Two Faced investing

4 Flares Twitter 1 Facebook 3 Filament.io 4 Flares ×

This is my final post in my “cowards” investment philosophy series.*

You will remember that the central thread tying this series together is the idea that when it comes to investing, not losing money ends up being far more powerful than making extra money in the long term.

To this end we have discussed aggressive diversification, risk barbelling, value investing, and trend following.

And today we get to my chosen strategy, dual momentum.

So what is dual momentum?

Dual momentum is a philosophy that uses two forms of momentum to increase returns in bull markets, and more importantly to decrease losses in bear markets.

The most widely appreciated form of momentum is “relative momentum,” which compares the performance of a number of assets, and invests in the one that has performed the best (in total returns) over a pre specified look back period. Such an approach capitalizes on the enduring observation that assets which have performed well in the near to medium term past, have a tendency to continue to do well in the near future.

Relative momentum has been shown to improve returns significantly in almost every type of market and every time period studied, going back to the 1700s.  It is a robust pattern, referred to by Eugene Fama as “the premier market anomoly.”

But it’s not the sort of approach that minimizes losses. It’s more of a high risk high reward sort of a strategy.

To illustrate this, let’s look at a simple relative momentum strategy toggling monthly between emerging markets, US, and foreign developed markets based on a six-month look back period.

Back test rules:

  1. 3 index funds (VFINX,DFALX,DFEMX) are backtested on the last day of each month.
  2. Best performing fund based on a 6 month lookback period is bought on the first trading day of the next month and held for a month.

rel mo

Screenshot 2015-01-19 21.29.40

Screenshot 2015-01-19 21.30.54

So the relative momentum portfolio outperforms the 100% US stock portfolio, but it does so at the cost of higher volatility, and larger drawdowns.

The other (And lesser-known) type of momentum used in dual momentum investing is “absolute momentum” or “time series momentum.” This approach also looks back at asset total returns over a set time period, but it doesn’t compare two or three assets, it simply compares one asset to a risk free asset (namely cash or short-term treasuries) and stipulates that any asset that has not outperformed cash over the look back period, should be sold and held in cash instead, until it’s losing trend has reversed.  It treats the loser assets as the grimy outcasts that evidence suggests they will continue to be in the near term.

loser

What smart investors think about assets with negative absolute momentum…

Absolute momentum is a terrific cowards play. Instead of getting the investor into the best possible assets at the best times, it gets him out of risky assets at the worst times.

And it works pretty well.

Let’s backtest an absolute value approach investing only in the S&P 500, (or short-term treasuries) based on a six-month look back period.

Back test rules:

  1. 2 index funds (VFINX,VFISX) are backtested on the last day of each month.
  2. Best performing fund based on a 6 month lookback period is bought on the first trading day of the next month and held for a month.

Screenshot 2015-01-19 21.42.51

Screenshot 2015-01-19 21.43.49

Screenshot 2015-01-19 21.47.21

So these results are interesting, I think. The absolute momentum portfolio actually outperforms the S&P 500 by quite a bit. But it has much less volatility, and much smaller drawdowns. In fact, its max drawdown is even less than a 60/40 stock bond (“balanced”)portfolio. So this is really a low risk/high reward strategy.

This is actually a pretty similar strategy to trend following using the 200 day moving average that I previously wrote about here.

But absolute momentum has some distinct vantages over the moving average approach. In Gary Antonacci’s book Dual Momentum, he compares a 200 day moving average approach to absolute momentum and finds almost identical performance and Sharpe ratios (ie efficiency) looking back over a 39 year back test. But importantly, he finds that the moving average approach yields more trades (about 1.2 trades per year versus 0.83 trades per year using absolute momentum.) And in the real world more trades equal more costs both in the form of trading fees and in the form of the friction of bid/ask spreads. All things being equal, less trading is always better.

Dual momentum, as you’ve probably guessed by now simply combines relative and absolute momentum, to both harness the upside of relative momentum, and the downside protection of absolute momentum.

