Dual Momentum Investing

[willthrill81]

Since you only get one shot at investing in your life, using a strategy that might not work during your lifetime doesn't seem a very well thought out method. Besides, you might end up abandoning it at the wrong time, after many years of underperfomance, because you think it's now stopped working.

Precisely the same argument could be used with regard to buy-and-hold of any chosen AA; there are no guarantees that any specific strategy will meet one's goals going forward. All roads carry risk.

[Lauretta]

Since you only get one shot at investing in your life, using a strategy that might not work during your lifetime doesn't seem a very well thought out method. Besides, you might end up abandoning it at the wrong time, after many years of underperfomance, because you think it's now stopped working.

Precisely the same argument could be used with regard to buy-and-hold of any chosen AA; there are no guarantees that any specific strategy will meet one's goals going forward. All roads carry risk.

Yes, but for B&H you don't incur taxation every time you sell with a profit (I think Paul wrote he uses DM in his taxable account too); you don't have losses for bid-ask spread etc. Psychologically it's also easier for a number reasons other posters have also explained.
IMO the wise thing to do is to diversify across strategies, not go 100% in DM

[Solo Prosperity]

I am a 55 year old still working in the Tech Industry. After a couple of months of research I took a plunge and implemented “Dual Momentum”. I have convinced myself enough that I have put 100% of my tax sheltered as well as taxable liquid assets to use this method . It has been about four months since I started. It has been an unusually good stock market - so far so good. Even though I think I have thought about this investment method from every angle, understandably I am still a little bit nervous. But I am quite optimistic about my prospects. If you think I am a crazy person to take such a plunge, please let me know in a constructive way.

As a fellow DM investor, I will not be touching on the criticisms of the strategy in any detail. There are downsides, as there are with every strategy, and yogi's amazon review covers many such as: Backtests v. Future Performance, Whipsaw Trades (Shifting Risk), Tax-Efficiency (Although slightly overblown as most positive trades get taxed at LT CGs), Increased Trading Costs (Again, slightly overblown given the investment vehicles used, but could change in the future), etc.

What I will suggest is that even with a strong faith in an investment strategy (Required), I would be hard-pressed to ever put 100% of my savings into a single strategy. This is something I would say to B&H Market-Cap Weighted investors as well. We have ~100 years of solid US data, ~50 years of solid Developed Markets data and ~40 years of solid Emerging Markets data. All of our investment beliefs come from back-tests/historical returns of a very small sample size. This leads me to be very aware of how I invest in any strategy. This is why I believe in diversity by strategy and factors, not just asset class and geographies.

Yes, I have an overall big-picture view of my global asset allocation of stocks and bonds, but I also have exposure to value, momentum & trend via different strategies. One of which is Dual Momentum (20%). One of which is a slight variation of Trending Value (15%). Another is Meb Faber's Global Value strategy (15%). The last of which is a simple Boglehead B&H portfolio (50%). All strategies are implemented with cost's (Both $ and emotions) in mind. I have been very happy with this setup and will continue this approach. Some years my strategies lose to B&H and some years they beat it.

Much like Benjamin Graham recommended to never have more than 75% or less than 25% in stocks, I have a slightly modified belief about investment strategies (Including Passive Market-Cap Weighted)...Never have more than 75% in a single investment strategy but Fun/Gambling/Active strategies can and probably should be less than 25% imo. They can fall out of favor for as long as 2-3 decades and could severely impact your path towards Financial Independence.

Unlike back-tests, we do not have 100 years for a strategy to work-out. By the time that most of us have any level of serious money to get compounding on for our FI, we get about 20, 30, maybe 40 (Max) years before the need to start de-risking begins. Putting all of that chance into one strategy is not a risk I would be willing to take.

Note: I understand most people don't want to take the time to manage active investment strategies and I get that. They take less time than most people think (I spent about 10-15 minutes a month on my portfolio throughout the year) but they can be a nuisance to folks who have no interest in markets beyond saving/investing for the future as they do require quite a bit of up-front research. For the hardened Boglehead, I would still caution being 100% invested in Market-Cap weighted funds, and would suggest people at least learn about factor diversification, but a simple 3-Fund is probably the best strategy for a completely "set it and forget it" mentality.

