Looking Under Rocks

3 Flares Twitter 2 Facebook 1 Filament.io 3 Flares ×

Sun Tzu observed that “if you know the enemy and know yourself, you need not fear the result of 100 battles.”

And as pat as it is to compare investing to war, it is also apt. For every investment is a transaction between a buyer and a seller, which means that every trade, on some level, is dualistic. You either win or you lose.

In my last post in this series I looked at the broad philosophical justification for dual momentum investing. Namely the empirical observation that momentum is everywhere, and the hypothesis that as long as humans are involved in markets, momentum is likely to persist. Performance chasing is just that essential to our make up.

So in this post I want to make like Sun Tzu and to consider my own chosen strategy’s weaknesses before entering into battle. For in battle there is no time for doubt. And since there is no strategy that does not contain weakness, it stands to reason that it is better to ponder any strategy’s faults in order to see trouble coming, and to react to it with calm rather than panic.

And so to begin this exercise let me lay out my fundamental assumptions about dual momentum. For without a doubt it is what I am sure of that is not in fact true that has the most potential for my own destruction.

Here then are my honest assumptions about the nature of capital markets and dual momentum investing that have informed my decision to adopt this strategy for my own tax exempt portfolio.

1. Since relative price momentum has existed everywhere that it has been studied, and has persisted long after it was first described, it is likely to persist into the future.

2. Absolute momentum (or timeseries momentum) is a form of trend following. As such by its very nature it must provide persistent and predictable downside protection.

3. Short-term treasuries are a risk-free asset.

So let’s come up with some counter arguments or hypothetical scenarios which would weaken my underlying assumptions and cause the persistent underperformance a dual momentum portfolio.

Hypothesis 1: Momentum Is Pervasive And Persistent in all markets.

Counter argument: The make up of the market is changing.

Although dual momentum has been present everywhere it has been looked for to date, markets are rapidly changing. Each year passive index funds claim a larger proportion of invested dollars (though at this point they’re still the vast minority.)

And if the performance chasing active investor is apt to perpetuate momentum by buying recent winners, shouldn’t the rebalancing of passive investors perpetuate the exact opposite? (Rebalancing should move in opposition to performance chasing, because by it’s very nature rebalancing means selling recent winners and buying recent losers. )

Furthermore the last few years have seen the rise of the Roboadvisers. If future decisions are to be made more and more by perfectly rational computer algorithms, and if momentum is an expression of human irrationality, then isn’t there a decent chance that the momentum effect will become ever more dilute with automation?


I have not been programmed to chase performance…

Takeaway: At this point the market is still very much driven by human decision-making, and active management. Given this, my feeling is that momentum continues to be a very good bet. But it’s worth always keeping an eye on the evolution of the market because the forces that drive market moves may change and at that point relative price momentum may become a bad bet.

Hypothesis 2: Absolute Momentum Protects Investors Against Big Drawdowns.

Counter argument: There are at least two exceptions to this rule.

For the most part bear markets are all similar versions of each other. There is a fundamental change in the marketplace that shocks the market, traders panic, volatility goes through the roof, and the stock market stairsteps downwards over the course of months and years. It reaches bottom, and then begins a gradual recovery in fits and starts.

And it is this classical version of a bear market that packs most of the returns beating punch in the dual momentum strategy. For although relative price momentum allows the investor to gain incrementally more during bull market periods, absolute momentum allows the investor to lose dramatically less during Bear market periods. Why? Because once the market starts stair stepping down, and the returns of the invested asset class over short-term treasuries during the look back period have been erased, then the investor switches fully into the safe asset of short-term treasures. In this way the dual momentum investor is safely on the sideline for most of the bear market (and a brief time at the beginning of the recovery.)

But when would such an approach not work? The answer is obvious:

In A Flash Crash.

A flash crash such as what occured on black Monday in 1987 is the worst case scenario for a dual momentum investor. Why? Because it causes big problems in 2 separate phases.

Phase 1: the crash.

The crash itself will almost definitely hurt your dual momentum portfolio.

Unless the crash occurs in the midst of a bear market at a time when you have already divested of stocks or other high risk assets,you will feel the full weight of the flash crash’s destruction.

