Portlandia
Dual Momentum is a combination of Relative Strength Momentum and Absolute Momentum. Relative strength momentum compares the trend of one security with respect to another security. Platinum members are familiar with this concept as we are always comparing the strength (price performance) of one ETF with another ETF. Absolute momentum or absolute acceleration examines the trend of an asset (ETF or stock) with respect to its own past history.
Dual Momentum Investing is Gary Antonacci’s book (copyright @ 2015) that lays out a momentum plan to increase portfolio returns while at the same time seeking to reduce risk. The plan is described below and ITA readers will find few differences from the Rutherford Portfolio model.
On page 101 of the book is a diagram laying out the dual momentum strategy. Here is the model verbalized.
1. Select three broad based index instruments.
- U.S. Equities – VTI is my choice for this asset class.
- International Equities – VEU is my choice as it also includes emerging markets. In our list of assets we use VEA and VWO to cover both developed and emerging equity markets.
- Bonds or U.S. Treasury bills – BND is my choice, but one could use AGG.
2. At the beginning of each month run a screen to see if VTI performed better or worse than VEU (VTI > VEU). The look-back period recommended by Antonacci is 12 months. Here at ITA the default settings are 60- and 100-day combination. If VTI is performing better than VEU then run another relative strength momentum test, this time comparing VTI vs. TLT. If VTI > BND, invest all money in VTI.
3. If VTI < BND, buy and hold BND.
4. Going back to step #2, if VTI < VEU, then compare the relative strength momentum of VEU and BND. If VEU > BND, buy and hold VEU until the next review, but if VEU < BND, buy and hold BND until next month. In one month, repeat steps 2 through 4.
Antonacci does not mention what one does if all three ETFs are showing negative absolute momentum. At least I missed it on the first read if he does. One option would be to go to cash or SHY. With inflation on the horizon, consider investing in TIP or SCHP. These are inflation protection ETFs.
While the Dual Momentum model is simple to follow, the book contains much data as to how this method of investing worked over the past 40 or more years. There is still no rational explanation as to why it works.
Discover more from ITA Wealth Management
Subscribe to get the latest posts sent to your email.
Rereading Appendix B, if both VTI and VEU (my proxies for U.S. Equities and International Equities) are negative, we flee to TLT as the safe haven from a declining market.
Lowell
Lowell,
Simply put, could you just pick the top performing ETF (VTI,VEU,TLT) monthly based on your ITA relative strength rankings (91 days 50%, 181 days 30%, volatility 20%) then go to cash (SHY) if none of the 3 ETFs has positive absolute acceleration? Positive acceleration would be that the 3 month, annualized rate of return is greater than the 6 months?
Thanks,
Richard
Richard,
Yes, that is it. A very simple model that will keep us out of those deep bear markets.
Lowell
For anyone wishing to use Antonacci’s simple strategy using the 7.1.3 Ranking SS with a 12 month lookback period:
1) Setup the Portfolio assets/ETFs to be used (e.g. VTI, VEA and TLT – together with SHY) by deleting existing tickers (Col C of Portfolio Tab) and replacing with the required tickers. Download data in Data Sheet.
2) Set the value of ROC1 (in Cell G10 of the Rankings Sheet) to 365;
3) Set the weighting in Cell H8 to 100% and the values in cells K8 and M8 to 0%;
4) Click on the “Sort by Overall Rank” button (Top Right of Ranking Sheet).
Rules:
Buy the Top Ranked asset if ROC1 is greater than zero. (Lowell is not quite right here – ROC1 is the measure of Absolute Momentum – not the value in the Absolute acceleration column). The ETF at the Top of the List is the ETF with the highest Relative momentum – as stated.
If the Top ranked asset has negative Absolute Momentum (ROC1 value) check the next highest ranked asset until you find one that does;
If no ETF ranks higher than SHY AND has positive momentum – go to Cash/SHY.
[Note: since the “normal” correlation between equities and bonds is expected to be negative, if equities are showing negative absolute momentum (ROC1 negative) we would expect TLT to show positive absolute momentum (ROC1 positive). However, this is not always the case and bonds (especially TLT) sometimes move in the same direction. Requiring Absolute Momentum to be positive for any asset should keep us out of major trouble]
David
David,
I added another another screen shot to the post. I think it does what you laid out above. I used VEU as it includes some emerging market equity stocks.
Lowell
Richard,
Your suggestion is far more sophisticated 🙂 but that’s essentially what we do with most of these portfolios – we just make it a little more complicated by adding more ETFs and (usually) selecting more than 1 in which to invest (people get nervous only picking 1 😉
Your statement regarding the acceleration numbers is correct – but Antonacci doesn’t mention/use this – it’s exclusive to this site 🙂 lol
David
David, Thanks for your reply. I have done some back testing on a similar strategy without an acceleration filter that looks very robust. My feeling is that adding the acceleration piece would make me a little less nervous. With that said, this portfolio has plenty of versatility, would be simple to manage and may result in less trades and less trading friction than a portfolio with more holdings. Hope so because I plan to use it with real money.
Richard
Lowell, I don’t know if you miss that part in the book, and I’m not saying that it’s right or wrong (let me know what you think). But Gary was specially leaving emerging Market out of his portfolio because he was saying that EM was not offering much diversification (everything being more correlated now), and was adding volatility. This is why he was using VEA instead of VEU?
Martin
Martin,
You are right. I see where I highlighted the last sentence on page 52. “We could easily drop emerging markets entirely from our dual momentum portfolio without a significant dilution in our results.”
