Disclaimer from [[Investing Playbook]] still applies. :) # Introduction - The Dual Momentum investing strategy was invented by the economist Gary Antonacci. He has published a book on his work in 2014: [Dual Momentum Investing: An Innovative Strategy for Higher Returns with Lower Risk - Amazon](https://www.amazon.com/Dual-Momentum-Investing-Innovative-Strategy/dp/0071849440) - The core idea of momentum strategies in general is that something that has grown for some time will continue to grow for some time and thus it should be possible to follow the trend and perform better than the SP500 over the long term. - Antonacci pushes the momentum strategies further by combining absolute momentum and relative momentum to improve performance. # GEM - Dual Momentum - Global Equities Momentum - [Extrategic Dashboard - Dual Momentum (Updated)](https://extradash.com/en/strategies/models/10/dual-momentum/) ## Introduction video - "More Profit with Less Risk through Dual Momentum" by Gary Antonacci (2016) ["More Profit with Less Risk through Dual Momentum" by Gary Antonacci - YouTube](https://www.youtube.com/watch?v=LMIo3-S3jfk) - Newton: "*A body in motion tends to stay in motion*" - Riccardo: "*Cut your losses short, and let your profits run on*" - Alfred Cowles III & Herbert Hones, NYSE stocks from 1920 to 1935, in 1937: "*The tendency is very pronounced for stocks that have exceeded the median in one year to exceed it also in the following year.*" - Fama & French: "*The premier market anomaly is momentum. Stocks with low returns over the past year tend to have low returns for the next few months, and stocks with high past returns tend to have high future returns.*", "Dissecting Anomalies", Journal of Finance, July 2008 - 2014 Dalbar Report - Past 20 year annualized returns: - Average U.S. equity fund investor: 5% - S&P 500 index: 9.2% - 2 types of momentum - Relative (cross-sectional): compare performance to our peers - Absolute (time-series): compare performance to ourself ![[Dual Momentum - S&P 500 Absolute Momentum.png]] - Dual Momentum is similar to moving average, but better 1. Absolute momentum: First we use momentum to see if we want to be in stocks or bonds (Barclays US Aggregate Bond index) 2. Then, relative momentum: we pick the best index between the S&P500 and the MSCI ll Country World Index (ACWI) ex-US 3. Finally, monthly rebalancing, 12-month look back ![[Dual Momentum vs Relative & Absolute Momentums.png]]![[Dual Momentum vs Relative & Absolute Momentums 2.png]] - CAGR, Sharpe ratio & Worst drawdown are much better with Dual Momentum. And compared to stock momentum (picking individual stocks and doing momentum), there is no scalability issue, you can put as much money as you want in the index. - 1.3 trade / year on average. - And you avoid bear markets for the most part. - Advantages of Dual Momentum: - High returns - High scalability - Low trading costs - Low left tail risk - Risks: - Tracking Errors: 1979-1980 and 2009-11 - Whipsaws and lags in bull market - Short term volatility: it fuels long term returns - 3 secrets of success in Dual Momentum: - Patience - Discipline to stick with the model - Understanding - How do mean-reverting theories (something that has performed well will then underperform) coexist with momentum theories (something that has performed well will continue to perform well)? - Both are valid. Just not on the same time scale. - Mean-reversion is more the foundation of value investing and works on a 3-5 year horizon, with the goal of buying the dips. - Mean-reversion also applies to the short term, with periods of one month or less. - Momentum is an intermediate situation: 3-12 months. - If volatility is super high, the model doesn't work as well. Thus focusing on individual stocks performs less. - 12 month momentum allows to do less trades, thus less fees. ## Backtests - [Gary Antonacci - Extended Backtest of Global Equities Momentum - DUAL MOMENTUM](https://dualmomentum.net/2018/10/16/extended-backtest-of-global-equities-momentum/) - [Geczy & Samonov](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2607730) (2015) show that momentum consistently outperformed buy-and-hold back to the year 1801. - Momentum applied to geographically diversified stock indices outperformed momentum applied to stocks, bonds, commodities, currencies, and sectors.  