It will be difficult to tell your portfolio from that of a 100% “risk on” stock portfolio on the day of the crash.

But if you are anything like me, it will be worse than that. After all, dual momentum’s primary attraction  is it’s a promise of diminished drawdowns with unfettered returns. But here is an instance when “temporal diversification,” offers none of the benefits of plain old asset class diversification.  In this instance, the dual momentum investor’s portfolio is indistinguishable from that of a 20 year old stock trader who wants big rewards for taking big risks.  In other words the concentration of stocks within the portfolio at the time of the crash creates a bad mismatch between investor risk tolerance and portfolio risk.

But don’t worry, it gets worse…

Phase 2: the recovery.

If the flash crash is a plain old panic such as what was seen on Black Monday in 1987, then the market will react rationally afterwards, and prices will bounce back relatively quickly. Equities will soon recover to reflect their true value in the market.

So what would be expected is a very rapid drop, followed by a pretty rapid recovery.

This is the truly the worst case scenario for the dual momentum investor. For as soon as the month of the flash crash ends, the strategy will dictate the selling of high risk assets and the buying of safe assets. So just as the market begins its recovery from its hick-up, the dual momentum investor will be on the sidelines, losses locked in.

This nasty scenario really poses the same problem for any tactical asset allocation strategy including the 200 day moving average approach. (And frankly I lack the back testing tools to display an isolated look at this time period with dual momentum) so let’s look at the  years surrounding Black Monday using a 10 month moving average strategy with the S&P 500.


moving ave flash


(Tough coupla years for trend following…)


Take home: flash Crashes are kryptonite to trend following strategies like dual momentum.

One could come up with strategies to avoid this doomsday scenario, such as filtering out large single day drops in bull markets, but since there’s only been one black Monday in our history, it would be impossible to test such a strategy in any meaningful way. So my feeling is that it is wiser to take the bad with the good and stay true to dual momentum even should such an event occur again in the future.

I comfort myself with the rarity of such events in the past in contrast to the pervasive outperformance of dual momentum in long term backtesting.

So flash crashes are absolutely a risk to be aware of and to prepare oneself for psychologically.

But they by no means the only risk….

There Is Also The Risk Of Whipsaws.

Imagine you’re a dual momentum investor who uses 12 months as your look back period.

Now imagine the market slowly trends down over 12 months erasing the prior year’s gains. On the 12th month in keeping with the strategy you convert to a risk-free asset such as short term treasuries which have just begun to outperform the alternatives in your portfolio. As luck would have it the market starts to rally shortly after you move to the sidelines.

And several months later, when prices recover enough for you to become fully invested in stocks once again, the stock market once again takes a turn for the worse.

In this scenario you are “risk on” when the market goes down and “risk off” when the market goes up. Your timing is perfectly out of sync with the market. Never a good place to be.

This turn of events is termed a “whipsaw”.  And it can be a random but destructive turn of events for a tactical strategy such as dual momentum.

Take-home message: market whipsaws when perfectly out of sync with the momentum look back period can be very disruptive for the dual momentum practitioner.

One way to avoid this risk would be to use multiple look back periods when implementing your dual momentum strategy. For instance you could manage one third of your portfolio with a three-month look back, one third of your portfolio with a six-month look back, and one third of your portfolio with a 12 month look back.

This is probably something worth considering.  The main downsides to such an approach would be the increased complexity, and the increased trading costs (both in terms of commissions and bid ask spreads.)

Hypothesis 3: Short-Term Treasuries Are A Risk-Free Asset.

Counter argument: risk-free assets are only risk free until they are not.

I’m fairly certain that during the time of Alexander the Great, Greek government debt must have been considered “risk-free.”‘ The same must certainly have been true of Rome under Julius Caesar. Since Greek and Italian debt is no longer considered “risk-free,” the lesson is clear.  There is no such thing as a risk-free asset.  With time every asset is risky.

But the day that US short-term treasuries become a risky investment will be bad news for almost every investment class and approach, not just the dual momentum strategy. So while this is a real risk, it’s a tough one to hedge against regardless of your investment approach.

So those are some the risks that I can perceive lying out there waiting for me as a dual momentum investor. But I’m quite certain these are not the only risks. They are simply the ones that I can conceive of at this point in time.