It sounds more complicated than it is. In fact to execute a dual momentum strategy, One need only start with a relative momentum strategy, and add short-term treasuries to the asssets in the lookback period. To illustrate this why don’t we add short term treasuries our initial backtest of emerging markets, and foreign and US developed markets.

Back test rules:

  1. 4 index funds (VFINX,DFALX,DFEMX,VFISX) are backtested on the last day of each month.
  2. Best performing fund based on a 6 month lookback period is bought on the first trading day of the next month and held for a month.

Screenshot 2015-01-19 21.55.35

Screenshot 2015-01-19 21.57.53

Screenshot 2015-01-19 21.59.00

If you’re anything like me you find that result impressive.

And aside from past performance (and the caveat that accompanies it) there are many factors that appeal to me about this investment philosophy. (Which is why I have adopted dual momentum in my tax sheltered accounts. )

  • Momentum is a bet on human irrationality, and a bet against that flavor of the efficient market hypothesis that posits that all individual humans are Homo Economicus capable only of making rational investment decisions. (I have yet to meet even one such human,so why would I bet on a collection of such humans driving the market?)
  • The dual momentum strategy is essentially agnostic. As an example my friend Robert and I have gone back-and-forth over and over about whether it is wiser to be internationally diversified or not. But in executing the dual momentum strategy ones own biases become unimportant. If  US equities have outperformed international equities (and cash) for the lookback period then the dual momentum investor will be 100% invested in US equities (and vice versa.)  This rules based approach acts as kind of an irrationality filter.
  • The dual momentum strategy is easy to implement , able to be executed using only low-cost index funds,and dependent upon only upon rare trading.

But the most important reason why I’ve chosen dual momentum as my go to strategy is simply this:

  • I am a investment coward. And this strategy both avoids large drawdowns (a.k.a. putting your principal at risk,) and beats the market most definitively when the market is at its worst.

*I reserve the right to come back and write about out of the money options or some such hedging maneuver at some point in the future.

4 Flares Twitter 1 Facebook 3 Filament.io 4 Flares ×

46 Responses to “Two Faced investing”

  1. Robert January 20, 2015 at 6:07 am #

    Good post. I’ll simply add your own concern that you failed to mention, which is the concern over tracking error. I personally am not too worried about it; I’ve been out of phase with the markets many times before and it doesn’t bother me too terribly much (though certainly some). Maybe its my contrarian personality. But in response, it would surely be useful for us to have discussions here at major turning points, when dual momentum signals to switch out of or back in to equities. I’ll probably be tempering my use of dual momentum with input from other technical indicators (I’m a huge fan of James Stack of Investech Research) and macroeconomic trends (I wouldn’t be switching into emerging markets, for instance, anytime soon, given current trends and expectations in currency markets).

    • Miles Dividend M.D. January 20, 2015 at 10:52 am #

      Tracking error is the issue with any active strategy. That is absolutely true.

      But I am suprised by your comment on “tempering,” if current trends or currency market expectations are expected to have a negative impact on EM performance wouldn’t this be reflected in the performance over the look back period? As an example, last year when the trend favored long term treasuries, that was avery uncomfortable trade. But momentum held (and that’s the point that matters.) Perhaps I am a momentum fundamentalist, but why not keep it simple and follow the trend blindly. Human irrationality is the commonest mistake for all investors.

      • Robert January 20, 2015 at 11:59 am #

        As I said, I have a huge respect for James Stack, and have been following his newsletter since 1992. He has mostly been correct in his calls, including moving his readers out of equities before the 1987 crash, avoiding the late 90’s tech bubble/bursting, and have greatly reduced exposure before the 2008 financial crisis. He helps manage exposure/risk over the business cycle. I also have learned a lot from John Mauldin on macroeconomic factors, which factor into this as well. Where I see myself using these two inputs is to compensate for DM’s failure in market spikes. In other words, if another 1987 crash type situation develops, I may be able to pick it up with these other indicators and get out before it happens, whereas DM would be too late. For more typical rounding tops, DM does a good job. That’s why I said “temper”; I see it as a small modification, but a significant one.