Just for fun I checked EFA and EEM, developed international and emerging markets respectively, in the Schrodinger. I’ve held these ETFs for nearly 14 years. While the portfolio is up 5.8% annualized, EFA is down 0.1% (-0.1%) while emerging markets EEM is up 5.5%.
Lowell
Lowell,
5.5% for 14 years? The questioning is: what will emerging market will do in the coming years??? 🙂
Personally I still have some in my portfolios, I’m just including them when they’re above there EMA’s……(usually less volatile)
Martin,
Yes, that is the annualized figure. VTSMX was up an annualized 5.8% so there is not a huge difference. This includes two major bear markets. The Dual Momentum model would have made a significant difference.
Lowell
Lowell,
I agree, Thanks
Should we use VEU or VEA to cover international equities when applying the Dual Momentum strategy. Antonacci is not consistent in his writings. At one place he says we can neglect emerging markets while on page 97 he advocates using MSCI All Country World Index (ACWI) which includes 24 developed markets and 21 emerging markets. We are better off using VEA and VWO as both carry lower expense ratios vs. ACWI which is rather expensive.
This debate is not an issue for those of us using the “Rutherford 10” plus SHY as that list includes both VEA and VWO.
Lowell
“The look-back period recommended by Antonacci is 12 months. Here at ITA we use a 91- and 182-day combination.”
Please direct me to a link if my following question has already been answered, but I’m curious why ITA deviates from Antonacci’s 12-month look-back period by using much shorter 3-month and 6-month look-back periods?
On page 21 of Antonacci’s book it states:
“Jegadeesh and Titman’s research clearly showed that stocks strongest over a 3- to 12-month look-back period are also the strongest ones over comparable future periods. This was especially true with respect to a 6-12-month look-back window.”
On page 94 it states:
“The majority of academic literature covering both relative and absolute momentum agrees that a 12-month look-back period gives the best performance.”
On page 105, table 8.5 shows that a 12-month look-back period for Global Equities Momentum generates much better results than 9, 6, or 3-month look-back periods.
Thank you for any guidance,
Jim
Jim,
Thanks for your comment. If you are wondering why the delay, I need to approve the first message of any user as this cuts down on rogue comments. Your next question will show up immediately unless it is captured by my spam filter.
I’ll start the answer, but Ernie and David will likely respond as well since they have both done quite a bit of back-testing. The longer one waits or extends the look-back time frame, the greater the draw-down. Shortening the look-back period lowers the volatility.
The three- and six-months periods were selected as they seemed reasonable without going into a “curve fitting” or data mining operation to determine the very best look-back period.
David and Ernie – If your data shows something else, please correct my answer above.
David did an 8-year backtest between 2006-2014 that showed slight outperformance of the 3/6 hybrid lookback period compared to Antonacci’s 12-month lookback period in the 8 asset model:
https://itawealth.com/2014/11/12/dual-momentum-back-tests-part-2-adding-diversification-dual-momentum-strategy/
In the 3 asset model, however, the 3/6 hybrid lookback significantly outperformed the 12-month lookback:
https://itawealth.com/2014/11/10/dual-momentum-back-tests-part-1/
But I also note David’s caveats:
1. A backtest based on 40 years of data (1974 – 2013) is more robust and sustainable than a backtest based on only 8 years of data (2006-2014).
2. The 6 month (100% weighted) and 3/6 month (50%/30% weighted – 20% Volatility) options obviously involve more transactions, and associated costs (including more realized short-term taxable events).
3. David would rather use a system that had a high probability of generating a 13% annual return over any 8 year period than a system that might generate an average return of 17% over 40 years but that might have some 8 year periods with returns of only 6%. BUT — is there any reason to believe that the superior performance of the hybrid 3/6 lookback in the most recent 8-year period is sustainable or is it simply data mining?
4. One can usually look at shorter amounts of data and find a better parameter than the one that has worked best over the long run. The problem is that the best parameter over one short period of time may not hold up during subsequent short periods of time.
Best,
Jim
Jim,
The tax issue is a serious one. According to my calculations, the momentum model must generate about 2% higher than a passive approach if one is churning each ETF before one year passes.
As for Antonacci’s 40 year study – I question how he can come up with his results as the ETFs specified in the GEM model do not have anything close to a 40-year record. Even if one can come up with equivalent fund, would an investor have known to pick those index funds 40 years ago? Just another uncertainty tied to back-testing.
Lowell
Hi James, David, and Lowell,
I don’t think you can take the “best look back” question in isolation. David and Lowell may not agree, but I think it depends on the update period one is using to check and update the portfolio. This also ties with the tax question.
One of the reasons why I like the 33-day checkup period is it gives me some comfort that I will be able to catch a downturn soon enough to avoid taking the assets of the portfolio through a major bear episode. Of course, crashes can occur over a few days, but most of these turn out to be noise in the market and recovery occurs quickly. Doing a 33-day check period means one is going to forego capital gains tax rates on most trades. If one puts a high value on capital gains, then set your check period to one year, realizing that in doing so you may ride the beginning of a bear episode longer than you would like.
If one uses the 33-day checkup period, then the momentum one seeks is the momentum that provides the greatest likelihood that an asset will continue positive results over the next 33 days. That may well mean that shorter look back periods are more appropriate for the 33-day checkup period than would be appropriate for a longer checkup period, say, 90 or 180 days. There is at least one study I have read that confirms this. I do not remember the checkup period for Antonacci’s study.
Also, on the matter of how many assets to fund in the selected portfolio, my back test studies show that one trades off lower returns for reduced volatility when more than one asset is included (tested for the 33-day checkup period). That is what one would expect.
Those are my thoughts.