Since I’ve chosen to go down this avenue, it is fairly obvious that I like the rewards of this approach enough to knowingly take on these risks. But if there are other risks to dual momentum that concern you, by all means share them below.

Because as Sun Tzu points out the more we know about our own weaknesses and strengths and as well as those of our competitors, the less we have to fear.

3 Flares Twitter 2 Facebook 1 Filament.io 3 Flares ×

15 Responses to “Looking Under Rocks”

  1. Grant April 12, 2015 at 7:59 am #

    I think every strategy has it’s risks. The big risk for buy and hold is the bear market. Probably the biggest risk for dual momentum is the whipsaw. Maybe a good strategy is to diversify these risks and split one’s portfolio in some fashion between the two strategies which is what you are doing using dual momentum in your tax protected accounts and buy and hold for the rest. I’m still in the “thinking about it mode”. With 1/3 of my assets in tax protected accounts, it would take a lot of rearranging in my taxable accounts (and cost due to capital gains tax) to maintain my asset allocation there and use dual momentum in tax protected.
    Do you think the look back period of 6 months or one year would have an impact in the likelihood of the whipsaw problem? I see from the charts that U.S. equuty, although still “in the lead” (at least in Canadian $), is rapidly being challenged by emerging markets. Do you use emerging markets in your dual momentum mix?

    • Miles Dividend M.D. April 12, 2015 at 12:58 pm #


      I don’t know that you need to keep your overall asset allocation static when converting a portion of your portfolio to a dual momentum strategy. After all the risk profile of a dual momentum portfolio is equivalent to that of a very bond heavy buy and hold portfolio (though its upside is considerably greater.) I certainly wouldn’t take a tax hit for this purpose.

      I think of the whipsaw problem as really a problem of randomness. My guess is that at some point the market will randomly move out of sync with the periodicity of any look back period.

      I don’t use emerging markets in my strategy simply because there are no low cost EM fund options in my 403B account. But if I had that option I would strongly consider including EM, as back testing shows that adding EM has significantly increased returns at the cost of a mild increase in drawdowns/volatility.

      Even with my 6 month lookback I have yet to make a trade. But the principle cost of shortening the lookback period is increased trading costs. By any measure and with any periodicity DM is a very low cost, infrequently traded strategy.

      I expect foreign developed markets to overtake the S & P in a month or 2 for the 6 month look back period. It will take a little longer if the trend holds, for the 12 month look back period.


  2. Robert April 14, 2015 at 8:46 am #

    Alexi, another good post!
    A few comments regarding the first assumption…
    1. Several recent studies have shown that as passive investments have grown, the stock market has increased volatility. It is a cause-effect relationship. Active trading makes the markets more efficient. As the % of passive trading increases, market efficiency decreases. One of the effects is that correlation between stocks increases, so stocks may persist in being over or under-priced. See http://tinyurl.com/n53qqvs. All these factors suggest to me that increases in passive trading won’t necessarily undermine dual momentum, and might even enhance it.
    2. Andrew Lo has done some nice work on the dynamics of the evolution of markets. Growth vs. value has come and gone in cycles, though the disparities decrease over time. I suspect that even if momentum went away, the same would apply, and it would come back in style again, though perhaps at a lower return premium to buy-and-hold.
    3. I’m doubtful that it is possible to keep an eye on the markets and switch from a momentum strategy if it ceases profitability. More likely that will only make you trade out just when you should be staying in. The timescale for knowing it isn’t working is too long to act on it.

    • Miles Dividend M.D. April 19, 2015 at 9:58 pm #

      Thanks Robert,

      1. First of all correlation is not causation so is theincreased volatility coming from the rise indexing or algorithmic (high frequency) trading? Second of all, Volatility can be good or bad for dual momentum, but a 100% passive market would not have price momentum as it (almost assuredly) is generated by human decision making and the flow of money in and out of funds based on past performance.

      2. It is hard to to understand a model where momentum is as susceptible to crowding as value or size, since multiple expansion decreases expected future returns for value/growth/size groupings, but paradoxically increases the momentum effect!