That is how we use momentum. - **GEM components**: - US stocks: S&P 500 index for U.S. stocks - Non-U.S. stocks: MSCI All Country World Index ex-U.S - Bonds : Barclays U.S. Aggregate Bond Index (pas un actif sans risque, donc bien une stratégie différente de Duo Momentum de l'étude précédente) - **GEM management strategy** - When the trend of stocks is up according to absolute momentum applied to the S&P 500, we use relative strength to determine if we will be in U.S. or non-U.S. stocks. - When the trend of stocks is down, we invest in bonds. - We use a 12-month lookback period for both types of momentum and rebalance monthly. - [Extended Backtest of Global Equities Momentum - DUAL MOMENTUM](https://dualmomentum.net/2018/10/16/extended-backtest-of-global-equities-momentum/) - Backtest that starts in 1950, to 2018. ## How to reproduce this strategy + Real time - [TrendXplorer: Prospecting Dual Momentum With GEM](https://indexswingtrader.blogspot.com/2016/10/prospecting-dual-momentum-with-gem.html) + GEM in real time - **Important point:** In contrast to common practice, GEM’s official rules administer absolute momentum before the application of relative strength momentum. - According to recently confirmed academic research the U.S. stock market tends to lead the rest of the world especially during recessions, hence the trend of stocks is determined first using just the S&P 500 index. Only when an uptrend in the S&P 500 is identified by comparing its 12-month total return against that of the 3-month U.S. T-Bill, relative strength momentum is applied to select the best performing stock index for allocation: the S&P 500 index or the All Countries World Index ex U.S. Otherwise the U.S. Aggregate Bond index is selected for capital preservation. - Regarding this point, this is exactly what the following two backtests examine: - GEM Reproduction: [Global Equities Momentum: performance analysis - Le Quant 40](http://www.lequant40.com/2016/09/global-equities-momentum-analyse-des.html) - GEM Reproduction 2: [Global Equities Momentum: a review of performance analysis - Le Quant 40](http://www.lequant40.com/2016/09/global-equities-momentum-retour-sur.html) - => Spreadsheet: [Global Equities Momentum - Reproduction - Google Sheets](https://docs.google.com/spreadsheets/d/1XBXQA_ZWTVdkwsZaSvBlzaoeRh4ODQiY9pHtru0h9Xg/edit#gid=333408813) - The difference between the _Reproduction GEM_ strategy and this new strategy, which I will call _Reproduction GEM 2_, lies in the order of comparison of the absolute and relative performances of the S&P 500 and MSCI ACWI ex-USA indices: - In the _Reproduction GEM_ strategy, the relative return of the two equity indices is compared first, and then it is checked that the absolute return of the better equity index is positive before investing. - In the _Reproduction GEM 2_ strategy, the absolute return of the S&P 500 index is first checked to be positive before investing in the best of the two equity indices. ## Fragility analysis of GEM (2019) - [Fragility Case Study: Dual Momentum GEM | Flirting with Models](https://blog.thinknewfound.com/2019/01/fragility-case-study-dual-momentum-gem/) - To be clear, no amount of diversification will protect you from the risk of the _style._ As we like to say, “risk cannot be destroyed, only transformed.”  In that vein, trend following strategies will always incur some sort of whipsaw risk.  The question is whether it is whipsaw related to the style as a whole or to the specific implementation. - => A solution could be to do dollar cost averaging to start GEM strategies. Don't invest everything at first, but something 5-10% of portfolio every month, to decrease any whipsaw risk (market changing direction rapidly). ## E-GEM - Enhanced Dual Momentum Models - There are improved versions of GEM (the Dual Momentum strategy presented in the book), but they are not available to the general public. - [Dual Momentum Performance - Optimal Momentum](https://www.optimalmomentum.com/dual-momentum-proprietary-models/): - Enhanced Global Equities Momentum (E-GEM) - 16.8% CAGR - E-GEM is an enhanced version of our book’s GEM model. E-GEM includes multiple criteria for deciding when to be in stocks or bonds. E-GEM is for aggressive investors willing to accept the short-term volatility that comes from investing in the stock market. - Enhanced Global Balanced Momentum (E-GBM) - E-GBM is a balanced allocation between stocks, bonds, and other assets. E-GBM is for investors with moderate risk tolerances. It is a general-purpose model suitable for those wanting exposure to a variety of assets and less volatility than E-GEM. ![[Enhanced Dual Momentum Models.png]] - [Global Equities Momentum - Optimal Momentum](https://www.optimalmomentum.com/global-equities-momentum/) # Details on Dual Momentum