Dual Momentum: 6 Month or 12 Month Lookback

Dual Momentum, discovered and popularized by Gary Antonacci, is one of the investment strategies that I endorse and personally use.

Over time, Dual Momentum has performed extremely well compared with simple buy-and-hold, lazy style portfolios. It has also outperformed most complex investment strategies and most active managers of mutual funds.

The best part about Dual Momentum is its ease of management for any amateur investor. Dual Momentum hardly requires any work at all.

Once a month you evaluate three different investable asset classes, find the highest performing of those three asset classes over a specified historical time period, and invest your entire portfolio in that asset class.

These are very broad parameters, but are based on verifiable academic research. Dual Momentum is branded as such because it evaluates two simple forms of momentum: relative momentum and absolute momentum.

Relative momentum is the comparison of two assets within the same broad asset category. For the purposes of Dual Momentum, our relative momentum category is the equity universe: the U.S. stock index compared to the International stock index.

Relative momentum evaluations are not exclusive to broad stock indices. For example, if you compare the recent historical performance of Apple, Inc. (AAPL) and Microsoft (MSFT), you would also be conducting a relative momentum evaluation within the large-cap tech stock asset category.

Absolute momentum evaluations are the comparison of any asset class to a risk-free asset, usually represented by Treasury bills. In Dual Momentum, we compare our relative momentum evaluation winner to the recent historical performance of short-term government bonds.

Once you have determined the preferred asset after your relative momentum and absolute momentum evaluations, you would ensure your entire portfolio is invested in that preferred asset. If you already hold that asset (which is the case the majority of the time), you would do nothing that month. If you hold a different asset, you would sell the asset you have and purchase the asset dictated by your Dual Momentum evaluations.

Dual Momentum Variables

In Dual Momentum there are two main variables to consider that will impact your strategy in a meaningful way.

The first is choosing the lookback period you will use to determine your momentum evaluations. We'll cover a bit of that in this post.

The second is choosing the number and type of investment assets you will track. I have talked about that a bit in the past, but will do a new post in the future.

For the purposes of this post, we'll maintain the standard asset classes that Gary Antonacci has set out: U.S. stocks vs. International stocks, and the winner of that evaluation goes up against Treasury bills.

It's worth pointing out Antonacci introduces a small twist on the bond part. While he uses short-term government bonds (T-bills) when conducting his absolute momentum evaluation, he actually invests in the broad bond universe when the signal is in the risk-free asset. This small change increases overall returns over time.

Choosing Lookback Periods

One variable that is always up for debate with any momentum analysis is choosing the best lookback period to use when evaluating historical performance.

Gary Antonacci advocates a 12-month lookback period when sharing his strategy with the investing public. However, he also has made it clear he offers more responsive programs to institutional investors which purchase his services as a investment consultant.

Academic papers have evaluated lookback periods quite intensively. While the exact conclusion can depend on the asset being evaluated and the time frame used for the academic analysis, it is generally agreed that a lookback period between 3 months and 12 months can offer a momentum based advantage for investors. This is to say, assets which have outperformed in the past three to twelve months continue outperforming relative to other assets for some future time.

I don't disagree with Gary's public assertion to just go with 12 months because it is generally good. Something like choosing a time frame for performance evaluation can become a bit of a cherry picking exercise.

Maybe the ideal is 6.5 months for U.S. stocks between 1970 and 1996 and 11 months from 1996 until today. International stocks might be 10 months until 1988 when Emerging markets were added and just 5 months since that time. I don't know... but you get my drift here.

At the end of the day, you need to choose a standard metric that works generally well across all assets being evaluated.

Shorter Lookback Period

There are a few benefits to using a shorter lookback period. A short period will translate to a more responsive system. The signals are changing much quicker from month to month. This means you will have more trades.

At some point, the investor using short lookback periods will end up actually losing their performance advantage thanks to whip-saw trading. You are in one asset, get bounced out of that holding after a bad month, during the month you are out the first asset does very well compared to your new holding, and at the end of the month your evaluations put you back into the first asset again. If you would have simply held on, you would be better off.

For example, using a 6-month lookback since 1970, a simple Dual Momentum system would have counted a little over 100 trades. The largest drawdown would have occurred in the 1987 period with a -22.3% drop on a monthly calculated basis.

Longer Lookback Period

Using a longer lookback period, your system will be less responsive to changes in the market which occur in more recent months. The signals will change much slower, resulting in fewer trades.