[kober.paul]

Hi Paul,

being 100% in Dual momentum seems to me irrational since, as for example Clifford Asness noted in one of his papers, 'while momentum has performed well over the long-term, it has suffered periods of sharp underperformance'.
Thus if you begin the back-tests in the 1970s and end today, you get good results.
But if you begin in the late 1940s and end today, you don't.
If you go back to the 1920s, the results improve again.

Since you only get one shot at investing in your life, using a strategy that might not work during your lifetime doesn't seem a very well thought out method. Besides, you might end up abandoning it at the wrong time, after many years of underperfomance, because you think it's now stopped working.

I am also curious, as I have seen on Amazon people like 'Max Henke' spend a lot of time and effort defending Antonacci's book in the review section, and now I see new people registering on Bogleheads and doing the same. May I ask what your motivations are for defending this book? In the end having many people acting as a mouthpiece for the author (so to speak) may have the opposite from the desired effect. At least it has done so for me.

I am going to be exploring a bit more about back testing beginning late 1940s. Thanks for pointing me that.

The main reason I am investing according to DM is precisely because based on my research I believe it to provide me with best chance of investment success and that I only have one shot at investing. No matter which methodology we choose, there will always be some possible outcomes which are undesirable. Given that we assign likelihoods to various outcomes and if the worst outcomes among them is acceptable. One of the reasons DM is attractive to me because of it protection against huge drawdowns.

I can't answer your last question with regard to why many people are acting as a mouth piece etc., I am an investor who chose DM for my investment strategy and I seek people to challenge my notions so that I become more confident, hence I am more likely to stick to it when going gets tough which is inevitable. Whether or not there are such mouth pieces and their motivations, we can still have rational discussions.

Thank you for your response.

[Lauretta]

I am an investor who chose DM for my investment strategy and I seek people to challenge my notions so that I become more confident, hence I am more likely to stick to it when going gets tough which is inevitable.

Yes I also think it's a great idea to seek people to challenge your notions, I am doing the same also on other subjects in this forum Anyway, good exchanging with you.

[kober.paul]

As a fellow DM investor, I will not be touching on the criticisms of the strategy in any detail. There are downsides, as there are with every strategy, and yogi's amazon review covers many such as: Backtests v. Future Performance, Whipsaw Trades (Shifting Risk), Tax-Efficiency (Although slightly overblown as most positive trades get taxed at LT CGs), Increased Trading Costs (Again, slightly overblown given the investment vehicles used, but could change in the future), etc.

What I will suggest is that even with a strong faith in an investment strategy (Required), I would be hard-pressed to ever put 100% of my savings into a single strategy. This is something I would say to B&H Market-Cap Weighted investors as well. We have ~100 years of solid US data, ~50 years of solid Developed Markets data and ~40 years of solid Emerging Markets data. All of our investment beliefs come from back-tests/historical returns of a very small sample size. This leads me to be very aware of how I invest in any strategy. This is why I believe in diversity by strategy and factors, not just asset class and geographies.

Yes, I have an overall big-picture view of my global asset allocation of stocks and bonds, but I also have exposure to value, momentum & trend via different strategies. One of which is Dual Momentum (20%). One of which is a slight variation of Trending Value (15%). Another is Meb Faber's Global Value strategy (15%). The last of which is a simple Boglehead B&H portfolio (50%). All strategies are implemented with cost's (Both $ and emotions) in mind. I have been very happy with this setup and will continue this approach. Some years my strategies lose to B&H and some years they beat it.

Much like Benjamin Graham recommended to never have more than 75% or less than 25% in stocks, I have a slightly modified belief about investment strategies (Including Passive Market-Cap Weighted)...Never have more than 75% in a single investment strategy but Fun/Gambling/Active strategies can and probably should be less than 25% imo. They can fall out of favor for as long as 2-3 decades and could severely impact your path towards Financial Independence.

Unlike back-tests, we do not have 100 years for a strategy to work-out. By the time that most of us have any level of serious money to get compounding on for our FI, we get about 20, 30, maybe 40 (Max) years before the need to start de-risking begins. Putting all of that chance into one strategy is not a risk I would be willing to take.