strategy ## Only with stocks or with other asset classes? - In his white paper ([[Investissement/Dual Momentum/Risk Premia Harvesting Through Dual Momentum - Antonacci.pdf]]), Antonacci clearly states that the Sharpe Ratio when combining different asset classes is better than being 100% invested in Dual Momentum stocks only, for example. - However, on his website (GEM), Antonacci only discusses his strategy with stocks exclusively. - [Forum - View Thread](https://www.portfolio123.com/mvnforum/viewthread_thread,11513) - Other comments online: - 1. I've seen some conflicting information in Gary's research about whether one should invest in cash (SHY) or aggregate bonds if neither equity market satisfies the absolute and relative criteria. But in his book he recommends aggregate bonds so I'm following that advice here. - 2. In his original white paper, he also revealed a composite index (4 asset classes: Equities, Credit, Real Estate and "Stress") using the same momentum principles across all 4 asset classes, with very good absolute and relative performance. ## Dual momentum works better if done in the US stock market - [Dual Momentum for non-US Investors](https://www.optimalmomentum.com/dual-momentum-for-non-us-investors/) ## Official FAQ on Dual momentum - [FAQ - Optimal Momentum](https://www.optimalmomentum.com/faq/) - Here are two of these references that are especially relevant.  They contain long-term validation of both relative and absolute momentum. - Geckzy & Samonov (2017), - Greyserman & Kaminsky (2014). ## Other pros & cons? [What are the pros and cons of the 'dual momentum' investment strategy? - Quora](https://www.quora.com/What-are-the-pros-and-cons-of-the-dual-momentum-investment-strategy) # New Dual Momentum strategies? Not as convincing ## Accelerated Dual Momentum (2018) - **TLDR;:** - The major difference with classic Dual Momentum is that instead of using a 12-month momentum, this model combines 1-month, 3-month, and 6-month momentum (each with a 33% weight) in order to react more quickly to trend changes while avoiding false signals (and thus unnecessary trades). - **However, the results for the Accelerated Dual Momentum strategy seem quite optimistic and based on extrapolated data:** - Actively managed ETF - A data history limited to only 20 years (compared to 50 years for GEM) - Trades are made on the last day of the month, whereas if they are made at another time, performance drops significantly. - Charts - [EP's Accelerating Dual Momentum - TuringTrader.com](https://www.turingtrader.com/ep-accelerating-dual-momentum/) - [Accelerating Dual Momentum Performance | line chart made by Swhanly | plotly](https://chart-studio.plotly.com/~swhanly/7/#plot) - [Accelerating Dual Momentum - PortfolioDB](https://portfoliodb.co/portfolios/accelerating-dual-momentum/) - Original Accelerated Dual Momentum paper: [Accelerating Dual Momentum Investing – engineered portfolio](https://engineeredportfolio.com/2018/05/02/accelerating-dual-momentum-investing/#data) - Now let’s take the “original” dual momentum strategy and make some tweaks. The links are to the individual backtests in Portfolio Visualizer, final balances shown are for a $10,000 investment at the beginning of 1998 growing until February 2018. - $80,296 – Original Strategy - $159,464 – Simple Accelerating Dual Momentum - Same assets as original but use 1-month, 3-month, and 6-month returns - $206,161 – Replace Total Bond with Long-Term Treasuries - $331,003 – Replace Global Large Cap with Global Small Cap - $426,408 – Accelerating Dual Momentum - Make both replacements above, global small cap and long-term treasuries ### An analysis on Accelerated Dual Momentum [Accelerating Dual Momentum - AllocateSmartly](https://allocatesmartly.com/taa-strategy-accelerating-dual-momentum/) - Unlike traditional dual momentum, the strategy measures momentum across multiple, shorter-term time frames, making ADM more responsive to market changes. - Accurate data for international small-cap equities doesn’t exist that far into the past. As a result, we were only able to extend this test to 1990. That matters. - So while we think that the idea of replacing international large-cap exposure with small might have merit, readers should understand the limitations on historical data that decision introduces, and temper expectations accordingly. - Readers know that we do something pretty nifty on our site for strategies like ADM that trade once per month: we show the results of trading on other days of the month as well. We’re not simply moving the date of