Dual Momentum: A Look at Market Cycles

Dual Momentum is one of my favorite investment strategies for do-it-yourself investors. The historical performance of Dual Momentum—even in the now effete twelve month lookback form—has been fantastic going back nearly seven decades.

There is no other strategy I can think of where investors can experience double digit annualized gains over full market cycles with 25 percent drawdowns (in the worst case) and just a couple trades per year.

Although the long backtests are beautiful and the results are solid, Dual Momentum has come under a lot of fire lately. Investors who employed the strategy are losing faith after several shakeouts in the past couple years, the quants are poking holes in the mechanics of the strategy, and the buy-and-holders are flaunting massive gains as the market keeps jumping up on every dip.

An argument that's been made before, and I've been on that side as well, is that we can expect equity trend strategies to underperform the index in the upside portion of market cycles. This is often observed in single market backtests. As a trend strategy, Dual Momentum would fit this argument.

However, Dual Momentum doesn't just beat the benchmark by a few points annualized as we see with single market trend models by avoiding long downtrends. Dual Momentum has crushed the market by also picking the strongest equity class—U.S. or International—in the broader market upwards half-cycle.

I want to focus on the "bull market" or upside half-cycles in U.S. equity markets because this is where trend investing can tend to underperform equity portfolios. To help put this relative performance into a visual format, I decided to look at all the "bull market" phases since 1970. We'll see how a Dual Momentum investor would have felt standing beside the S&P 500 and a balanced portfolio (33.3 percent U.S. stocks, 33.3 percent International stocks, and 33.3 percent bonds).

We know Dual Momentum has avoided the deep drawdowns of extended market downwards half-cycles. In other words, I believe it is unlikely to see a Dual Momentum investor complaining about their recent performance at a time like the summer of 1974, summer 2002, or fall 2008.

July 1970 to December 1972

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

In this short bull market Dual Momentum underperformed the benchmark portfolios for more than a year into the cycle. Dual Momentum quickly caught up in the latter half of this half-cycle, essentially meeting U.S. stock performance at the cycle peak.

October 1974 to November 1980

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

The dynamics of the late-70s upwards cycle was very similar to the brief earlier cycle. Stocks ran hard at first with Dual Momentum lagging behind for the first half of the cycle. However, later in the cycle Dual Momentum caught up again, barely edging out stocks over the entire half-cycle period.

August 1982 to August 1987

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

While the mid-1980s market half-cycle started off similar to the previous bull markets in the 1970s, it certainly finished different. Dual Momentum lagged at first, for approximately a year and a half, but the signal moved the Dual Momentum investor into International stocks. For the next few years International stocks wildly outperformed U.S. stocks and a Dual Momentum investor finished the half-cycle much wealthier than an American focused investor.

Even the normally lagging Balanced Portfolio performed neck-and-neck with U.S. stocks. While the bond component may have held returns back, the International stock allocation gave the Balanced Portfolio a lot of strength into the latter half of the upwards portion of the cycle.

We all know what happened to U.S. stocks in October 1987. What many people don't know is that International stocks did much better in that crash comparatively. By holding International stocks, Dual Momentum did better in that sudden crash, even with a full equity allocation.

November 1987 to August 2000

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

As the chart shows, this long market upwards cycle was almost all U.S. stocks driving the performance. Dual Momentum got a bit of an edge on U.S. stocks when the market shifted to bonds in 1990 and to International stocks in 1994.

Overall both Dual Momentum and U.S. stocks gave a 9x return in little more than a decade. The momentum was so strong that the Dual Momentum signal held in U.S. stocks for the 1998 Bond and Currency Crisis, riding to the peak of the U.S. telecom and tech bubble.

October 2002 to October 2007

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

This half-cycle we're back to the familiar performance of Dual Momentum lagging the market for the first half of the upwards cycle. Again, Dual Momentum finishes strong as International stocks took off in the mid-2000s. In fact, Dual Momentum had investors in International stocks for the vast majority of this half-cycle.

March 2009 to April 2019

Credit: TheRichMoose.com, MSCI Inc., FRED-Federal Reserve St. Louis.
Right-click to enlarge image.

This market half-cycle has been a significant deviation from the past cycles we've looked at. Normally, a couple years into the cycle, Dual Momentum would surpass the returns of the Balanced Portfolio and U.S. stocks. and pull away from there.

Although I don't have data going as far back as Dual Momentum's publisher, we have never seen Dual Momentum perform this poorly in a market half-cycle. Dual Momentum has never lagged the S&P 500 or a Balanced Portfolio by so much for so long.