  2. PowerPlayer January 20, 2015 at 10:29 am #

    Maybe I’m missing something, but I see no references to either the absolute or relative momentum strategies or details of your own strategy. In principle what you say makes sense but the screenshots aren’t of tremendous value to me when the “timing portfolio” rules aren’t provided. Can you clarify?

    • Miles Dividend M.D. January 20, 2015 at 3:21 pm #

      PP,

      Thanks for your comment.

      That’s a valid criticism. I’ll ammend my post later with the rules for each of the backtests.

      AZ

      • PowerPlayer January 21, 2015 at 5:56 am #

        I see the updates but am still wondering….at least for the absolute momentum case, would you not go to cash if neither of the two funds are performing better than cash? Is this effectively VFISX in your example?

        Also, what are Balanced, Bond, and Stock Portfolio (at least in terms of ratios of stock to bond for the screenshot backtests)? 60/40 (VFINX/VFISX), 100% VFISX, and 100% VFINX?

        • Miles Dividend M.D. January 23, 2015 at 12:32 am #

          PP,

          Yes short term treasuries, (VFISX,SHY, etc) are “cash.”

          The benchmarks are the standard portfoliovisualizer.com selections.

          Stocks are total US stock market.
          Balanced are 60/40 “stock/bond”
          Bonds are 10 year US treasuries.

          these are asset class returns, not fund returns, which should favor the benchmarks since there are no expense ratios deducted from performance.

          AZ

          • PowerPlayer January 24, 2015 at 2:01 pm #

            Thanks for the info….very useful!

  3. Jonathan January 20, 2015 at 1:50 pm #

    Wow! Great article. Thank you for introducing this to me. I always thought being invested in low cost index funds with proper asset allocation was the way to go. Which funds are recommended for this strategy (S&P 500, total US market, especially what emerging market fund and international fund)? Also how often do you reassess and look to see if you should trade funds?

    • Miles Dividend M.D. January 20, 2015 at 4:15 pm #

      Jonathan, thanks.

      I added the backtesting rules for the post, so hopefully that helps.

      There is nothing wrong with a properly diversified passive buy and hold portfolio, (That’s my approach with my taxable accounts.)

      In terms of fund selection, whatever is cheapest available index ETF or mutual fund that covers the part of the market that you are interested in. The differences here are small potatoes.

      The funds in my backtests were chosen simply because they had the longest histories to test.

      I highly recomend Gary Antonacci’s book Dual Momentum, if you arre interested in momentum investing…

      AZ

    • Robert January 20, 2015 at 4:24 pm #

      Seems like it might be useful to refer readers to Gary Antonacci’s website: http://www.optimalmomentum.com/index.html

      • Miles Dividend M.D. January 20, 2015 at 5:02 pm #

        Of course! What Robert said!

  4. Jonathan January 24, 2015 at 12:03 pm #

    I would love to see this video.

  5. Ming January 28, 2015 at 4:09 pm #

    As a Boglehead believer and a subscriber to general life rule of thumb that…anything that looks too good to be true most likely is. And so when I looked at the chart of the dual momentum strategy and that insane gain WITH lower volatility, I was both very excited and also very skeptical.

    I am in the process of acquiring the Dual Momentum Investing book (price at Amazon is way too high imo).

    But my question is this. And during my brief research into momentum strategies, I saw a video by a Columbia professor discuss this in 2010…so this has been known for a while. Why then, if this is so obvious and relatively easy to implement and seems to have almost guaranteed higher returns AND lower risk…why then are not all the professional managers and active mutual funds on Wall St making a killing from this strategy? Is this something new that has been discovered? Why should we lay people then just invest in a “dual momentum mutual fund” pay a management fee, but then get rewarded with much higher returns and lower risk without having to do the monthly calculations?

    So in other words, if this active management/market timing strategy works so well, then why isn’t everybody in on this already?