Note: I understand most people don't want to take the time to manage active investment strategies and I get that. They take less time than most people think (I spent about 10-15 minutes a month on my portfolio throughout the year) but they can be a nuisance to folks who have no interest in markets beyond saving/investing for the future as they do require quite a bit of up-front research. For the hardened Boglehead, I would still caution being 100% invested in Market-Cap weighted funds, and would suggest people at least learn about factor diversification, but a simple 3-Fund is probably the best strategy for a completely "set it and forget it" mentality.

Thank you very much for your thoughts. I hope I am able to spend a lot more time and understand various strategies that you are using. A couple of my thoughts reading your post ...

1. Whichever strategy that I use, in addition to sufficient back testing I also need good theoretical underpinnings. For example, using B&H because you believe in EMH. Using momentum strategies because you believe that momentum is a sustainable anomaly and you understand why it exists in the first place and why you think it is sustainable etc.. It does take a lot of time and effort to research for strategies that you feel comfortable about.

2. When you are diversifying with asset classes, you get your free-lunch because your assets are either poorly or negatively correlated. Is there such a thing when you are diversifying with strategies? I am not challenging this at all - it makes common sense ... but are there any papers or books that you like that talk to diversifying strategies and how to do it?

3. Am i willing to put 50% in B&H for strategy diversification? I will feel more comfortable for sure. But the additional alpha that DM seem to promise is too tempting.

Once again, thank you so much for taking time to write such a detailed response. It is very thought provoking for me.

[LadyGeek]

^^^ I want to make sure you're aware of our backtesting project: Simba's backtesting spreadsheet

Questions and comments can be posted in the support thread: Simba's backtesting spreadsheet [a Bogleheads community project]

[patience...]

Yes, but for B&H you don't incur taxation every time you sell with a profit (I think Paul wrote he uses DM in his taxable account too); you don't have losses for bid-ask spread etc. Psychologically it's also easier for a number reasons other posters have also explained.
IMO the wise thing to do is to diversify across strategies, not go 100% in DM

What makes you think B&H is psychologically easier to implement?
If positions are held for longer than a year (or in IRA accounts) there is no tax difference. I wouldn't base my decision between B&H and momentum on this.
And transaction costs using ETFs really? Unless you have a large multi-million dollar portfolio > $10M I don't see the large cost.

[willthrill81]

Yes, but for B&H you don't incur taxation every time you sell with a profit (I think Paul wrote he uses DM in his taxable account too); you don't have losses for bid-ask spread etc. Psychologically it's also easier for a number reasons other posters have also explained.
IMO the wise thing to do is to diversify across strategies, not go 100% in DM

What makes you think B&H is psychologically easier to implement?
If positions are held for longer than a year (or in IRA accounts) there is no tax difference. I wouldn't base my decision between B&H and momentum on this.
And transaction costs using ETFs really? Unless you have a large multi-million dollar portfolio > $10M I don't see the large cost.

I tend to agree, though trend following does face a greater headwind in taxable accounts. There are many situations in which trades are made more often than once per year, even with long timing periods. And note that long-term capital gains tax rates are not zero for everyone; only those in the 12% bracket or lower get that benefit.

[Riley15]

I had posted this in another thread but I think it's more relevant to this topic.

As to using moving averages to time the market or make buy sell/decisions. Any type of moving average is just that. A statistically calculated number used for technical analysis, it may not have any relevance to increasing or decreasing returns in any meaningful way when used. It's not any different then 52 week lows/highs or round numbers that end in 00. Which type of moving average would be ideal?? There's simple moving average, weighted average, exponential moving average, etc.. They may all have different outcomes in various time periods.

[aegis965]

Diversify your signal.

[willthrill81]

I had posted this in another thread but I think it's more relevant to this topic.

As to using moving averages to time the market or make buy sell/decisions. Any type of moving average is just that. A statistically calculated number used for technical analysis, it may not have any relevance to increasing or decreasing returns in any meaningful way when used. It's not any different then 52 week lows/highs or round numbers that end in 00. Which type of moving average would be ideal?? There's simple moving average, weighted average, exponential moving average, etc.. They may all have different outcomes in various time periods.

This is a common criticism of any trend following method, but over 100 years of data have shown it to be a straw man.

Long periods of historical analysis have shown that most methods lead to similar outcomes. Yes, there will be differences created by using a 3 month simple moving average instead of a 15 month exponential moving average, but over long-term periods, they have a surprisingly strong tendency to converge. Meb Faber examined many of these over a decade ago (2006) in his now famous paper. Since that time period, the simple trend following method he outlined in that paper (10 month simple moving average) has had higher absolute and risk-adjusted returns than buy-and-hold and with much lower drawdowns. Further, most of the moving averages used are not "statistically calculated."