For some perspective, U.S. stocks would have to fall 60 percent to match Dual Momentum's returns over this market half-cycle. It would take a 25 percent decline for the Balanced Portfolio. This number isn't impossible. In fact, very prominent market researchers have been suggesting a 60 percent decline in U.S. stocks would be necessary just to get us back to a baseline historical valuation.

For myself, I am not sure where markets are going to go. One thing I do know is that zooming into shorter period performance can help us get a picture of how a strategy performs.

If there are clues that something may be shifting in the performance of a strategy, we need to keep a more careful eye open. In my research I've seen long-term shifts in other investment strategies; once winning strategies becoming laggards. Of course they are only seen in hindsight, but it is important to look for these shifts before they drag down your portfolio returns for decades.

While I'm not abandoning Dual Momentum, I am somewhat less confident in the strategy than I used to be. It would take a pretty solid relative performance in the next downwards half-cycle to renew full confidence in the strategy. Gary Antonacci has previously mentioned that he would look to full market cycle performance to determine the effectiveness of Dual Momentum. I would say that a potential challenge is in the works.

For now, I am still comfortable staying with Dual Momentum for a good portion of my portfolio. However, as I've advocated for many times before, don't put all your money in a single strategy. Mix it up. Spreading your bets across a few strategies can reduce a multitude of risks much better than small adjustments within the strategy that try to adapt to market conditions or reduce some of the risks present the strategy.

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: A Look at Market Cycles”

  1. Yanniel says:

    I also started to doubt traditional DM.

    So I moved away from it recently. Instead I am
    running a customized version of it:

    I have 3 accounts in which I run modified DM. The holding period is 1 month for each account; but I place the trades (and compute my signals) on different dates: roughly on the 1st, 11th and 21st of each month. Trying to avoid timing luck.

    Each account looks at 10 different look-backs and holds 10 different equally weighted “sub-portfolios”. This translates in the full portfolio of the account to be allocated in some combination of stocks and bonds.

    I use vanilla ETFs like VOO, VEU if the percentage to allocate to stocks is less than 100%.

    If I reach the point in which the portfolio as a whole is to be allocated to either US or Int stocks; I use 3x leveraged ETFs. I split it like 33.33% UPRO and the rest to SHY. I am willing to let equity exposure go beyond 100% so long the it does not go beyond 1.5x.

    If on the next month the signal says to be 90% on stocks, I get out of the leveraged ETF and use plain vanilla ones.

    I don’t incur commission fees by using NBDB.

    Can I get your input on this Daren? Does it make sense to you.


  2. Daren (Editor) says:

    I think that adding some different mechanics to the strategy can increase the stability of the expected returns the overall strategy can provide. You would be less likely to have poor performance on the basis of bad timing luck on those close signals. However, I’m not sure it will change the risk of the overall strategy underperforming. The post I did looking at the Newfound system covered some of these ideas.
    I’m in the camp that–provided your portfolio is sufficiently large–it is much better to diversify across strategies than within them. Even then, it depends on what you are looking for in your portfolio and what you are using. A lot of strategies and a lot of factors (smart beta applications) are proving themselves to be unsustainable for long-term investing. They either add costs with no added returns, or they are subject to front-running reducing their effectiveness, or they come with obscured risks.

  3. Thanks for your opinion Daren. I am running two strategies. One based on DM and the other is just buy and hold with some rebalancing. The latter has done pretty well this year. Not so much for DM.

    I have been looking at factors; but I haven’t ventured beyond momentum.

    I have also started to look at things like all weather portfolios with some tactical allocations. Have you written about this before?

  4. One concern that I have about diversifying across strategies is that although the strategies might seem different, under the hood they might be exposed to the same sources of risk. I don’t know how to quantify this.

    Something similar happens when I consider a barbell portfolio. Say I divide it in 10 different asset classes. One might thing this could potentially mean 10 different bets; but are there really 10 or less? When s$&& hits the fan all those bets might turn to be just one big bet. These are things that concern me when I try to diversify; specially because I don’t know how to quantify any of this.

    Any tips, if you have them, are most welcome.

  5. Michael says:

    Hi Daren, thanks for your hard work on this article.

    I too am concerned about the underperformance of DM during this bull market, however, from reading Antonacci’s data, it appears that there is a precedent for DM significantly underperforming the S&P 500 during a bull market:

    Contrary to your data for the Oct 1974 – Dec 1980 bull market (he included one extra month than you did), Antonacci reported that DM returned 103.3% while S&P 500 returned 198.3%. Your data shows that DM narrowly beat the S&P 500. Do you know why there is such a large discrepancy? Hard to believe 1 month made all the difference.