      3 You are probably right.

  3. Grant April 14, 2015 at 11:40 am #

    Alexi, I didn’t mean keeping my overall asset allocation static on an ongoing when doing partial DM, just the one time at the beginning. I believe you set up and maintained your original asset allocation in taxable and went with DM in tax preferred? For me, just one time change in taxable would be costly due to capital gains taxes. So, as you say, the volatility of DM is similar to a bond portfolio, so it would make sense for me to convert tax preferred to DM ( mostly bonds in my case) and leave taxable as is. That would more or less maintain the volatility of the portfolio, but increase the expected return – the beauty of DM, assuming all goes according to plan!

  4. Robert April 16, 2015 at 10:16 am #

    Alexi, I’m a few days behind on the AlphaArchitect blog, but found this paper there: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2585056. It is quite interesting, particularly with regard to time-series momentum, which it tests back to the mid-19th century. It digs into the data-mining issues, among other things. Here are a few of my take-aways relative to our interests here:
    1. Some support for 7 month lookback period.
    2. Puts some meat on your concerns about tracking error. Note particularly the discussion around p. 36. Essentially, a single unit measure like Sharpe ratio calculated over an extended period (such as 20 years) is quite deceiving for investors with shorter time horizons (say, 5-10 years). He shows in Fig. 3 how much momentum outperformance is due to a few isolated negative events. Outside of those events it may underperform. The chance of momentum outperforming the market over 5-10 years is only on the order of 50%. (Or, to put it another way, the chance of it underperforming the market over that time horizon is about 50%).
    3. Over long time horizons, MOM did outperform on a risk-adjusted basis.

    • Miles Dividend M.D. April 19, 2015 at 9:38 pm #


      As I may have mentioned this week was a call week and was an absolute bear, so I haven’t had time to actually read the paper yet.

      But my take away from your takeaways is:

      1. Interesting! As I mentioned in this blog post I am intrigued at The possibility of diversifying away some of the period specific whipsaw risk by using three separate look back periods equally weighted in my portfolio. I haven’t figured out how to backtest it yet though.

      2. I’m not surprised that momentum is a coin flip depending on the decade. What I think dual momentum does extremely well is limit drawdowns in all instances other than flash crashes. That has always been its prime attraction to me. (But don’t get me wrong I am as susceptible to the siren song of increased returns as much as anyone.)

      The correlate to this limitation of drawdowns is that it allows you to take on more risk. As an example when I was buy and hold I had a 70/30 allocation. Now I am 100% stocks and I’ve gotten increased returns because of this. I am only 100% stocks because of the promis of a reliable exit signal though, not because my risk tolerance has changed. But I suspect dual momentum will beat a 70/30 allocation most decades.

      3. I would think so.

  5. Fred C Dobbs April 18, 2015 at 11:28 am #

    Alexi – It’s good to think about the possible risks of any strategy, but you also need to realistically think about the likelihood of those risks occurring and what the overall impact will be if and when they do occur.

    Flash crashes have been extremely rare. I noticed only one or possibly two during the past 80 years. Short term Treasuries becoming risky may or may not ever occur. Whipsaws have occurred during Antonacci’s backtests. Yet momentum results have still been strongly positive, not only in Antonacci’s backtests, but also in other tests of relative momentum going back 200 years and absolute momentum going back 800 years.

    • Miles Dividend M.D. April 19, 2015 at 9:40 pm #


      I agree with you! Obviously I do feel that dual momentum is a good bet, or I wouldn’t have placed all of my tax sheltered retirement money into such a strategy.

      I still think there is value at looking at these risk situations because long-term dual momentum investors will experience periods of severe underperformance just because of randomness. So recognizing these events may make it easier to stay the course when they happen.

      Most important with any strategy is screening for events that can blow us up. Because even a low probability event went extremely powerful can ruin everything.

      Thanks for your comment!

      • Samir April 22, 2015 at 7:01 pm #

        1. Statistics and probabilities have never helped the individual, only populations. Prudent risk management calls for considering the worst case scenario and asking oneself whether it’s bearable or not. If it’s not bearable, then consider the next best (or less worse) alternative (this could very well be inaction) and pursue it if its bearable

        2. Have you looked at GMOM, Meb Faber’s fund of funds in which he’s executing a global dual momentum strategy using a much more diverse set of funds. It has an ER of 0.94%. What’s your projected aggregate ER?