People have been using metrics like the 200 day moving average for at least the last 60 years. And it has continued to perform at least as well as buy-and-hold over long-term periods since then.

I've never completely understood this criticism. It's a bit like saying that because different AAs have different returns (including some bond heavy portfolios occasionally outperforming stock heavy portfolios, as in 2000-2009) over different periods that AA doesn't matter or is just a bogus idea altogether.

Diversify your signal.

I think that's a fine approach. For instance, one could use a 50 day moving average, a 100 DMA, and a 200 DMA simultaneously, each used for one-third of the portfolio.

That being said, over the long-term, all of these timing periods tend to lead to similar results.

[Riley15]

Long periods of historical analysis have shown that most methods lead to similar outcomes. Yes, there will be differences created by using a 3 month simple moving average instead of a 15 month exponential moving average, but over long-term periods, they have a surprisingly strong tendency to converge. Meb Faber examined many of these over a decade ago (2006) in his now famous paper. Since that time period, the simple trend following method he outlined in that paper (10 month simple moving average) has had higher absolute and risk-adjusted returns than buy-and-hold and with much lower drawdowns. Further, most of the moving averages used are not "statistically calculated."

People have been using metrics like the 200 day moving average for at least the last 60 years. And it has continued to perform at least as well as buy-and-hold over long-term periods since then.

I can understand lower drawdowns should definitely reduce volatility drag as you stated earlier. Actually that is probably the biggest advantage, I think many people overlook how volatility effects long term returns but it's one of the biggest obstacles to compounding. So much of planning is done using annual returns with compounding without considering volatility. That may be a topic for another thread..

But on the other hand, the market is known to be in a long-term drift upward so that fact itself would result in moving average underperforming buy and hold over the long term. So that may be a losing strategy in that regard. Would the advantage of reduced volatility drag outweigh the underperformance in a upward trending market?

I will review the paper when I get a chance, always interesting when there is academic literature to backup a theory.

Another sticking point for me is, when would you trade? As the market keeps drifting above and below the 200 day MA, would you keep buying and selling mindlessly. One would have to set limits on how often they trade.

[Lauretta]

Another sticking point for me is, when would you trade? As the market keeps drifting above and below the 200 day MA, would you keep buying and selling mindlessly. One would have to set limits on how often they trade.

In the models propounded by e.g. Faber and Antonacci you trade on a day of the month of your choice (usually at the end or the beginning of the month), thus once a month. As the link I gave in the thread on Larry Swedroe shows, there is a large spectrum of possible results in backtests, simply depending upon the day you trade. So this means there is a big element of luck.

[willthrill81]

But on the other hand, the market is known to be in a long-term drift upward so that fact itself would result in moving average underperforming buy and hold over the long term. So that may be a losing strategy in that regard. Would the advantage of reduced volatility drag outweigh the underperformance in a upward trending market?

In a long bull market such as we have experienced for the last nine years, virtually no trend following system will consistently outperform buy-and-hold. Historically, this underperformance has been compensated for by trend following's outperformance during bear markets. So the 'cost' of buy-and-hold is the occasional big drawdown that may take a long time to recover from, and the cost of trend following is underperforming the market during the good times. Individual investors need to be aware of these differences and, IMHO, select the strategy that works best for their personality and situation.

Again, this is somewhat analogous to selecting an AA when using a buy-and-hold approach. Over the long-term, you should expect a more stock-heavy portfolio to outperform a bond-heavy portfolio, but that doesn't mean that everyone needs an all stock portfolio (nor does it mean that stocks will always outperform bonds). Other important factors come into play here.

Another sticking point for me is, when would you trade? As the market keeps drifting above and below the 200 day MA, would you keep buying and selling mindlessly. One would have to set limits on how often they trade.

I only trade once per month, the first business day of the month, since I am only 'paying attention' to the closing price of the last trading day of the month. Most months, I do not make any trades at all.

