Dual Momentum: Evaluating the Bonds Component

This is the final post in a three post series of evaluating each signal within the Dual Momentum model: "U.S Stocks", "International Stocks", and now finally "Bonds".

In the Dual Momentum strategy, which I share each month with the blog readers, I use a customized 3-month Treasury bill rollover index to analyze 12-month performance of cash (the absolute momentum hurdle).

To calculate the past 12-month return of T-bills, I mimic what an investor would have achieved for a gross return if they had purchased a 3-month T-bill a year ago and rolled it into a new 3-month T-bill each time the T-bill expired, creating a compound return.

For example, if I was evaluating the cash return on November 30, 2018, I would simulate the following:

  1. Buys 3-month T-bill on December 1, 2017 and holds until maturity; then
  2. Buys 3-month T-bill on March 1, 2018 and holds until maturity; then
  3. Buys 3-month T-bill on June 1, 2018 and holds until maturity; and finally
  4. Buys 3-month T-bill on September 1, 2018 and holds until maturity.

At each T-bill rollover, the investor would invest their original investment and the interest they earned from the prior T-bills to generate a compound return of the four T-bills. The purpose is to mimic the 12-month return of holding the safest asset currently possible.

While I use the 3-month Treasury bill rollover index to generate the signal, each investor must choose a fund they will actually invest in when the signal is in "Bonds".

I do not recommend buying T-bills as you may need to hold them for less than 90 days, although you could buy a money market fund or cashable GIC if you don't want to use bond ETFs.

When it comes to bonds, there are almost countless choices for investors to implement this signal their portfolios.

Some of the most common bond fund types which are available in a low-cost ETF format include:

  • Aggregate bond funds—invests in all investment grade bonds of all maturities in a market weighted format
  • Short-term bond funds—invests in investment grade bonds up to 5 year maturities in a market weighted format
  • Government bond funds—invests in only government issued bonds
  • Corporate bond funds—invests in only corporate investment grade bonds

There are endless variants of the above, including corporate junk bonds, short-term government bonds, short-term corporate bonds, intermediate-term government bonds, intermediate-term corporate bonds, government TIPS bonds, international bonds, green bonds, etc.

Given all these options available, most at a relatively low cost, what should investors choose when the signal is in "Bonds"?

Aggregate Bond Funds

Gary Antonacci is very clear that he prefers using a low-cost aggregate bond index fund when the signal is in "Bonds".

Gary argues that aggregate bonds represent a safety asset that benefits from investors' shift from stocks to bonds during a stock bear market. As well, they benefit from loosening monetary conditions as central banks attempt to revive the business cycle.

Aggregate bond funds are highly liquid and available at extremely low management fees of 0.05 percent. They consist 70 percent AAA rated bonds, including government Treasury bonds, with the remainder from large corporate issuers.

Their average duration is approximately 6 years, pulled down mostly by the tilt to generally shorter duration on corporate bonds.

Given the inclusion of corporate bonds and long-term Treasury bonds, aggregate bond funds will see somewhat more volatility than a pure short duration bond fund.

Short-term Bond Funds

Short-term bond funds are much more stable than aggregate bond funds as they do not suffer from duration risks. These funds will not hold bonds with maturities longer than 5 years while the average duration is just 2.5 years.

The short-term bond funds typically have somewhat fewer government bonds as a percentage of total holdings than the aggregate bond funds. This is because many corporations issue short-term bonds.

Just 60 percent of bonds in short-term bond index funds are AAA rated bonds.

Short-term bond funds are available at extremely low management fees of approximately 0.06 percent. Like aggregate bond funds, they are very common from several low-cost providers and often highly liquid.

Given their short average duration, short-term bond funds are normally more stable and benefit less from loosening monetary conditions when compared with aggregate bond funds.

Government Bonds

Government bonds, in the context of U.S. Treasury bonds or even Canadian government bonds, are the most stable and secure form of credit investment.

Governments also issue the most investment grade bonds on the market. Funds holding government bonds are often highly liquid and available at a low cost as it is easy manage a fund with only one credit issuer.