    • Miles Dividend M.D. February 6, 2015 at 12:55 pm #

      Ming,

      This is a great question and worthy of an entire blog post which I will work on. Very busy with another venture right now so until then…

      Momentum is a weird story. We humans are much more prone to adopt intuitive approaches such as “value” (purchasing products priced below their intrinsic value,) and “size” ( a risk story) than we are to stick to an empirical observation (such as the momentum anomaly) particularly when the going gets tough.

      Larry Swedroe (whom I greatly admire) admitted on Masters in Business that he dousn’t tilt hard momentum, because it doesn’t conform to his underlying understanding of the efficient market hypothesis (that risk and reward are linked.) (never mind that momentum is a far more pervasive and convincing observation than any of the other factors mentioned.

      AZ

    • Graham March 4, 2015 at 2:06 am #

      Part of the issue is tracking error vs. indexing in bull markets. Funds that under-perform for years lose all their capital and the manager is out of a job.

      There are many constraints that institutional investors have that individuals don’t.

      Better to do your research, after all, no one cares more about your money than you.

      • Miles Dividend M.D. March 8, 2015 at 5:28 pm #

        Graham,

        I am not following your point here. You seem to be describing survivorship bias, and calling this tracking error?

        It’s always a good idea to do your research, as you point out, however.

        AZ

  6. PowerPlayer January 31, 2015 at 8:31 pm #

    I’ve spent a couple hours playing around with the data and while my timing portfolio is definitely outpacing some of the asset classes you describe (and significantly), my returns are still 50k shy of what you demonstrated. ($194K in my case). I also can’t get portfolio visualizer to return the numbers you demonstrated for your comparison asset classes. Any light you can shed would be very helpful.

    Still very interested in your analysis.

    • Miles Dividend M.D. February 6, 2015 at 12:44 pm #

      Try this:

      Portfolio visualizer>tools>timing model:relative strength>timing period:6 months.

      AZ

      • PowerPlayer February 11, 2015 at 7:27 pm #

        Thanks! This enabled me to get similar results to yours.

        Similar to the last comment posted, I can’t help but wonder how this platform would have performed in 20-year segments prior to 1995. I plan to dig into this a little more (find equivs to DFEMX, DFALX, VFISX etc)

        • Miles Dividend M.D. February 13, 2015 at 2:47 am #

          Good luck! those were the oldest index funds I could find! If you find older, please share.

          AZ

  7. RichG February 1, 2015 at 4:32 pm #

    Great article. Can you recommend a website or App that lets you do a quick comparison between funds, looking back 6 months, without having to research each fund individually.

  8. Jesse February 8, 2015 at 6:04 pm #

    Your strategy works well during the time period you tested but not well during other time periods. It performs very poorly during sudden drops in the market. Ultimately any strategy that attempts to time the market is likely doomed to fail. I am reminded of this graph: http://static2.businessinsider.com/image/53309b9eecad04e412a1750b-1200-900/cortd.jpg

    • Miles Dividend M.D. February 8, 2015 at 9:20 pm #

      Jesse,

      Thank you for your comment.

      I’m willing to entertain your criticism, but please share with me any 30+ year time period where dual momentum hasn’t “worked.” Evidence please! Empty claims need not apply.

      Flash crashes and whipsaws are examples of market conditions that will treat dual momentum Poorly, but please name the investment strategy that performs well in all market environments. I can’t think of one, certainly not traditional buy and hold.

      As to the “10 best day” chestnut depicted in your graph, it is incredibly misleading since most “best days” occur in the midst of volatile bear markets which are better avoided altogether.

      See: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1908469

      • Jesse February 11, 2015 at 7:06 pm #

        Thanks for the reply! It is very difficult for me to give you great evidence. I spent hours using portfoliovisualizer after your post and concluded that it might be dumb luck. Unfortunately I can’t run the exact same simulations because mutual fund products change over time and the index information seems to be limited. But I can say that If you include pre 1987 data in your models the timing method does not seem to work. This is precisely because you take a massive hit on the worst days in the late 80s and then miss the best days. So this method does not handle flash crashes very well.