This has pros and cons. Academic studies show that markets have mean reversion characteristics in short time-frames and over very long time periods.

Longer lookback periods can reduce disruption caused by short term mean reversions.

However, a drawback of a longer lookback period will occur when the market has a fast run up quickly followed by a steady fall down over a period of several months. You may be in a trade through the worst drawdown periods, only to get signaled out of the asset when it is recovering.

This was the case in 1987 where Dual Momentum would have moved you out of equities for a few months in the beginning of 1988, way to late to avoid the October/November Crash. By the time you were cycled out of equities in that fast crash, the market was in full recovery causing more portfolio damage.

If you use a 12-month lookback system since 1970, Dual Momentum would have counted fewer than sixty trades in 48 years. The largest drawdown on a monthly basis would have occurred in the 1973-1974 bear market with a loss just under 20%.

Time Averaged Dual Momentum - Combining Lookback Periods

For my own evaluation, I use both the 6-month and 12-month historical performance of each asset class by averaging the two period performances.

Using performance data since 1970, this strategy results in just four more trades being made over those 48 years when comparing to a simple 12-month lookback.

When comparing the combined lookback system's trades to the simple 12-month system's trades, the combined strategy is clearly more responsive. The combined system will move to a new asset class up to four months earlier than the 12-month system.

This small difference helps add nearly 0.5% per year in compounded annual returns.

The combined system also showed a smaller drawdown in nearly all drawdown situations over 10%.


When making a decision about lookback periods, I believe it is important not to overthink the process. You can do that by keeping your investing goals in mind.

By and large, a 12-month lookback period is fine. It gets you out of the market for those long painful bear markets, exactly like the kind we've experienced in 2000-2003 and 2007-2009 and will experience again at some point in the future.

It's easy to debate a half percentage here, or one percent there to find the "best" system out there. It all becomes largely irrelevant when the system is doing what it should: getting you out of a bear market at a 10% or 15% drawdown and helping you avoid the nasty 30% or 50% drawdowns that make serious dents in your financial situation.

The system I use is my Time Averaged Dual Momentum model--the system with a combined 6-month and 12-month lookback. Time Averaged Dual Momentum is easy to calculate, it is responsive, it has performed great over time, and there is no significant difference in the number of trades executed. Until the evidence demonstrates otherwise, I will predominantly share this system on the blog.

Dual Momentum based on longer lookback periods (combined or not) works well. It handily beats the S&P 500 over full market cycles, it beats passive buy-and-hold portfolios, it significant reduces drawdowns on your portfolio, it is very low cost, and it is an easy system to run in any self-directed portfolio.

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.

14 Replies to “Dual Momentum: 6 Month or 12 Month Lookback”

  1. Thanks for the excellent article- I’ve been reading allot about this strategy and find all of this info very interesting.

    Few questions:
    1- I can see that you’re investing in unhedged CAD ETF’s for this strategy. When comparing relative and absolute performance, are you using USD ETF’s or unhedged Canadian ETF’s? Basically, are you including fluctuations of currency in your performance comparisons, or just comparing ETF’s priced in USD, as Gary did.
    2- Do you know of an automated (or semi automated) way of calculating the 6/9 month performance every month?
    3- Do you think this strategy is still effective in a registered account? Would be subject to capital gains more frequently with more trades.


    1. Daren (Editor) says:

      Thanks for reading! If you haven’t already, reading Gary’s book and blog is a great way to understand the strategy, why it works, and where the downsides are.
      1) It should not make a significant difference if you measure performance off Canadian listed unhedged ETFs, or if you use US ETFs/mutual funds. In Gary’s evaluation he does not use a hedged version of the International stocks to strike the currency factor on the US side. This means both are evaluated absent currency adjustments in US dollars. I think the bigger issue that may slightly change some of the signals is the difference in cost between Canadian listed international stock ETFs and US listed international stock ETFs. Right now the cheapest international stock ETF in the US is a full 12 basis points cheaper than the comparable Canadian version.
      2) I just use an Excel spreadsheet formula to average the 6-month and 12-month returns. It’s easy and takes me a minute or less.
      3) I am currently using a version of Dual Momentum in my registered accounts (RRSP & TFSAs). I think the strategy will work well on a risk adjusted basis in any account. Some capital gains will be triggered in a non registered account, but the impact of that is quite minimal as capital gains taxes here are quite low unless you have a very high income. Thankfully we don’t have the US issue where short-term capital gains are taxed at much higher rates than long-term capital gains.