Government bonds are issued across the entire duration curve, ranging from 1 month bills to 50 year bonds (Canada). Most commonly, the U.S. treasury issues 3-month Treasury bills through 30-year Treasury bonds.

Although there is inflation risk due to a floating fiat currency system, a bond holder of U.S. government debt is certainly going to be repaid. It would not make sense for the government to declare bankruptcy on their debt when they can just print more dollars to pay it off.

Corporate Bonds

Corporate debt securities range from AAA rated bonds through to unmarketable junk bonds. Investment grade credit, rated above BBB (Standard & Poors), is the most common corporate bonds for index ETF investors.

Corporate bonds suffer from inflation risk like government bonds, but they also suffer from default risk. If the government prints dollars and stokes inflation, corporations will benefit as their debt value decreases compared to asset value.

At the same time, if the corporation runs out of money, they will declare bankruptcy. This could be a near total default, or a partial restructuring where bond holders often take large losses.

The risk for corporate bond funds is higher than government bonds, but bond funds are well diversified so isolated bankruptcies are not likely to cause significant losses for corporate bond fund investors.

Due to the default risk on corporate debt, investors should avoid investing in long-term corporate bonds.

Junk bond ETFs do exist and are widely available; however, they are much more correlated to stock market performance than they are to bonds, especially in crisis periods. For this reason, you should avoid junk bonds in Dual Momentum!

Comparing Bond Options in Dual Momentum

Before we recommend any one option over another, we will evaluate the historical performance of several major bond fund categories when used in the Dual Momentum model.

Government Bonds

Due to a data issue, I will start with a comparison of short-term government bonds, intermediate-term government bonds, and long-term government bonds since 1978.

For these backtests, I maintained the exact same signals using the MSCI indices which I use for all my Dual Momentum analysis.

When the base signal was in "Bonds", I put the portfolio either into short-term, intermediate-term, or long-term government bonds for each time period. This in effect isolates the performance of each option so you can see how it works in the Dual Momentum model.

The backtest runs from January 1978 until October 2018.

Sources: TheRichMoose.com, MSCI Inc., FRED Federal Reserve St. Louis, Vanguard Funds

In this backtest, long-term treasuries very clearly outperformed. The short-term bond option generated a 16.51 percent compounded annual return over the 40 year test period while the long-term bond option generated a 18.04 percent compounded annual return.

Clearly the intermediate bond option (average duration of 5 years) and the 10-year bond option were in the middle. You can see the month-to-month volatility on the 10-year line and it is a good reflection—amplified or reduced—of the monthly changes of the long-term, intermediate, and short-term bonds.

In my backtest, the long-term bonds were the best holding in nearly every single period where the signal was in "Bonds". Although long-term Treasury bonds were more volatile within the holding periods, the end result was repeatedly better.

Corporate Bonds and Gold

In this next backtest, I will compare short-term corporate bonds, aggregate bonds, and gold (often cited as a crisis asset) from 1987. The time period is shorter, again due to limited available data.

For these backtests, I maintained the exact same signals using the MSCI indices which I use for all my Dual Momentum analysis.

When the base signal was in "Bonds", I put the portfolio either into short-term corporate bonds, aggregate bonds, or gold for each time period. This in effect isolates the performance of each option so you can see how it works in the Dual Momentum model.

The backtest runs from January 1987 until October 2018.

Sources: TheRichMoose.com, MSCI Inc., FRED Federal Reserve St. Louis, Vanguard, London Bullion Exchange

In this comparison, the results for 10-Year Treasury bonds (the standard in my model), short-term corporate bonds, and aggregate bonds were very similar.

Gold underperformed significantly over most of the period, but made for lost ground in the 2000-2003 and 2008-2009 holding periods.

The top performer, 10-Year Treasury bonds, returned 14.62 percent compounded annually in this backtest while the bottom performer, short-term corporate bonds returned 14.24 percent compounded annually.

Although Gary Antonacci shares his preference for holding a low-cost aggregate bond fund when the signal is in "Bonds", 10-year Treasury bonds performed slightly better in nearly every holding period.

This may be a reflection of Treasury's preferred safety during crisis events and the slightly longer average bond duration.