        Another issue: If we are investing in the 1980s what international mutual fund would we look to for growth? Japan I imagine. If you include Japan you take an even larger beating. And a big reason why your timing model is so successful in the 2000s is the inclusion of China in emerging markets. Would we have known that China was about to kick that much butt and that we should be tracking the emerging market index? I’m not certain.

        If there happens to be a slow downturn before a crash then your model does well (2000 and 2008). If one of your positions takes off and soars like an eagle then you do very well (China). These three events explain all of your success. Will the same pattern happen for the next 25 years?

        And if it does do well for the next 25 was it just dumb luck? If we look back 25 years from now and we talk about how great our timing strategy worked will we remember that back in 2015 no one was excited the African economies which in retrospect (25 years from now) seemed poised for incredible growth?

        • Miles Dividend M.D. February 13, 2015 at 2:45 am #

          Jesse,

          It is easy to come to a conclusion that something will not work without examining the data. Running the numbers often suprises.

          I chose the time series in my backtests simply because it was as far back as I could go with passive index fund data. But Gary Antonacci does back test as far as 1971 in his book Dual Momentum, using index data, and the results were equally stunning, both in the 80s and the 70s. I highly encourage you to read the book.

          Everywhere I have looked I see that dual momentum’s behavior is fairly consistent across all time series and markets. It decreases drawdowns significantly early in bear market’s, (with the exception of “flash crashes” but we’ve really only had one of those ((in 87,)) and boosts returns significantly after the early emergence of bull markets. The most important factor , of course is the mitigated drawdowns. Also important, is that it is cheap, with little trading.

          Again, I am open to new evidence, but theoretical arguments about when any approach might not work just aren’t convincing to me. Let’s keep it empirical.

          AZ

  9. Bob Werner February 16, 2015 at 8:57 am #

    Alexi, very nice information. I’m one of your students in you and Brad’s miles class and very much appreciate the type of well beyond the box information you are presenting.

    On this particular investment strategy I would like to be all in. You mentioned that you only use it in your non tax accounts. I was wondering why that is? It would seem that a 7% differential would offset and taxes from flipping the taxable accounts.

    It would seem that losses and loss carry overs would go a long way as well.

    Have you taken the time to see what the actual after tax return would be in taxable accounts?

    Special Thanks, Bob W

    • Miles Dividend M.D. February 17, 2015 at 3:40 pm #

      Bob,

      Welcome to MD2! I’m so glad you are here.

      Gary Antonacci, the author of Dual Momentum, does argue that this approach is tax efficient because you tend to take short term losses and long term gains, but I am skeptical that it can consistently beat a tax optimized buy and hold approach in taxable accounts. I haven’t modeled it, but losing 39%+ on short term capital gains (at my tax bracket) seems a high hurdle indeed. If I were in the 15% tax bracket or below, I might entertain it for taxable accounts as well…

      AZ

  10. Bob Werner February 16, 2015 at 9:05 am #

    Off topic — After my last comment I started thinking about Panama. Searched your site and see no mention.

    Just curious, have you looked into moving investment assets to Panamanian trusts.

    Any knowledge or ideas you could share about this off shore haven.

    You’re welcome to PM me if you prefer.

    • Miles Dividend M.D. February 17, 2015 at 3:43 pm #

      Bob,

      I don’t mean this in a judgmental way at all, but moving my money offshore to avoid taxes is not an approach that I would feel comfortable with personally. Although I am quite comfortable taking advantage of strategies embedded in our labyrinthine tax code (like the back door Roth) I arbitrarily draw my line of comfort there.

      AZ

  11. Grant February 23, 2015 at 2:05 pm #

    Alexi, there is an interesting thread on Bogleheads on Dual Momentum, including some criticism of the strategy (from yogiyoda with a link to his review
    on Amazon).

    https://www.bogleheads.org/forum/viewtopic.php?f=10&t=158297&p=2389185&hilit=dual+momentum#p2389185

    With your knowledge and experience with the strategy, I wondered if you might be interested in adding to the discussion?

