Last week we explored a simple dual momentum strategy using three portfolios consisting of three mutual funds. The strategy was inspired in part by Gary Antonacci’s ”Risk Premia Harvesting Through Dual Momentum” paper available on Optimal Momentum.
The first article generated interest and several readers emailed questions. To reiterate, mutual fund were used strictly because of their longer trading histories. Since most mutual funds have trading restrictions, a real-world application of the strategy would be better served with exchange-traded funds (ETFs) that do not have many of the same trading restrictions. However, with ETFs come commissions, which will reduce overall returns.
The portfolios in the first article were not meant as optimal portfolios but there was a rational behind the security selection. Funds were grouped into three asset classes – equities, bonds, and real assets – that historically perform differently in different economic conditions. Within each asset class two funds were selected that have historically low correlations, although there have been extended periods of time where US and emerging equities, for example, have been highly correlated.
The results from the first article were impressive when tested from 2003-present. However, results over the past two years have been more tempered. The recent returns are one example of why it is important not to rely on one test, portfolio, or time-frame to draw broad-based conclusions.
What happens if we modify the portfolio holdings? Does the dual momentum strategy hold up well? Also, what if we just equally weight the portfolio and re-balance annually? Does the dual momentum strategy still out-perform? I attempt to answer those questions in the tables below.
The strategy for the backtests involves purchasing one mutual fund in each portfolio and equal weighting the three portfolios to create a “complete” portfolio. Purchases are determined by the one fund in each portfolio which has the highest trailing 6 month returns. The strategy rebalances each month, selling the current holding if it is no longer the top ranked fund in its portfolio and replacing it with the fund which has the highest momentum.
You will notice a short-term treasury fund is in each portfolio. Combining absolute and relative momentum means a mutual fund outperforms a risk-free asset class such as cash in addition to outperforming all of the other asset classes in the portfolio Thus, for the purposes of this test I added a short-term treasury fund to each portfolio to represent a risk-free asset and to act as a comparison point for absolute momentum.
Portfolio #1 consists of the same securities as my first article:
Portfolio #1 | |
Equity Basket | Symbol |
Emerging Market | VEIEX |
US Total Stock Market | VTSMX |
Short Term Treasury | VFISX |
Bond Basket | |
Emerging Market Bond | PREMX |
Long-Term Investment Grade US | VWESX |
Short Term Treasury | VFISX |
Real Asset Basket | |
Precious Metals & Mining | VGPMX |
MSCI REIT | VGSIX |
Short Term Treasury | VFISX |
Portfolio #2 consists of the following securities:
Portfolio #2 | |
Equity Basket | Symbol |
Emerging Market | VEIEX |
US Total Stock Market | VTSMX |
Short Term Treasury | VFISX |
Bond Basket | |
Emerging Market Bond | PREMX |
Long-Term Investment Grade US | VWESX |
Short Term Treasury | VFISX |
Real Asset Basket | |
GSCI Index | GTY |
MSCI REIT | VGSIX |
Short Term Treasury | VFISX |
Portfolio #3 consists of the following securities:
Portfolio #3 | |
Equity Basket | Symbol |
Total International Stock | VGTSX |
US Total Stock Market | VTSMX |
Short Term Treasury | VFISX |
Bond Basket | |
Emerging Market Bond | PREMX |
Long-Term Investment Grade US | VWESX |
Short Term Treasury | VFISX |
Real Asset Basket | |
GSCI Index | GTY |
MSCI REIT | VGSIX |
Short Term Treasury | VFISX |
The returns for the different portfolios and different strategies are in the next table. The time-frame for the tests is 2003-December 21st, 2012. Two benchmarks have been added, a 60/40 Balanced Vanguard Fund (VBINX) and the SPDR S&P 500 Index ETF (SPY).
The equally weighted portfolios all have 7 positions weighted at 14.285% and include one position in the Vanguard Short-Term Treasury Mutual Fund (VFISX). The equally-weighted portfolios include results with an annual rebalance of each position back to 14.285%.
