Alpha Architect on Momentum & Trend Following Rules

A couple months ago I wrote a post on Gary Antonacci’s book, Dual Momentum, and received a ton of questions and reader feedback on his absolute and relative momentum trend-following rules. I’ve been meaning to write a follow-up.

After reading the new book, DIY Advisor,  from the Alpha Architect team of Wes Gray, David Foulke and Jack Vogel, I thought it would make more sense to get their thoughts on the topic since they cover momentum so extensively in the book. Here’s a deep dive from a recent email Q&A:

​1. The ROBUST strategy you outline in the book utilize​s 2​ ​price-based market timing rules. Can you explain how ​these 2 rules work in a way that’s easy to understand for individual investors?

​Sure, but before getting into the nitty gritty of our downside protection system, we want to convey the strategy’s mission. Unfortunately, we can’t set a goal of no drawdowns, no volatility, tons of upside, and a Sharpe ratio greater than 4, even though that’s what people want. ​That goal is impossible because markets are too competitive. So given that constraint, an investor has to focus on a few things that matter most and optimize around those parameters. With ROBUST, we’ve chosen to minimize extreme drawdowns, while maintaining an opportunity to participate in the upside of various asset classes. ROBUST is not going to minimize volatility, or maximize Sharpe, and there is a good chance it will not have returns as high as a buy-and-hold strategy. But that’s okay. We’re trying to capture most of the upside, but more importantly, we are trying to preserve the kitty from a total implosion!

Now, with all that said, and with our mission carefully defined, what have we concluded from our 5-year research and development adventure into risk-management? We have concluded that 2 simple rules seem to work reasonably well: Invest in assets with strong-long-term trends and invest in assets with strong recent performance. With respect to strong-long-term trend, we simply mean buying assets that have current prices that are above a long-term average (say 12 months). When we talk about strong recent performance, or in Gary Antonacci’s words “absolute momentum,” we are talking about buying assets whose performance over the recent past (say 12 months) is positive. These 2 rules are highly correlated and in many respects the same thing, but we like looking at both for robustness.

2​. Why should we expect these rules to continue to work in the future?

​We wrote an extended piece on our downside protection model and highlighted some psychology research​ that suggests that human beings get more risk averse when they are stressed over a long duration. Why is this interesting? Well, when things get hairy and volatility spikes–over an extended interval–people get afraid. And when people get afraid, the price of holding risk shoots skyward, which leads to large drawdowns. Empirically, we see price charts that drop fast and furiously, we don’t see price charts that slowly erode along to an orderly 50%+ drawdown. Trend-type rules avoid these situations. Who knows if human behavior will change in the future, but the wiring in our brains seems to suggest that good old fashioned fear and greed will rule the day for many years to come.

3. What are some of the psychological barriers that prevent people from following such a model?

Following a trend-type model can be excruciating. To start with, they require one to make bold moves and either invest or not invest. This means we can sometimes be wrong, and additionally, we are naturally inclined to slow, incremental change. But there are studies that suggest folks can sell effectively using trend rules. So perhaps this aspect isn’t too difficult. However, the biggest challenge is the buy signal, which is usually at the trough of a massive face-ripping drawdown where everyone is contemplating a dive off a 100-foot cliff. The same study above suggests that humans have a much harder time following rules that force them to buy after a drawdown.

So on net, following these sorts of systems is difficult and there are a multitude of psychological barriers; however, all is not lost. One can either outsource this function to a fairly priced advisor, or one can maintain an iron will and stay disciplined to an established process.

4. What are the biggest potential benefits and drawbacks to utilizing trend-following rules from an asset allocation perspective?

By far the biggest drawback is tax-managing the risk-management events. Say we own an asset with a basis of $1 and it grows to $10. And now one of the trend triggers breaks and we need to sell the asset. On the one hand we don’t want to keep betting on risk, but on the other hand we don’t want to eat a $9 tax liability! This is a serious issue for taxable investors, which the vast majority of our own clients happen to be. We’ve come up with a set of principles and techniques we use in our own business and solutions one could use as a DIY Financial Advisor.

1) Annual rebalance on the core weights–For example, if your equity has a good run and your bonds stink, you sell some equity and buy some more bonds. If the allocations are close, don’t worry about it (ie 35% is target, but they sit at 40%), but from a risk management perspective you want to avoid “allocation creep”….so you don’t want 95% equity, 5% other stuff if equities have an epic ride relative to other stuff.

2) Harvest losses in the short run. Let’s say you have a really bad run on one of the assets, sell it, book the loss, and pop in an equivalent for 30 days. We usually use a 10% loss rule, so the loss we’re booking is big enough to overcome the transaction/complexity costs.

