Swensen 6 ETF Portfolio.
Cluster Weighting Momentum Analysis.
Over two years passed since I posted an article on the Swensen's 6 ETF Portfolio. How did the original portfolio perform if no changes were made over the past 25+ months? If one had done nothing and stayed with the original percentages, the Swensen 6 generated an Internal Rate of Return (IRR) of 13.2% (plus dividends of approximately 2.9%) vs. the VTSMX benchmark of 19.7%. Of the six ETFs, which were a drag on the portfolio, as the Swensen 6 did not match the benchmark? Here are the individual performance results.
- Vanguard's Total Market ETF (NYSEARCA:VTI) - 19.6% or nearly a match of VTSMX. Something we expected.
- Vanguard's Developed International Market ETF (NYSEARCA:VEU) - 30.1% this was a surprise.
- Vanguard's Emerging Markets ETF (NYSEARCA:VWO) - 3.7% I expected more, as this ETF has been performing better recently.
- Vanguard's REIT Index ETF (NYSEARCA:VNQ) - 10.2% or about as expected.
- iShares TIPs Bond (NYSEARCA:TIP) - negative -0.7% but not quite this bad, as dividends are not included in this calculation.
- iShares 20+ Year Treasury Bond (NYSEARCA:TLT) - 0.4%, and a drag on the overall portfolio performance.
No dividends are included in these individual IRR percentages.
Were I writing the original article today, I would use VEA instead of VEU for the Developed International Markets representative, as VEU includes a number of Emerging Market stocks and that asset class is covered by VWO. However, I will not quibble with the return for VEU.
Instead of buying and holding these six ETFs, what if one employed something known as the Cluster Weighting Momentum [CWM] model? The CWM model breaks the six ETFs into clusters that have a correlation less than 0.8. One can vary this value, but 0.8 is what I use throughout this analysis. Based on three weighting factors, the ETFs are ranked every quarter. This is explained in more detail in this Seeking Alpha article. ETFs are bought and sold based on the CWM software recommendations.
While the improvement is not huge, the Swensen 6 Portfolio increased by 110 basis points annually over the Buy & Hold portfolio. A bull market is not a true test of the CWM model, as it works best when a bear market is part of the testing period, as it really is a volatility reducing application.
Here are the individual ETF performance results when the CWM is applied from 4/26/2012 through 6/10/2014.
- VTI - 17.5% or a little higher than the Buy & Hold.
- VEU - 31.2% or a little higher than the B&H.
- VWO - 13.7% or a major improvement over B&H. Since the initial allocation was 5%, this was an asset class where recommendations over the two-year period had an opportunity to shine.
- VNQ - 13.8% Again, a small advantage over B&H.
- TLT - 1.2% A move out of negative territory.
- TIP - 4.7% or a non-trivial improvement over B&H.
Once more, dividends are not included in the above percentages.
What are the disadvantages of the CWM model?
- There are tax implications.
- Bid-Ask spread can be a problem. Price slippage is inevitable when there is buying and selling. I use limit orders almost exclusively, but this also has inherent problems in a bull market.
- It requires some work to go through the CWM process each quarter, a minor inconvenience.
- There will be periods when one is not invested in income generators such as VNQ.
The big advantage of the CWM model shows up when bear markets hit such as we experienced in 2002 and again in 2008 and early 2009. The CWM model shines brightest when more low-correlated ETFs are included for possible investing. Additional out-of-sample testing of CWM is ongoing so stay tuned.
Disclosure: The author is long VTI, VEU, VWO, VNQ, TLT, TIP, SHY. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.