Dual Momentum – The Famous 5 Portfolio

This article describes a Dual Momentum study over a multi-asset ETFs basket with a new attempt to improve this well-known investing style.

Dual Momentum strategies rely on two different very simple filters: absolute momentum and relative momentum.

Absolute momentum (rule 1) is a trend following filter used to switch any selected assets that have a negative excess return over the risk-free rate to cash while relative momentum (rule 2) criteria compares returns of a basket different assets through a ranking list to highlight those assets with the highest return.

I considered a time window of three months in my backtest, as it has shown the strongest results.

Rule 1: If the asset shows that the last three months return is higher than the risk-free rate then buy or keep the asset, if not just switch to cash.

Rule 2: Buy or keep the # assets with the highest three months return if Rule 1 is satisfied, otherwise just move or stay in cash.

I backtested the following portfolio of multi-assets ETFs going back to January 2006 and holding each time the two assets with the highest three months return (holding 2 assets instead of 1 improves the results). I considered the five most common stocks ETFs:

GLDSPDR Gold Shares ETF
VOVanguard Mid-Cap ETF
VBVanguard Small-Cap ETF
VTIVanguard Total Stock Market ETF
QQQPowerShares QQQ Trust ETF
SPYSPDR S&P 500 ETF
TLTiShares Barclays 20 Year Treasury Bond Fund ETF

wordpress post Untitled-1

PortfolioInitial BalanceFinal BalanceCAGRStd.Dev.Best YearWorst yearMax.
Drawndown
Timing Portfolio$10,000$54,48518.99%12.98%37.74%-7.01%-14.90%
Equal Weight
Portfolio
$10,000$23,3569.09%12.02%27.03%-22.06%-33.70%

Applying Dual Momentum strategy more doubles the Final Balance and CAGR and cut by half Max Drawdown of the Equal Weight Portfolio.

This improvement from the Equal Weight Portfolio can be clearly seen on the equity curve too:

Equal Weight Portfolio. Dual momentum

Sitting on cash while out of the market can obviously be a waste of time, and good results can be achieved simply by switching to SHY (iShares Barclays 1-3 Year Treasury Bond Fund) instead of cash.

Hereafter, the new performance metrics:

PortfolioInitial
Balance
Final
Balance
CAGRStd.Dev.Best yearWorst yearMax.
Drawdown
Timing Portfolio$10,000$56,07519.34%12.99%37.74%-6.74%-14.67%

GAGR is above 19% and maximum drawdown has been reduced below 15%. Equity curve follows:

drawdown below 15%. Equity curve. Dual momentum

Adding a safe instrument of the monetary market helped both rising return and cutting risk. I considered it “safe” due to the next multi-year coming Fed rising rates environment. With the duration limited to a maximum of three years, new higher rates can easily be exploited.

Of course, nothing can be taken for granted and that is why until Fed does not start the rising rates process another adjustment should be implemented to the Dual Momentum Countries Portfolio: using TLT (iShares Barclays 20 Year Treasury Bond Fund) instead of SHY. Results are the followings:

PortfolioInitial
Balance
Final
Balance
CAGRStd.Dev.Best yearWorst yearMax.
Drawdown
Timing Portfolio$10,000$67,34321.61%13.78%37.74%-5.18%-14.02%

GAGR rised well above 20%.

GAGR rised well above 20%. Dual momentum

To sum up, adding long term treasuries (TLT) might be perceived as an additional risk factor due to the coming Fed restrictive monetary policy, but as far as this has not occurred yet, a portfolio with a growth rate higher than 20% looks very attractive to investors.

The overall strategy performance should turn in favor of the short term treasuries (SHY) option as soon as rates begin to rise, in contrast to the last 15 years decreasing rates/zero rates/quantitative easings financial environment that led TLT-based strategy to higher returns.

That is why I shall expect a 20% returns range for the SHY-based strategy and a lower one for the TLT-based one in the coming years once rates start rising.

–By Marco Simioni of  Nightly Patterns

About the Author Marco Simioni

  • frank says:

    Marco, Jeff

    Can you give a example in excel, EasyLanguage on how you calculate absolute and dual momentum. Maybe attach a excel file or easylanguage statement.

    Thanks

    Frank

    • Frank,

      You may be more familiar with momentum as the ROC indicator, for you classical technical analysis people.

      Or, put another way, the cumulative return over the past three months.

  • So, why all of these separate equity assets? Why a three month rotation, as opposed to 2, or 4? Why go all-in on one security? What are the performances on momentum settings ranging from 1 to 12 months? How sensitive is the strategy to which day of the month it’s rebalanced on?

    -Ilya

    • Thank you for comments Ilya. I just wanted to run through big, medium and small cap stocks. 3 months shows the strongest results.
      I didn’t checked for the day of the months effect.