3) Tax managing the risk management signals–Let’s say you own REIT at $1 and it’s now at $10…but the rule says you need to hedge to 0% exposure. Instead of selling the REIT, you can short something that is 99% correlated to bring the effective exposure down to ~0%. We use futures to do this, but you can also use ETFs. This can get complicated, but the tax benefits of continued tax-deferral can be worth it.

4) Managing across tax-status–If you have different tax buckets (IRA, 401k, personal, trust, etc) you want to run the system across all your accounts, but dump the tax-efficient exposures in your non-qualified accounts and dump tax-inefficient in your qualified accounts. ETFs make equities tax-efficient, but the income distribution on the REITs and bonds makes them–at the margin–less tax-efficient. So you’d want to fill your qualified accounts with the REITs/Bonds as much as possible to shield their income distributions. Of course, there are other considerations with qualified accounts, but the general concept if you want to leverage your differentiated tax-status across your accounts to maximize overall portfolio tax-efficiency.

5. What percentage of investors, both on the individual and institutional side, currently utilize trend-following rules in their portfolio?

I would venture to guess that a majority of investors utilize some form of risk-management rule and/or trend-following rule. However, I would guess that VERY FEW actually follow them and often override their own rules. So the problem in the marketplace is not a lack of good ideas that work over the long-haul, the problem is a lack of discipline and overconfidence in one’s instinct to “beat a simple model.” We outline why this is the case in the first 4 chapters of the book.

You can learn more in the new book, which is out now:
DIY Advisor: A Simple Solution to Build and Protect Your Wealth

Further Reading:
The Opportunity in Active ETFs

Subscribe to receive email updates and my quarterly newsletter by clicking here.

Follow me on Twitter: @awealthofcs

My new book, A Wealth of Common Sense: Why Simplicity Trumps Complexity in Any Investment Plan, is out now.

Download PDF

Full Disclosure: Nothing on this site should ever be considered to be advice, research or an invitation to buy or sell any securities, please see my Terms & Conditions page for a full disclaimer.

What's been said:

Discussions found on the web
  1. Gregory commented on Sep 15

    Momentum is said to be the 4th factor in the Fama-French Model but I
    can’t find any DFA momentum funds. By the way as I know there is no monentum Vanguard fund either.

  2. drc commented on Sep 15

    hey ben
    how would you think about using this in an institutional context for an endowment or pension plan? thx.

    • Ben commented on Sep 15

      that’s a great question. my sense is most of the bigger plans would have a difficult time implementing this for either career risk reasons or market impact/trading costs. They could maybe utilize futures or options to hedge, but I think most would have a difficult time going all-in or all-out.

      I acutally think this is something of an advantage that smaller insti’l funds or even individuals have over the behemoths.

  3. Brian Hatch commented on Sep 16

    Ben, so with your own money (or that of a good friend asking for advice if you prefer) would you be more inclined to go with more of a buy and hold model of asset allocated ETFs like that offered by Vanguard or Betterment or Wealthfront, or would you tend to do these momentum and value tilted splits like recommended in this book?

    Having read his last couple books, I’ve got to say I’m intrigued by the backtesting numbers they think might be there for future investors, but have a hard time with the tax inefficiency and higher costs.

    • Ben commented on Sep 17

      I have no problem with either approach. It really depends on how much tracking error from the broad indexes you’re willing to take on. Plus, it doesn’t have to be an either/or situation. You could also use a mix of index etfs and value/momentum etfs. The beauty of that is you get a more diversified approach.

      But it all depends on how complicated you want to make things and how comfortable you are with those types of strategies.

  4. Jim Haygood commented on Sep 17

    For what it’s worth, my testing has shown that mixing two momentum criteria (even very similar ones) tends to muddy the signal, such that the mix usually performs a little worse than either criterion on its own.

    Both rate of change (e.g. today’s value divided by 12 months ago value) and moving averages have their indispensable uses.

    But combining them is a bit like wearing your camel hair sport jacket with your black-striped tuxedo pants — an awkward compromise.

    • Ben commented on Sep 17

      Do you have a preference between rate of change or MAs? Seems to me like the results would be fairly similar.

      • Jim Haygood commented on Sep 17

        No preference, except for what works. For instance, MAs seem to work best for timing the S&P, while ROCs work better for timing sectors and bonds.

        Maybe it’s just an illusion, or an artifact of the data. But splitting the difference by using both has never shown superior results, in my experience.

        Simplicity is always good, both for robustness, and as an aesthetic goal. Truth and beauty!