This post is the first of a series where we will be providing some of our own thoughts and commentary on the conversations we had in the first season of our new podcast.

This post covers our conversation with Adam Butler, which you can listen to here.


1:57 – Corey introduces Adam via a blog post Adam wrote about his experience with the emerging market and commodity super cycle theory of the early 2000s.

Corey Hoffstein (“CH”): One of the great things about interviewing bloggers is that their writing captures the real-time evolution of their thinking.  But few bloggers were writing more than a decade ago, so there is a lot that is lost to their personal history.  The post I quote was one written in reflection by Adam and I think really helps highlight that past experiences can influence current thinking.

Nathan Faber (“NF”): Hearing the, “I read Dennis Gartman” quote from the blog post and knowing some of Adam’s thinking now seemed like an about face in approach to investing.

3:43 – Adam discusses when he first got interested in investing and managed a bit of money for friends and family.

NF: It seems like we’ve all been there at some point.

5:03 – Adam enters a trading competition and parlays it into a trading job.

CH: I thought this was a great example of how ego can cause us to conflate luck and skill.  I firmly believe that one of the worst things that can happen for anyone early in their career is easy success – particularly if the source is luck and not skill.  Failure, on the other hand, can help keep us humble and open our eyes to the large role that randomness plays in our lives.

NF: “The right side of the market with leverage” is a great way to mistake luck for skill. The rise and fall of the inverse volatility ETF XIV is a great example of this on a large scale.

6:09 – Adam grows his trading account up to $8mm before blowing up and getting dismissed from the desk.

CH: I think the part that astounded me most, here, was that the desk did not seem to apply any risk limits to Adam’s trading book.  Instead, they seemed to keep increasing the size with which he could trade, doubling down on his success.  This can be a great way to compound skill, but it’s a phenomenal way to go bust if it’s actually all due to luck.

Justin Sibears (“JS”): This is a great example of procyclical risk management where risk limits are increased (decreased) as the book rises (falls) in value.  This creates a situation where we continually increase our bets until we blow up.  Part of the reason effective risk management is so hard is that it can require a measure of countercyclical thinking, lowering risk after runs of success and increasing risk after runs of failure.

NF: Thinking the darts is skill is dangerous. Sometimes superiors can’t even tell skill from luck.

8:54 – Adam turns his career towards technology with IBM.

CH: For as long as I have known Adam, I’ve been incredibly impressed with his ability to communicate highly complex topics.  I think that skill is paramount for quantitative managers, as we are the bridge between our models and our investors.

NF: Business and technical are often conflict. Being able to explain complex topics in simple ways is a very valuable skill.

JS: The asset management industry is such a great example of how business/sales and product can conflict.  Imagine you developed a strategy that over the long-run was negatively correlated with equities, but was expected to return 0% on average.  From a portfolio construction perspective, this strategy would have real value to investors.  Good luck raising money and keeping folks invested though.

11:01 – Adam discusses his dot-com boom-and-bust experience and his eventual transition to living in Bangkok.

CH: I’ve known Adam for years and had absolutely no idea he worked at a startup or that he had lived in Bangkok.

NF: A fresh perspective can often lead to a new direction.

12:59 – Corey tries to connect some dots between Adam’s time in Thailand and his eventual interest in the emerging market and commodity super-cycle narrative.

CH: I once read that history should be taught backwards, since when it is taught forwards it makes later events seem like foregone conclusions and eventualities, rather than just one of many possible outcomes.  In this instance, knowing where the story was going, I tried to connect the dots, only to find that what seemed like a natural connection did not actually exist.

NF: In hindsight, the trajectory is often easy to see, but we don’t often know what the current state is necessarily leading toward.

13:28 – Adam outlines the basics of the emerging market and commodity super-cycle theory.

CH: What I loved most about this moment of the podcast was that Adam was still able to recall, with extreme clarity, the entire argument behind the theory.  And he was able to tell it with such conviction that it almost had me believing by the end.

NF: Theory and practice often do not line up, especially when dealing with randomness.

JS: “The best investment opportunities come from areas where those who know it best love it least because they’ve been disappointed most.” – Don Cox.

16:27 – Adam and Corey talk the “illusion of knowledge.”

CH: I loved this phrase.  Informational edges are hard to exploit at a macro level.  There is an arrogance required to presume that, assuming the story is true, the market has not already priced it in.  I think there are also a sunk cost and confirmation bias effects at play here.  When you’ve spent a significant amount of time learning something inside and out, it can be hard to walk away from.

