever name, I hope I will find one I am comfortable with.

For example, I have read of a "Real Return" strategy wherein one can match or beat the S&P500 over the economic cycle if one can match 65% of the upside, and limit losses to only 50% of the downside. [Scott, any comment on this?]

I suspect the difficulty will be in having enough confidence in any system to stick with it long enough to be convinced it does or does not work.

Jul 11 12:29 PM

[Scott's Inv...:]

As they say, past performance is no guarantee of future returns :). However, check an idea I posted here about using multiple strategies, scottsinvestments.blog...

Jul 11 01:07 PM

[Scott's Inv...:]

As they say, past performance is no guarantee of future returns :). However, check an idea I posted here about using multiple strategies, scottsinvestments.blog...

Jul 11 01:08 PM

[richjoy403:]

Thanks Scott for leading me to those articles. There is a lot to digest there, and I will study them.

Jul 12 07:07 AM

[numbercrunc...:]

What gives? According to both Yahoo Finance Interactive Charts and StockCharts.com VTI at 46.27 was above the 200 day SMA I am assuming that is same as 10 month i.e. 20 trading days/month x 10 months??? Stockcharts.com has the SMA at 44 and change. My own calculations have 44.14 assuming price is adjusted for dividends. Curiously Yahoo interactive says the 200 SMA on 6/30 was 42.47. But both are below 46.27. Is the 200 day SMA not the same as 10 month or is there some other reason?

Jul 14 07:56 PM

[Scott's Inv...:]

They are different - the 10 month SMA is the last 10 trading months, the 200 day is the last 200 trading days, which is not equal to the 10 month. Some months there can be more then 20 trading days. My source on this article was my online broker's data, but going forward for simplicity I will probably be using Yahoo for everything so the information is accessible for everyone.

Jul 14 10:54 PM

[numbercrunc...:]

Thought of that after I posted the initial comment, but it doesn't make much difference. Using Yahoo historical price data (adjusted close) SMA using prior 10 month ends (only 10 data points) is 44.66, SMA for all trading days in prior 10 months (209 data points) is 44.90 and 200 day SMA is 44.14. In each case June 30, 2009 close was above all three. I reset the interactive chart parameters and now it is giving the correct 200 day sma. Not sure why it was giving bad data last night. Not reassuring!


On Jul 14 10:54 PM Scott's Investments wrote:

> They are different - the 10 month SMA is the last 10 trading months,
> the 200 day is the last 200 trading days, which is not equal to the
> 10 month. Some months there can be more then 20 trading days. My
> source on this article was my online broker's data, but going forward
> for simplicity I will probably be using Yahoo for everything so the
> information is accessible for everyone.

Jul 15 10:46 AM

[Scott's Inv...:]

Yes, you are correct yahoo is showing an SMA of 44.66. This article was written using data collected from Thinkorswim, my guess is their data is not using the adjusted close which could account for the difference but I'll follow up with them. Starting in August I will be posting on my blog links or spreadsheets for the data to make it more transparent; in addition, I'm finding that the 200 day SMA seems to be a more popular average for investors, so a permanent switch may be made there as well. Thanks!

Jul 15 12:45 PM

[dardhel78:]

What are your thoughts on the 200 day EMA? I read one ETF book, iMoney, that advocated using EMA over SMA. Just curious as to what you think.

Jul 15 05:45 PM

[Scott's Inv...:]

I think as long as your consistent (not switching back and forth) that is the most important element. EMA has advantages since it weights current data more heavily, but most academic studies I've read use the SMA which is the reason why I incorporate it into the article. Others suggest using EMAs for shorter durations and SMAs for longer durations. Consistency is key, no matter which one you choose.

Jul 15 05:53 PM

[User 457591:]

Good article. I'm looking for a spreadsheet or website that will easily do all the work around finding and averaging the trailing 3, 6, and 12 month total returns. And a site that would do this plus show the 200 day sma would be even better. Any suggestions on where I can get all this information, or do I need ot create my own spreadsheet and pull information from somewhere else?

Jul 25 03:14 PM