At the end of the day, the asset of choice for the end investor when the signal is in "Bonds" is largely meaningless when comparing the major options: U.S. Aggregate Bond funds, Intermediate Treasury Bond funds, 10-Year Treasury Bond funds, or Long-term Treasury Bond funds.

If the Dual Momentum investor wanted to pursue the highest possible returns with the increased risk, Long-term Treasury Bond funds would be the better choice.

I believe U.S. Aggregate Bond funds, 10-Year Treasury Bond funds, or Intermediate Treasury Bond funds are all good choices for the typical Dual Momentum investor. Short-term bond funds are probably unnecessarily safe for most investors.

As always, choose a low cost option that is highly liquid to minimize drags in returns caused by management fees and bid-ask spreads.

In Canada, this leaves us with three viable options to invest when the Dual Momentum signal is in "Bonds":

  1. We could choose XBB.TO/VAB.TO/ZAG.TO which invest in the aggregate Canadian bond market at 0.08 MER; or
  2. We could choose ZDB.TO or HBB.TO which are more tax efficient choices of the above at 0.09 MER; or
  3. We could convert our accounts to U.S. dollars and invest in a number of ETFs that hold aggregate bonds, long-term Treasury bonds, or intermediate Treasury bonds all for under 0.07 MER.

Comments & Questions

All comments are moderated before being posted for public viewing. Please don't send in multiple comments if yours doesn't appear right away. It can take up to 24 hours before comments are posted.

Comments containing links or "trolling" will not be posted. Comments with profane language or those which reveal personal information will be edited by moderator.

12 Replies to “Dual Momentum: Evaluating the Bonds Component”

  1. Thanks for the analysis, this is a timely post since DM may soon switch to bonds. For a future post, I would be curious whether there is any significant difference between the choice of Canadian, US or global bonds and whether currency hedging bonds matters for Canadian DM investors.

    I think Vanguard’s all-in-one funds don’t currency hedge equities but they do currency hedge their global bonds.

  2. Is it sensible to use Relative Momentum (RM) to determine what bond fund to use? If we use relative momentum for the equity portion; why not do something similar for the bonds? I am not suggesting to switch between bond funds back and forth while the signal is in bonds. My thinking is that upon making the switch from equities to bonds we could use RM to determine which bond fund to use and remain in that until we are taken into equities. I am not sure what look back period to use in this case.

  3. If you were in US dollars and in a non-registered account, what etfs would you consider in this case?

  4. Daren (Editor) says:

    I believe that Schwab, iShares Core, and Vanguard have some of the better low cost options in each bond category.

  5. Daren (Editor) says:

    Relative momentum doesn’t seem to work as well in the bonds I mention (provided you are not including Corporate Bonds and Junk Bonds which are much more correlated to stocks). When looking at Short-term Bonds, Aggregate Bonds, and Treasury Bonds, the performance is most directly related to average coupon duration. Relative momentum, when comparing the major differences of Short-term bonds vs Long-term bonds actually underperforms an equal weight holding using almost any lookback period.
    If you hold an Aggregate Bond fund, it performs better than a Short-term Bond fund. If you have a 50/50 allocation to Short-term bonds and Long-term bonds and re-balance annually or semi-annually, you will slightly outperform Aggregate and Intermediate term bonds in most time periods.

  6. Daren (Editor) says:

    There are sound reasons to hedge international bonds, particularly in emerging markets. If you don’t hedge bonds, they tend to become a currency play. That said, long-term results show that returns are the same so the differences are mostly in shorter time frames where currencies can swing much more.
    Gogi Grewal’s research shows that investors are best to hold U.S. bonds in U.S. currency. The difference between that option (Case 1 in his post) and the local bond option (Case 4 in his post) are really minor. Either way you are probably good. https://www.dualmomentum.net/2015/06/dual-momentum-for-non-us-investors.html

  7. Yanniel says:

    Daren, I see your point about RM and bonds. I did not know that. Thanks.

    What about picking short vs long term bonds depending on the yield spread? I’ve read (but I haven’t estimated this myself) that a yield spread of 150 bps or more between the 2 Year Gov Canada Bond and 10 Year Gov Canada bond represents a sweet spot to transition from short to long. I realize that this is a “fundamental way” to look at this which departs from the technical approach and trend following in general.