  • frank says:

    Thanks, Ilya
    I am trying to do this in excel.

    1.
    This is regular momentum in Easy Language.
    momentum = Momentum ( Close,90)
    ROC = RateofChange( Close, 90 )

    2.
    What will be the syntax of Dual Momentum

    3.
    What is the syntax of Absolute Momentum

    • Dual momentum strategy is really quite simple: every rebalancing period, rank assets by the way you measured momentum (3 month lookback, standard deviations above a moving average, etc.). Invest in the top N assets (sometimes, N is 1 for a “rotation” strategy), and invest anything that isn’t higher than the risk-free rate in the risk-free asset (usually something like SHY or BIL).

    • Rod says:

      Ilya

      Do your own work. He clearly tried 3 months, and compared it to 2 or 4. He is not going all-in on one security – he said he is using two.

  • Mitch says:

    It is actually not the ROC unless your dataset is dividend adjusted. You actual want the total return which included dividends.

    The rotation is monthly, not every three months. The Total return is 3 months, though the actual strategy recommends you use the 12 month total returns. 3 month only works better in recent history but when you go back decades not so much. you will also trade more with a shorter lookback. One of the benefits of not trading so much is that you will accrue more often than not capital gains instead of income as the trade holding time will be longer. You really need to look at after tax returns.

  • bernhard says:

    hi !

    what is if you only have one asset which has positive absolute return ? you than go 100% of your funds in that asset?
    or 50% in that asset and 50% in cash ?
    how did you do it in your backtest?

    thanks !

  • gregor says:

    I cannot reproduce the results. I use the following method: On the first trading day of the month, buy the top two of GLD, VO, VB, VTI, SPY, TLT on the basis of the returns for the three calendar months ending on the last trading day of the prior month. If any of the top two have negative return during the prior three months, replace it by cash.

    Here is what I get. Close, but not the same. Perhaps the author can clarify if he is using something different from what I outlined.

    2006 24.76%
    2007 22.80%
    2008 -8.61%
    2009 31.08%
    2010 23.82%
    2011 29.83%
    2012 10.25%
    2013 32.93%
    2014 13.67%
    2015 -8.86%

    CAGR 16.46%

  • I am trying to replicate this too and come up with much worse MDD. Is your ETF list correct? You call this the “The Famous 5” but have 7 ETFs in the list. Are GLD & TLT supposed to be on there? If I remove them, I get much closer results, but still not close enough for me.

    To calculate the 3 month return do you use 63 bars or compress the data to monthly bars and use 3 bars?

    Does your data include dividends?

    Thanks,
    Cesar

    • Marco says:

      Hi Cesar,
      Well, I call 5 the 5 stocks etfs, so not thinking at TLT and GLD.
      I used adjusted monthly close price.
      I used the 3 monthly bars.
      And the drawdowns are the monthly ones, not the daily ones.

  • Don says:

    Would it be reasonable to expect that in a rising interest-rate environment, the portfolio is likely to spend more time going forward (relative to the 2006-2015 backtest) in cash/SHY/TLT rather than the 4 stock ETFs (VO, VB, VTI, QQQ) and hence expect lower overall returns for this system?

  • Marco says:

    Hi Don,
    I think the time spent out of the market will be more or less the same. If the markets rally, momentum will follow them. We just need to choose one of the 3 cash/SHY or TLT. I think TLT to be too risky to be used as a cash proxy in such an environment. I would use SHY instead of cash.
    But I would still rely on TLT when he’s rising and momentum calls for it as an asset.

  • Ron says:

    Dear sir, author of the articles. You did not backtest anything yourself.

    You just used this website:

    http://www.portfoliovisualizer.com

    and you did not give them credit. This isn’t good and I ask Jeff to place a note on top of your article giving credit to the people of that website you used to present yourself as a quant to your audience. Probably Jeff was not aware of this failure to give credit where due.

  • Mark says:

    Ron–

    This would be a serious ethical problem for the author. He would also hardly be in a position to field questions from others (e.g. Gregor, Cesar) since he didn’t do the work himself!

    If you’re going to launch such an accusation, though, then you really should explain why you believe it.

    Mark

  • George says:

    I will say this Jeff, PortfolioVisualizer.com looks very useful and interesting!

  • Drew says:

    It appears when I run the same portfolio scenarios using Portfolio Visualizer (which by the way is an excellent tool so thank you for suggesting it) I am able to reproduce the results fairly closely but what concerns me is the recent performance of the Dual Momentum strategy. Almost all derivations have performed poorly over the last 12-18 months and would have been down 5-10% over that period. Has anyone found a derivation of the system that has both performed well historically and in the current time frame?

  • Chris says:

    Following this strategy will have you owning two ETFs at any given time. If the signals indicate that you change one of them, do you switch only the one out of favor or do you rebalance the portfolio so that it is evenly divided between the two funds.

  • >