    • Miles Dividend M.D. February 23, 2015 at 7:56 pm #

      Grant,

      That’s an interesting thread. Thanks for alerting me to it.

      Unfortunately I can’t jump in because I was kicked off of bogleheads for writing about travel hacking, which the moderators were conviced was illegal! I love the depth of conversation in that forum, but it is no haven for free speech.

      Ming wrote an interesting question earlier along these lines and I am thinking through a deeper critique of dual momentum along with a more complete exposition of my own perspective on why I have chosen this path despite its risks.

      • Grant February 24, 2015 at 5:45 pm #

        Alexi, I’m sorry to hear about your Bogleheads experience! There is certainly a great depth and breadth of wisdom to be found there, nonetheless. I look forward to your critique of Dual Momentum.

  12. Grant February 23, 2015 at 2:12 pm #

    Alex, there is a thread on Bogleheads on Dual Momentum. with some interesting comments, including some criticism (yogiyoda with a link to his review on Amazon).

    https://www.bogleheads.org/forum/viewtopic.php?f=10&t=158297&p=2389559&hilit=dual+momentum#p2389559

    With your knowledge and experience with the strategy, I was wondering if you may like to join the discussion?

  13. Jimbo April 29, 2015 at 7:43 pm #

    I’m trying to wrap my head around this. Do you hold only one fund at a time and then sell of it and buy the best performing of the others at the end of each month (best performing being over the previous 6 months)? That wasn’t entirely clear from the blog post. Thanks.

    • Miles Dividend M.D. May 3, 2015 at 3:08 pm #

      jimbo,

      Yes that is how you do it. Adjust your allocation each month based on the best performer for your lookback period, and allocate 100% to that asset.

      Of course you can couple multiple pairs of assets with cash to to greate several buckets that you adjust each month if you like, but each bucket will be 100% allocated to the winning asset.

      Hope that helps,

      AZ

  14. Thriphty August 3, 2015 at 12:28 pm #

    Thanks for your article. In his book Gary A recommends using S&P 500 ETF like SPY, IVV or VOO. However I saw this article http://seekingalpha.com/article/3203656-dual-momentum-a-simple-yet-successful-portfolio-management-strategy
    the author is using VTI (Vanaguard Total US Stock Market ETF) and many have commented that using VTI over SPY is a better thing to do. So what’s you take on this, should I go with VTI or SPY. I’ve done a lot of research on this and ready to start with Dual Momentum GEM, but still debating between SPY and VTI.

    • Miles Dividend M.D. August 5, 2015 at 4:10 pm #

      Thriphty,

      It won’t matter too much. VTI is highly correlated with SPY. My guess is that using VTI will mildly increase returns (and drawdowns) due to the shift towards smaller size stocks. (i would expectr similar effects by substituting all world ex US, in place of an EAFA fund)

      In the end I would probably just go with whichever is cheaper with your brokerage.

  15. Brandon Zingsheim November 15, 2015 at 9:27 pm #

    How’s it been going for you lately? I imagine you’ve had to keep you emotions in check over the last couple months as performance of this strategy hasn’t been exactly stellar…thoughts?

    • Miles Dividend M.D. November 15, 2015 at 9:51 pm #

      My emotions have been fine actually. As of today I am actually still ahead of where I would have been had I never changed course 14.5 months ago. I find it kind of liberating to entrust my strategy to an evidence based rules based approach. It means that my opinions of where the markets are going have become completely irrelevant. Which is a good thing, I think, because I know that my prediction skills are not to be trusted.

      • Brandon Zingsheim November 16, 2015 at 4:25 am #

        Good to hear. As someone who sort of went in on this strategy early summer, you can imagine the results haven’t been as kind. I’m sticking through it as I know that momentum trading doesn’t necessarily correlate with S&P and (more importantly) in the long run, the evidence I’ve reviewed will carry the day.

Leave a Reply


Visit Us On TwitterVisit Us On FacebookVisit Us On Google Plus