  8. Daren (Editor) says:

    I don’t have a lot of knowledge looking at bond investments in this way. It does raise some questions though: is the article advocating 150 bps as a switch point going both ways? Ie. you have a clear entry signal and clear exit signal? What was the historical performance and what time frame was used? How would you manage this within the DM bond signal?
    In the backtests I’ve done, I see a clear advantage to long-term Treasuries vs. short-term Treasuries in nearly every DM bond holding period. That is with buying a single bond holding and not changing it during the holding period for any reason.

  9. Yanniel says:

    The article is somewhat vague. It just talks about the transition from short to long term bonds. It puts an example from 2010 to 2015; but handpicking a specific period is not proof of anything. Maybe they did a more exhaustive research but it is not spelled out in that piece.

    It also talks about how it makes sense to buy short term bonds when we are in a hiking interest rate environment. This is seems like a no-brainer to me.

    I have the link to the article. It is a market comentary from a financial firm. I won’t post it here because I am adhering to the comments guideline of this blog. It is not my intention to promote anything indirectly.

    I found another article somewhat related to this topic (by the Montreal Exchange). It is more elaborated and it looks at bond futures. It is a short read. I think naming it here won’t violate the commenting guidelines so here it goes: “Trading on the Yield Curve”. Please, delete it if not appropiate.

  10. Daren (Editor) says:

    Thanks for sharing the Bourse de Montreal article, here is the link for other readers: https://www.m-x.ca/f_publications_en/strat7_bondfutures_en.pdf
    I should point out that this short paper is a spread trade strategy which effectively bets on central bank actions.
    As a note, I don’t mind if comments contain links to articles which are more academic in nature. For example, AQR, Longboard, Alpha Architect, and Capital Fund Management have some great academic papers on the trend following strategy. Research Affiliates does great papers on factors.
    What I don’t want is Fred from Fred’s Gold Sellers coming into the comment section and posting a comment that says “Gold is going up, gold is the best, come see the 9 reasons why you should have gold in your portfolio: w.fredsellsgold.co/buygoldnow2018”

  11. Hey Daren,
    My first comment here. Love your articles.
    Yaniel brought up a good point. A few months ago I read a seeking alpha article by Gary Antonacci on Dual momentum for fixed income that he wrote in 2014. The backtest is from mid 80s but he uses Barclays’ indexes all the way rather than ETFs for recent years and indexes for the years when ETFs were not around. The backtest annual returns are not very high (somewhere in the vicinity of 11%) but the drawdowns are quite low. This means that the strategy can be leveraged a bit either on it’s own. In his book Gary also mentions a strategy he calls Global balanced momentum where a 30% FI allocation is maintained with the rest (70%) being allocated to GEM. The FI component uses this dual momentum for FI strategy and so does the FI part of the GEM which has the 70% allocation.
    However, I am a little suspect of the backtest in the SA article as it’s during the bond bull period. The GBM results are from 1974 though and the annual returns to max drawdown is better than GEM. Sharpe ratio is better too.

    Also, I’m curious to know how do you do the backtests. Do you use a service or do you have a setup of your own?

  12. Daren (Editor) says:

    I agree the data analysis is tough because it is almost entirely in a bond bull market. That said, we don’t know how bonds are going to perform going forward. For example, investors lost billions betting against bonds in Japan. Arguably Japan is a decade or two ahead of Western Europe and the USA in terms of demographics and debt pressures, so it gives us a rough picture.
    We’re probably near the tail end of the current business cycle. Historically the Fed and other CBs have drastically cut rates thereby boosting bond valuations when business cycles flip. It’s a ballsy bet to go against the grain on that. But it could happen…
    I have looked closely at Gary’s 30/70 model; I think it’s mostly directed to institutional investors who obsess over Sharpe, Sortino, Kurtosis, and these types of metrics. With individual investors who would otherwise invest 60/40 or something similar the model makes less sense.
    For the most part I use my own Excel spreadsheets and try scrounge for free raw data all over the web. I prefer using index data where I can because fees are changing drastically within the US and by product type, not to mention wildly different fees in other countries.

Comments are closed