    Similarly, according to Antonacci, during the Jul 1962 – Nov 1968 bull market, DM significant underperformed the S&P as well (58% vs. 143.7%). I feel better about DM knowing that there are previous bull markets where it has underperformed to the degree that it is underperforming in the current bull market.


  6. Daren (Editor) says:

    The All Weather portfolio is a pretty good option for buy and hold that is less dependent on equities for returns. I like it in a lot of ways, although there are a few things I would tweak on it to try balance the portfolio.
    The issue with AWP and getting more tactical on that strategy is recency bias can come into play. I see people doing things like swapping out broad equities for Small Cap Value or Momentum Factors, using U.S. equities only, and dropping allocations to gold and commodities. With the All Weather you are also very likely to underperform the market in many situations by a small amount. Many investors just can’t handle that.

  7. Daren (Editor) says:

    I wouldn’t say that all strategies have the same risk exposure in a broad sense, although they do have certain similar risk exposures in a more narrow sense. For example, any portfolio with equities is subject to damage from a 1987 event. But a long-term trend portfolio can be much less exposed to a ’73-’74 or ’07-’09 event than a buy and hold approach.
    The idea with a barbell portfolio is not necessarily the avoidance of correlation of bets. It is based on the idea that we have generally reliable markets and good governance with a positive return on short-term lending (even if it is not big). Then, instead of taking large exposure to different assets (equities, commodities, etc.), we take small exposures and leverage that up carefully. I would rather take 10 bets with 2 percent allocation, leveraged up 3x each, for a total of 20 percent exposure to loss than taking 10 bets at 10 percent each without leverage.
    I think you can be assured that many bets are at risk to correlate to 1 or -1 if sh*t hits the fan. We’ve seen that before.

  8. Daren (Editor) says:

    Gary calculates momentum slightly different from my preferred method. I talk about that in this post: http://therichmoose.com/post20190215/
    Gary has discussed both methods, but he confirms his view that the S&P 500 is a leading market and bases his equity signal on that market. I take the position that it unnecessarily exposes DM investors to single market risk for the sake of a slightly lower backtested max drawdown. There are arguments to be made for both sides, but Gary is smarter than I am so take my position with a grain of salt.
    That difference in calculating momentum is ultimately the reason for the different performance from ’74-’80. I can’t speak for the ’62-’68 period because my data doesn’t go back that far and I’m not sure how much of that underperformance is due to the momentum calculation.
    We’ll see where DM takes us going forward!

  9. Hello Daren, thanks for the interesting post.

    I too am running classic Dual Momentum, and although I’m still pretty convinced by the strategy (I don’t expect the recent market regime to last forever) I’m thinking about diversifying and adding an allocation to a leveraged barbell strategy using an all-world index and margin, in order to get exposure to diversified buy and hold while controlling risk. My questions concern ideas used in several of your blog posts so I guess I’ll just ask them all here!

    1. I like the idea of smarter rebalancing using the trend, and I want to use something simple like a 12-month lookback, with rebalancing only when the trend changes in my theoretical barbell portfolio. My goal is to have a simple check one day a month. Do you think using the same signal as dual momentum adds risk compared to a different signal or simple yearly rebalancing?

    2. Since I’d only be using half of my portfolio for the barbell and I’d be using IB USD margin (I live in east Asia in a place without any king of investment tax) I could leverage even higher than 3x in combination with a trailing stop.

    An example portfolio (taken to an extreme with 5x leveraging of the all-world ETF) : 50% Dual Momentum, 62.5% all-world, 37.5% Bonds. A trailing stop order of 80% is applied to the all-world ETF. This should mean that the barbell part of my portfolio is equivalent to 25 % x 5 all-world and 75% in bonds.

    What would be the downside of this kind of approach? I assume that, if rebalancing on a fixed date yearly, there might be a lot of time spent out of the equity market if a 20% drop triggers an exit of the equity position the months after a scheduled rebalance. Would use a trend-following rebalancing system help with this?

    3. How would you handle periodic contributions in such a trend-following rebalancing barbell system? Imagine if the market was in a prolonged down turn and my equity portion of the barbell had reached 0. Should I still add to the equity part during my next monthly contribution or should I just add 100% to the bond portion of the barbell until the trend reverses?

    Thanks for all your ideas and help!