In each case, the dual momentum strategy still out-performs on an absolute and risk-adjusted basis its equally-weighted portfolio and the benchmarks. All test results courtesy of ETFReplay.com:
Strategy | Total Return | CAGR | Sharpe | Volatility | Max Drawdown |
Portfolio #1 Dual Momentum Strategy | 398.50% | 17.50% | 1.09 | 12.80% | -18.50% |
Portfolio #1 Equally Weighted, Annual Rebalance | 196.30% | 11.50% | 0.66 | 13.60% | -42.02% |
Portfolio #2 Dual Momentum Strategy | 357.30% | 16.50% | 1.14 | 11.50% | -13.40% |
Portfolio #2 Equally Weighted, Annual Rebalance | 153.10% | 9.80% | 0.61 | 11.60% | -40.94% |
Portfolio #3 Dual Momentum Strategy | 320.20% | 15.50% | 1.12 | 10.90% | -13.10% |
Portfolio #3 Equally Weighted, Annual Rebalance | 126.20% | 8.50% | 0.53 | 11.40% | -40.79% |
60/40 (VBINX) | 98.20% | 7.10% | 0.39 | 12.20% | -35.97% |
SPY | 97.20% | 7.10% | 0.3 | 20.70% | -55.20% |
as i recall mr. antonacci used a 6 month lookback for measuring relative strength and a 1 year lookback to compare the asset with the highest RS vs. cash. curious to know the lookbacks you used.
Hi Steve, I used 6 months for all look-backs in this article
Antonacci used 12 months. See page 7 of the Dual Momentum paper.
Using 6 months on ETFreplay is a bit of datasnooping, since it happens to capture the dramatic losses in 2008. Put back 12 months and the return drops. Although it’s still quite nice.
You are correct, he does, and I point that out in some of my other articles.
Hi Scott, is this a possible portfolio for 2013? The dual momentum strategy may have slowed down a little but still made money in 2011& 2012. The flexibility of ETF or mutual funds is a bonus, since most discount brokers charge hefty commissions for mutuals along with trading restrictions etc. Gary just updated his research at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2042750.
Risk Premia Harvesting Through Dual Momentum.
Keep us posted on 2013 developing a tracking portfolio or not. Thank again for all your research and wisdom. Happy New Year!! john b. in Nevada
Hi John, I will try to put together something along these lines for 2013, there will be some changes and additions in 2013. Happy New Year!
Yes a portfolio along these lines would be awesome – but using ETFs not Mutual Funds please!
ETFREPLAY and ETFSCREEN both provide a great resource for statistics on prices and various momentum and moving average measurements for ETFs.
Are you aware of a service which provides similar data for the common stocks of the NYSE?
Based on the growing interest in the dividend income payouts of common equities and fixed income securities,
Due to the growing number of seniors who seek a growing stream in income, as a private investor I would like to know of a service which provides data relative to the past, current and projected “growth of the dividend payout stream” generated by ETFS and Equites.Such a service would differentiate the dividend records by showing the “trailing 12 months payout vs current price”yield,the yield based on the annualized most recent monthly or quarterly payout, and the future dividend payout based on a stated projection of the next 12 month’s of dividend payouts
for the next twelve months.
I believe I am not the only one seeking such information and easy reference resources for such information.
Tal Fletcher
mailto:talfletcher@yahoo.com">talfletcher@yahoo.com
I am not, but there are several good authors on Seeking Alpha that focus on dividend stocks. Perhaps I will look to add more dividend features going forward
GTY is not the GSCI ETF.
You may have meant GSG?
GTY is Getty Realty. That will bias your simulations upward. At least it did for me.
No, I meant GTY but it is the symbol used on ETFReplay for the GSCI Index so it can be a bit confusing. It is an index, not an ETF, and the simulations were run using the GSCI Index.
Has anybody run a real time portfolio using this methodology since it appeared on this website? I have been running something since Feb 2013 and the return ex dividends is practically zero. The dividends may account for maybe 8% over the period but I don’t have reliable stats on that.
I would like to know how far back you backtested this strategy & if you could post annual returns, sharpe ratios & maximum drawdowns
They will be posted in January