execution forward. It’s a more complicated approach that maintains the “integrity” of the monthly data interval, but on other trading days - First, there’s a wide disparity in results. That’s not uncommon among strategies that hold concentrated positions using shorter-term indicators like ADM. The difference in performance trading on one day to the next can be drastic. Of course, members can mitigate that risk with portfolio tranching. - More important is the fact that there’s a wide disparity in results and the author’s original end-of-month variation happens to be the best of the bunch. Yes, trading on days near the beginning and end of the month tends to be good for TF/momentum strategies (read more), but the fact that EOM is an outlier suggests that these results may be somewhat over fit to history. - That doesn’t mean the strategy doesn’t have value, only that these results may be a tad optimistic. ### Response of Gary Antonacci on Accelerated Dual Momentum - 2018-09-13 [Perils of Data Mining - DUAL MOMENTUM](https://dualmomentum.net/2018/09/13/perils-of-data-mining/) - _**Shorter lookback periods**_ - In my book, I show that a 12-month lookback period outperformed 3, 6, and 9-month lookback periods with GEM. This lookback period has held up well since it was first identified by Cowles  & Jones in 1937. My website’s [FAQ](http://www.optimalmomentum.com/faq.html) page describes in more detail the advantages of a 12-month lookback. - Yet there are those who believe that because shorter lookbacks are more sensitive to market changes, they should give better results. - A 3-month look back did do well over the past 20 years. If you were to look only at that data, you might feel reassured about using a shorter lookback period. But this starts to unravel in 1979-80 when the markets were choppy. Choppiness gives lower returns and higher drawdowns. - **_EAFE instead of ACWI ex-U.S._** - But emerging markets and Canada missing from the MSCI EAFE make up a significant 24% of the MSCI ACWI ex-US index. Using the broader MSCI ACWI ex-U.S. avoids selection bias by being all-inclusive. - _**U.S. small and mid-cap stocks**_ - The reason broader indices give worse results may be due to there not being a small-cap premium despite many who think otherwise. See [here](https://aswathdamodaran.blogspot.com/2015/04/the-small-cap-premium-fact-fiction-and.html) and [here](https://www.researchaffiliates.com/en_us/publications/articles/284_busting_the_myth_about_size.html) for more on this. - **_Long term bonds_** - Some prefer to use long-term Treasury bonds as a safe harbor when they exit stocks because stocks and bonds have been negatively correlated. They think they will thus earn better returns being in long-duration bonds when stocks are weak. - It is true that stocks and bonds have been negatively correlated in recent years. But that has not always been the case. In fact, stock-bond correlations are as likely to be positive as negative over the long run. - On a risk-adjusted basis, 20-year Treasuries did not outperform intermediate-term aggregate bonds despite low stock-bond correlations and a strong bull market in bonds over the past 35 years. Under normal market conditions, long term bonds with their higher risk are more likely to be at a disadvantage as a safe harbor asset. - **_Recent data mining example_** - I have gotten some emails asking about a strategy called “Accelerating Dual Momentum” that was inspired by GEM. This model looks at 20 years of mutual fund data. Based on that data, the developer uses long-term Treasury bonds as his safe harbor asset and a combination of short (1, 3, and 6 months) lookback periods. We have seen that these lookbacks may not be the best choices based on long term data. We also know that intermediate-term bonds outperformed long-term Treasury bonds over a longer period. Using a combination of shorter lookback periods is not accelerating dual momentum. It is dual momentum with shorter lookback periods. - The developer also questions using the MSCI ACWI ex-U.S. index of large and mid-cap stocks as the best vehicle for non-U.S. equities. His argument is that companies are more globalized now so the correlation between U.S. and non-U.S. companies is higher than it once was. This may be true. But the following chart from my website’s [FAQ](http://www.optimalmomentum.com/faq.html) page shows something else. The relative strength difference between U.S. and non-U.S. equities is due mostly to macro-economic conditions reflected in the strength or weakness of the U.S. dollar. - At the end