If a market is moving 'sideways' as you describe, the frequent buying and selling is referred to as whipsawing. That effect is factored in to any backtested results. On average, trend following has been said to result in sub-optimal trades (compared to buy-and-hold) about 70% of the time. But the 30% of the time when it is 'right' at least makes up for this deficiency (i.e. small losses are compensated for comparatively big wins).

In the models propounded by e.g. Faber and Antonacci you trade on a day of the month of your choice (usually at the end or the beginning of the month), thus once a month. As the link I gave in the thread on Larry Swedroe shows, there is a large spectrum of possible results in backtests, simply depending upon the day you trade. So this means there is a big element of luck.

Faber actually found that most of the timing metrics used had similar long-term performance. Also, trading once per month as opposed to every time you got a buy/sell signal had very little impact on the long-term results.

[Lauretta]

In the models propounded by e.g. Faber and Antonacci you trade on a day of the month of your choice (usually at the end or the beginning of the month), thus once a month. As the link I gave in the thread on Larry Swedroe shows, there is a large spectrum of possible results in backtests, simply depending upon the day you trade. So this means there is a big element of luck.

Faber actually found that most of the timing metrics used had similar long-term performance. Also, trading once per month as opposed to every time you got a buy/sell signal had very little impact on the long-term results.

really? thanks for the information... If you read what I wrote above however, you'll see that's not what I was talking about. I was saying that if you designate a given day of the month to trade, your performance will greatly vary depending on whether you trade for example on the 1st or the 18th day of the month. Anyway best of luck.

[willthrill81]

In the models propounded by e.g. Faber and Antonacci you trade on a day of the month of your choice (usually at the end or the beginning of the month), thus once a month. As the link I gave in the thread on Larry Swedroe shows, there is a large spectrum of possible results in backtests, simply depending upon the day you trade. So this means there is a big element of luck.

Faber actually found that most of the timing metrics used had similar long-term performance. Also, trading once per month as opposed to every time you got a buy/sell signal had very little impact on the long-term results.

really? thanks for the information... If you read what I wrote above however, you'll see that's not what I was talking about. I was saying that if you designate a given day of the month to trade, your performance will greatly vary depending on whether you trade for example on the 1st or the 18th day of the month. Anyway best of luck.

The point I was making was that if it doesn't really matter over the long-term whether you use a 100 day moving average or a 300 DMA, nor does it matter if you trade once per month or multiple times per month, then it shouldn't really matter which day of the month you're making your trades either.

[Lauretta]

then it shouldn't really matter which day of the month you're making your trades either.

but it does matter:
https://alphaarchitect.com/2017/06/29/t ... th-matter/
you can also see this link I already provided yesterday:
https://blog.thinknewfound.com/2017/05/ ... e-details/

[willthrill81]

then it shouldn't really matter which day of the month you're making your trades either.

but it does matter:
https://alphaarchitect.com/2017/06/29/t ... th-matter/
you can also see this link I already provided yesterday:
https://blog.thinknewfound.com/2017/05/ ... e-details/

Interesting. That certainly flies in the face of the random walk hypothesis. I wonder if the effect they observed was also present in older data as well.

I wonder what the cause of the observed effect was. Perhaps trend followers trading based on the prior month's last closing prices?

[Lauretta]

then it shouldn't really matter which day of the month you're making your trades either.

but it does matter:
https://alphaarchitect.com/2017/06/29/t ... th-matter/
you can also see this link I already provided yesterday:
https://blog.thinknewfound.com/2017/05/ ... e-details/

Interesting. That certainly flies in the face of the random walk hypothesis. I wonder if the effect they observed was also present in older data as well.

I wonder what the cause of the observed effect was. Perhaps trend followers trading based on the prior month's last closing prices?

Yes it's interesting; like Walter Jones notes there's also the possibility of data overfitting.

[BlueEars]

Of course the outcome of monthly moving average methods is going to be somewhat dependent on the day of the month chosen. One only has to look at very sharp market declines like Oct 19 1987 (something like a -20% decline on that Black Monday). Selecting a mid-month day might well give a different outcome then a first of the month day as the market was somewhat gently rolling off in August-September 1987. So this should be unsurprising. But this is a very unusual example.

[willthrill81]

Of course the outcome of monthly moving average methods is going to be somewhat dependent on the day of the month chosen. One only has to look at very sharp market declines like Oct 19 1987 (something like a -20% decline on that Black Monday). Selecting a mid-month day might well give a different outcome then a first of the month day as the market was somewhat gently rolling off in August-September 1987. So this should be unsurprising. But this is a very unusual example.