  10. Daren (Editor) says:

    Some great questions Jon!
    1. I would personally use a different signal for re-balanced the LBP. Since you are using leverage, it is probably advantageous to use quicker signals. I used a 13-week P>SMA signal in this backtest with good results: http://therichmoose.com/post20190322/
    You might also consider other common trend signals: a moving average crossover system, directional moving average, or even an oscillator such as a weekly PMO or MACD. No system is perfect, but different signals can help generate somewhat differentiated returns and might reduce risk.
    2. Would you be using UCITS compliant ETFs? In theory anyone can leverage up a portfolio like that quite aggressively because you can borrow against your bond allocation and stock allocation with 30-50 percent margin. However, I’m not sure it is a great idea because there is little room for error. Personally, if I were to leverage up 5x on stocks, I would increase my bond holding to 80 or 85 percent and run a trend-following re-balancing system with a stop-loss (100 percent stocks, 80 percent bonds at the most).
    3. If I were making regular, moderate sized contributions I would always add to bonds and re-balance into stocks only with the signal. This might be different if I were making lump-sum contributions. The reason is if stocks are running higher there should be no need to add money to the stock portion.

  11. Thanks so much for your response! Everything sounds reasonable.

    Yes, I’m investing on the London Stock Exchange with UCITS ETFs, mainly for tax reasons. Just wondering why you asked, does it affect anything? The LSE actually has a 5x S&p 500 etf, SG92 – with tiny AUM, unsurprisingly. Not something I’m going to dabble with personally!

    When you mention little room for error, did you mean with regards to the margin cushion or to the need for a precise stop-loss order, or both? Also, do you worry about flash crashes triggering stop-loss orders?

    I think one of the problems with adopting a LBP for the casual do it yourself like me is simply the lack of ease in backtesting, especially with rebalancing and leverage – the leveraged ETFs don’t go back so far and they have limited offerings, and using margin would also create differences. Do you have any idea how performance would change by increasing leverage from 3x to 4x or 5x and reducing the capital at risk?

  12. Daren (Editor) says:

    UCITS ETFs are clearly the way to go for expat investors. Generally 15 percent withholding tax applies instead of 30, a pretty significant savings. I will be shifting from BSV to a UCITS alternative in the coming months.
    I took a look at SG92. Potentially great product for a LBP portfolio, but only trades once every few days. It needs more liquidity. Somewhat an aside, in my backtests, I generally see 5x daily leveraged ETFs as being the absolute cap on broad market indices.
    There seems to be little room for error on margin cushion (margin call risk) and prices gapping. If you get gaps down, they could devastate your portfolio and force selling in your bonds as well. A core idea of LBP is that your bonds stay intact and are not harmed by your stock investing. I’m thinking using e-mini futures like ES could be a better alternative if you want to avoid leveraged ETFs for liquidity reasons. ES is very liquid, trades 24/5 and allows for quite substantial leverage while still maintaining capital efficiency.
    It is relatively easy to simulate daily leverage in most major indices. I find that 3x to 5x is the magic window. 2x doesn’t allow for enough capital efficiency and more than 5x starts to get too affected by daily market moves in more volatile periods causing excessive drag on returns. 4x is basically a perfect balance.

  13. Thanks again for sharing your knowledge with us Daren. I’ve decided to go 100 % All-World/ACWI (20% x5) and 80 % treasury fund for the LBP half of my portfolio, leveraging up 40% on margin in total and use your idea of a 13 week SMA signal change for rebalancing. I’ll set price alerts instead of a stop loss as I’m too afraid of a flash crash situation robbing me. I don’t meet the net wealth requirements yet for options or futures trading on IB yet, and as I’m still in my 20s I figure I’ll wait another year or two to save more and learn more about these tools before I consider them. Hopefully this will let me capture most of the upside of buy and hold while avoiding some of the downside.

    How would IB’s CFDs on ETFs work compared to futures or options? They aren’t available to US, Canadian or Hong Kong (where I live) residents, but since you will be residing in Vietnam I think you could use them.. They seem to be a very convenient and accurate way to get leverage in a LBP.

  14. Daren (Editor) says:

    That sounds pretty reasonable Jon. It should work well with price alerts set a few percentage points above your stop loss.
    I haven’t looked a lot at CFDs because I can’t trade them in Canada. But they do look very appealing for capital efficiency. My primary concern is that the brokerage takes the other side of the trade. I don’t know how liquid these products are in times of crisis. I suppose I will need to look into them more closely as another option in Vietnam.

Comments are closed