of his discussion, the developer presented a chart showing the rolling real return of GEM versus his model back to 1871. But prior to 1970, he does not use international stocks. There was thus no dual momentum. He had only absolute momentum with the S&P 500 and government bonds (10-year bonds after 1953 and longer-term government bonds prior to that). He used bond yields and not their total return. He further misrepresented GEM by using EAFE instead of ACWI ex-US when he does use international stocks from 1970 forward. We saw earlier that the narrower EAFE index understates GEM performance by130 bps annually from when ACWI ex-US was introduced in late 1988. - Even after his data-mining efforts and distortions of GEM and dual momentum, the developer shows that the trailing 30-year annualized real returns of both models since the mid-1990s was about the same. ## Dual Momentum with Sector rotation ### Sector ETF Dual Momentum (2020) - Attempt at an improved version of Dual Momentum: - With sector ETFs (11) - Taking the top 4 each time - Changing the momentum durations analyzed - Issue: We only have data for the moment until 2004/2005, so we don't see performance across the 4 types of economic cycles (described in [[Permanent Portfolio - Investing strategy]]). - 2020-06-15 [Vanguard ETF Portfolio: Enhanced Momentum Strategies | Seeking Alpha](https://seekingalpha.com/article/4353915-vanguard-etf-portfolio-enhanced-momentum-strategies) - 2020-06-25 [Vanguard Sector ETFs Portfolio: Enhanced Momentum Strategies | Seeking Alpha](https://seekingalpha.com/article/4355636-vanguard-sector-etfs-portfolio-enhanced-momentum-strategies) - 2021-02 [Vanguard Sectors ETF Portfolio: New Momentum Strategies | Seeking Alpha](https://seekingalpha.com/article/4406225-vanguard-sectors-etf-portfolio-new-momentum-strategies) - The eleven sector funds are used during risk-on periods, while the treasury funds are used during risk-off. The total bond market fund together with the materials sector ETF and the metals ETF are used to determine the state of the market. Vanguard Communication Services Index Fund (NYSEARCA: VOX) Vanguard Consumer Discretionary Index Fund (NYSEARCA: VCR) Vanguard Consumer Staples Index Fund (NYSEARCA: VDC) Vanguard Energy Index Fund (NYSEARCA: VDE) Vanguard Financials Index Fund (NYSEARCA: VFH) Vanguard Health Care Index Fund (NYSEARCA: VHT) Vanguard Industrials Index Fund (NYSEARCA: VIS) Vanguard Information Technology Index Fund (NYSEARCA: VGT) Vanguard Materials Index Fund (NYSEARCA: VAW) Vanguard Real Estate Index Fund (NYSEARCA: VNQ) Vanguard Utilities Index Fund (NYSEARCA: VPU) iShares 7-10 Year Treasury Bond ETF (NYSEARCA: IEF) iShares 20+ Year Treasury Bond ETF (NYSEARCA: TLT) Invesco Base Metal Fund (NYSEARCA: DBB) Materials Select Sector SPDR Fund (NYSEARCA: XLB) Vanguard Total Bond Market Index (MUTF: VBMFX) - Initial Balance: $1,000 - Absolute momentum period: 3 months - Relative strength period: 3 months - Assets to Hold: 4 - Switching Frequency between risk-on and risk-off: Monthly - Trading Frequency during risk-on: Monthly - => 22.36% CAGR since 2008 ### Dual Momentum Sector Rotation by Antonnaci (2015) - There is also the Dual Momentum Sector Rotation (DMSR) model, conceived by Gary Antonacci, which involves Dual Momentum and investing in 4 rising sectors during up periods (rather than just the SPY). - Gary Antonacci's analysis: [Bring More Data - DUAL MOMENTUM](https://dualmomentum.net/2015/11/21/bring-more-data/) - By adding more years of historical data, it becomes apparent that the classic GEM model of Dual Momentum performs better than the one with Sector Rotation. - The monthly correlation between GEM and DMSR is only 0.59, sector rotation may still have a modest role to play in a diversified equities-oriented portfolio. But DMSR is not the best choice as a core portfolio holding. Sector rotation programs that use data no further back than the early 1990s to develop their models may be in for a rude awakening someday. Future drawdowns may be higher and returns may be lower than expected. - Along the same lines, there are also momentum-based portfolios popping up on the internet all the time now, some even labeled as “dual momentum,” modeled on the basis of only 10 or 15 years of ETF data. Momentum may be robust enough that future results won’t suffer much because of this. But those who think they are constructing optimal models this way are just fooling themselves. Overfitting modest amounts of data is one of the most pernicious problems in the development of investment models.