        • Miles Dividend M.D. April 23, 2015 at 12:51 pm #


          That’s a very provocative first sentence! (And I don’t agree with it, but you got my attention.)

          On the other hand I Completely agree with you on keeping a laser like focus on left tail risk, which is, after all, what my “cowards” series was all about.

          As to GMOM. I love Faber, but I don’t know why anyone would pay 94 basis points for GMOM, GEM is so easy to implement and to date has only cost me an ER 4 basis points (the ER for VIIIX). I predict that GEM will outperform GMOM going forward even before fees, and certainly afterwards. We’ll see.

          • Samir April 24, 2015 at 7:30 pm #

            What about my comment on statistics not helping the individual do you disagree with?
            With regards to GEM outperforming GMOM, I haven’t read Antonacci’s book, so I don’t have a basis to rebut. I’ve read Faber’s writings though (A Quantitative Approach to Tactical Asset Allocation, available at SSRN: http://ssrn.com/abstract=962461), and comparing that to what you’ve described, the difference appears to be in that GMOM uses more asset classes and sub-classes, presumably to take advantage of factor tilts. What makes you predict that GEM will outperform GMOM?

          • Miles Dividend M.D. April 28, 2015 at 9:15 am #

            I think that an individual repeatedly participating in a contest in which he holds a slight statistical advantage can be very lucrative for the individual.

            A few differences with dual momentum and Faber’s approach. GMOM has:

            More asset classes.
            More expensive underlying ETFs.
            200 day moving average in place of absolute momentum.

            More asset classes I’m ambivalent towards, but doubt it helps and suspect it probably increases trading costs.

            Including expensive underlying ETFs including Faber’s own companies ETFs (eg GVAL…), is a clear conflict of interest, and is very likely to be harmful to returns.

            12 month absolute momentum should have less trading than the 200 day moving average rule if past is prologue. Again lowering the cost.

            Most importantly paying over 90 basis points for a simple strategy is absolutely crazy in my view, but as I said above, there are other aspects of GMOM that I am deeply suspicious of, even assuming similar costs.

  6. PathtoFIRE April 22, 2015 at 1:17 pm #

    I’m definitely intrigued by this strategy, and am considering trying it in my 401k, which represents about 30% of my portfolio. I’m not exactly sure which classes to use. I have FXAIX (SP500, ER 0.025) and VTSNX (Total International, ER 0.12). My short-term Treasury fund is kind of sucky, FDLXX (ER 0.42). I also have VBMPX (Total US Bond, ER 0.05), VIEIX (Mid cap, ER 0.08), DFSVX (Small cap, ER 0.53), and finally IGREX (Global real estate, ER 0.89).

    Just doing some basic searches on stockcharts.com, and substituting FSBIX and FUSVX for FDLXX and FXAIX, I get at least 5 trades since April 2014: initially in FXAIX, into IGREX June 2, 2014, into FXAIX October 1, 2014, into IGREX on Feb 1, 2015, back into FXAIX on March 2, 2015, and then into VIEIX on March 1, 2015.

    Now these are all in my 401k, so no trading costs…but still this churning is a little disturbing, especially back and forth between SP500 and Global RE.

    • Miles Dividend M.D. April 23, 2015 at 10:53 am #

      Dual momentum just refers to toggling between 2 imperfectly correlated assets plus short term treasuries based on total returns for a look back period between 3 and 12 months.

      The simplest is just global equities momentum, which uses S&P, Non US developed (EAFE), and total bond market. (So I would use the good total bond fund over the expensive short term treasury fund in your case.)

      But you can pair up as many buckets as you want. eg: S&P/EAFE/SHY, real estate/EM,/SHY commodities/HY bonds/SHY…..

      you can do sector rotation as well which allocates to best sector in the US economy or cash.

      And don’t worry about churning. I have yet to trade out of S&P, (though next month could be the first time.) Tis is a rarely traded strategy..

Leave a Reply

Visit Us On TwitterVisit Us On FacebookVisit Us On Google Plus