That's a very good point. Meb Faber has noted that those who were using 10 month or longer moving averages as signals were in equities on Black Monday, but those using short-term moving averages were not. This is just an aberrant situation that is outside of one's control.

Regarding the 'end of month' effect, I really would like to see whether that could also be observed outside of their relatively narrow sample window.

[whodidntante]

For those who practice trend following, what are your related tools and information sources? I can certainly throw a 200 day SMA on an equity index ETF stock chart, but that is pretty cumbersome to do frequently and for several different asset classes.

[willthrill81]

For those who practice trend following, what are your related tools and information sources? I can certainly throw a 200 day SMA on an equity index ETF stock chart, but that is pretty cumbersome to do frequently and for several different asset classes.

Portfolio Visualizer is one of the most user friendly and free tools I've seen yet. It accommodates basic trend following, dual momentum, relative strength (what I use), and others.

[patience...]

Regarding the 'end of month' effect, I really would like to see whether that could also be observed outside of their relatively narrow sample window.

https://allocatesmartly.com/list-of-strategies/ has a comparative GEM backtest for all days of the month. One of the reasons I chose to use GEM for my investments is because the rebalance on the 1st day of the month does not show a large average return improvement. It is somewhere in the middle of the ranking. Most other asset allocation strategies backtested by AS have a much larger favorable returns skew.

[Riley15]

The point I was making was that if it doesn't really matter over the long-term whether you use a 100 day moving average or a 300 DMA, nor does it matter if you trade once per month or multiple times per month, then it shouldn't really matter which day of the month you're making your trades either.

Instead of trading once per month on any chosen day how about using a moving average envelope say like 2% above and below, in where you don't sell until you reach the lower band and buy back at the upper band. That way you have consistent mathematical buy/sell triggers instead of time based. Time based approach seems to throw in another random variable. I wonder if used consistent will yield similar results as the other methods? You can really get elaborate with this stuff, there are many other advanced methods as well.

Of course if one gains enough confidence then the next step would be to extend this into a growth fund or a growing sector where you can use the methods to reduce volatility on a downward trend but get a good portion of the upside.

[whodidntante]

For those who practice trend following, what are your related tools and information sources? I can certainly throw a 200 day SMA on an equity index ETF stock chart, but that is pretty cumbersome to do frequently and for several different asset classes.

Portfolio Visualizer is one of the most user friendly and free tools I've seen yet. It accommodates basic trend following, dual momentum, relative strength (what I use), and others.

What I had in mind was a kind of trend follower's dashboard where you can see factors that would influence buy, sell, hold decisions. Do you use something like that? Maybe it's on portfolio visualizer, but I didn't see it.

[willthrill81]

For those who practice trend following, what are your related tools and information sources? I can certainly throw a 200 day SMA on an equity index ETF stock chart, but that is pretty cumbersome to do frequently and for several different asset classes.

Portfolio Visualizer is one of the most user friendly and free tools I've seen yet. It accommodates basic trend following, dual momentum, relative strength (what I use), and others.

What I had in mind was a kind of trend follower's dashboard where you can see factors that would influence buy, sell, hold decisions. Do you use something like that? Maybe it's on portfolio visualizer, but I didn't see it.

I'm not aware of any such tool. Price over time is the only type of data I use.

[Lauretta]

For those who practice trend following, what are your related tools and information sources? I can certainly throw a 200 day SMA on an equity index ETF stock chart, but that is pretty cumbersome to do frequently and for several different asset classes.

Portfolio Visualizer is one of the most user friendly and free tools I've seen yet. It accommodates basic trend following, dual momentum, relative strength (what I use), and others.

What I had in mind was a kind of trend follower's dashboard where you can see factors that would influence buy, sell, hold decisions. Do you use something like that? Maybe it's on portfolio visualizer, but I didn't see it.

Well if you do Dual momentum you just need something like Perfchart by StockCharts.com (that's what Antonacci suggests you use) and compare the value of the 3 ETFs used in GEM now to what it was 12 months ago.
For SMA I think all platforms allow you to plot them.
So you don't need any dashboard.
Some quants consider other things like earnings momentum; they do rely on more complicated signals.