  2. What is the maximum drawdown of Dual Momentum while averaging the 6 and 12 month loopback periods?

    1. Daren (Editor) says:

      *Corrected in edit* The largest drawdown since 1970 (as far back as I can get International stock data) occurs from April 1973 to September 1975. If you had invested in T-bills as per the signals, the drawdown would be -22%. If you had invested in 10-year Treasury bonds, the drawdown would have been -24%.
      This compares with a similar -24% drawdown in the same time period using a 12-month only lookback (with 10 year Treasury bonds).

  3. Hello,
    thank you for sharing your improvements on the strategy. Why donĀ“t you use the AGG Bonds ETF instead of the 10Y Treasury Bonds? Do they have a better performance for the strategy? or are there any other reasons for choosing them.


    1. Daren (Editor) says:

      I use 10 year U.S. Treasuries to simulate bond returns in my model when the stock assets do not have positive absolute momentum. This is mainly because I don’t have access to monthly data for an AGG style bond fund going back to 1970.
      Personally I’m agnostic as to which fund the end investor chooses for their portfolio.

  4. Randy Harris says: Reply

    Daren, I have changed my personal implementation of Dual Momentum from one of scaling in and out to instead using a blended weighting of lookback periods. Unlike your blend of 6 month and 12 mont I have opted to go with 25% 1 month weighting, 25% 3 month weighting, and 50% 6 month weighting.

    It is far more responsive, yes, has far more trades, it also has less negative years and more shallow drawdowns.

    You can see my implementation here:


    Keep on doing a great job!

    1. Daren (Editor) says:

      Thanks for sharing Randy. I’ve actually gone to a 12-month look back only since this post. http://therichmoose.com/post20181030/
      Using your shorter system resulted in more than 2x the number of trades while the overall worst drawdown was just moderately better. I’m not sure if that makes a lot of sense to me, but I do understand the comfort in careful alignment with market moves.

    2. Randy Harris says:

      Daren, give my numbers a try, 1, 3, 6 month at 25% 25% and 50% back to 1970. I don’t easily have access to pre-2003 data for my model – but it far exceeds the 12 month look back period. Happy to do this in my retirement account where more trades don’t affect me negatively.

    3. Randy Harris says:

      As I see it, going back to 2003:

      12 month look back from 2003 – 2018 with 12 month returns 16.74% CAGR, -17.31 max drawdown
      1/3/6 look back from same time period returns 11.73% CAGR, -19.28 max drawdown

      Everything else remains the same, only changing look back period.

    4. Randy Harris says:

      sorry, I labeled those backwards! 12 mo look back returns 11.73, the 1/3/6 look back returns 16.74

    5. Daren (Editor) says:

      I ran a backtest to 1970. You are correct that the performance is good (16.8 CAGR). The 1/3/6 system weighted as you proposed outperformed 12M Dual Momentum in several periods (the 1973/1974 crash and 2009-2012 period) by getting out of the market earlier and back into the market earlier. However, 12M outperformed from 1975-2000 and 2013-2018.
      1/3/6M ended up with a better best calendar year and worse worst calendar year. The largest drawdown was quite a bit smaller than the 12M system (-16.7 percent vs. -25.4 percent). Overall, I would say the 1/3/6M system appears to perform well given it’s overall risk. It’s important to note the 1/3/6M system had 2.5x the number of trades over the total period. It’s probably a viable system in a registered portfolio with low trading costs.
      *I ran the test as (0.25*1M + 0.25*3M + 0.5*6M) and didn’t try other weightings to compare.*

  5. Ron Compton says: Reply

    Over a shorter time period, the Dual Momentum approach seems to give much lower return than the long term CAGR. As I see it, if the Dual Momentum approach had been adopted five years ago, for example, the CAGR would be only 6.6%. Is the Dual Momentum performance degrading in recent times?
    This is of particular interest to me because I’m retired and have only a 10-15 year investing time frame.

    1. Daren (Editor) says:

      That’s correct, current returns are much lower than the long term average.
      5-yr rolling returns have been as low as 3 percent CAGR and as high as 45 percent CAGR going back to 1970. My post tomorrow will demonstrate this.
      I don’t think DM is degrading. It is currently going through a period of returns on the lower side. However, we’ve been here before in the late-1970s and early-2010s.

Leave a Reply

sixty eight + = seventy five