NF: Thinking that we know enough can put blinders on us for seeing our gaps.

JS: It invites the question as to whether successful investing requires a degree of discomfort.

18:11 – Adam talks about the eventual bust of commodities and how “nobody goes to God on prom night.”

CH: When you’re winning, it is nearly impossible to look in the mirror and admit that the source of success might be luck and not skill.  So true change often requires failure, which can be very, very painful depending upon your prior level of self-confidence.

NF: Having mistakes early on is a benefit as long as we can learn from them.

JS: I think this is true as long as we don’t over specify our learning to the specific mistake.  The next crash may very well look nothing like the bursting of the tech bubble or the global financial crisis.  If our learning would only be helpful in avoiding crashes that look and smell a lot like past crashes, then I think the value of the lessons are limited.  If instead they inform our overall view of risk generally and managing for the unknown, then I think they are much more valuable.

20:29 – Adam stumbles upon the work of Philip Tetlock while at a pool in South Beach, Miami.

CH: I have found that the work of Tetlock is a common thread of influence for many in the new generation of quantitative investors.

NF: Experts are no better than randomness, are often more overconfident than non-experts, and are worse at forecasting in their own field than people not in the field… this confirms that most of the news cycle is very detrimental to decision-making.

JS: The impact of randomness on investing results is, at least in my experience, massively under appreciated.  Also, enjoyed Adam’s thoughts on forecasting.  Superforecasting is one of my favorite books on the topic.

25:24 – Adam adopts quantitative investing and comes across Meb Faber’s A Quantitative Approach to Tactical Asset Allocation.

CH: It was fun, for me, to hear Adam talk about the influence that Meb Faber had on his work knowing that I would eventually be talking to Meb later in the series.

NF: A simple model for measuring trends performs much better than “advanced” forecasting. This process is 100% systematic, which also can reduce the stress of operating a strategy like this.

26:34 – Adam begins running some simple, relative momentum strategies for clients that, in retrospect, was probably overfit and how August 2011 caused him to rethink his approach.

CH: The hits just keep on coming for Adam!  I think the important difference, here, was that the hit was less one of poor process (though he admits that the model was overfit) and more one of investor behavior.  We often say, “the optimal investment strategy is first and foremost the one an investor can stick with.”

NF: An overfit model can be yet another way to mistake luck for skill.

JS: Preaching to the choir on Excel.

29:01 – Corey asks Adam to discuss what he means by “first principles.”  Adam proceeds to discuss the different assumptions associated with different portfolio techniques and introduces his Adaptive Asset Allocation framework.

CH: Adam uses the phrase “first principles” a lot.  What I really appreciate about this line of thinking is that Adam is unforgiving in his connection between implementation and the implied assumptions.  “By doing X, then you must believe Y” is a powerful way of thinking that can cause us to re-evaluate our process.

NF: It is important to understand what assumptions go into your portfolio construction. There are always uncertainties in estimates, and not letting these uncertainties have too much influence is key.

JS: It’s interesting to contemplate how Adam’s comment about roughly equal Sharpe Ratios across asset classes changes based on the current market environment.  In a world where 1-year Treasuries are yielding 2.34% (as of 7/9) and 10-year Treasuries are yielding 2.86%, is it still appropriate to assume that bonds will deliver similar risk-adjusted returns as equities?

32:33 – Adam opines on the role and limits of optimization in portfolio construction.

CH: I dialed the geek factor up here.  Optimization often gets a bad reputation due to the fact that it can be a rather fragile exercise.  Traditional mean-variance optimization assumes the parameters are known with certainty, but in reality everything we input is itself a random variable, shrouded in its own distribution.

NF: Optimization risk is a risk that must be accounted for, and there are ways to reduce it in portfolio construction.

34:56 – Corey and Adam discuss the example of optimizing portfolios of sectors and the theoretical implications of the empirical success of a naive, equal-weight approach.

CH: I particularly appreciated how Adam tied the failure of optimization to add value in portfolio construction back to a lack of useful information available found in the data being utilized.  Again, it is the tie between the practical implementation back to the implied, theoretical interpretation.

NF: Equal-weight is beneficial when overlaying model signals since it diversifies your model risk.

JS: One key takeaway here is that it is very difficult to make broad conclusions about the relative merits of different portfolio construction approaches.  The best approach for a stock portfolio is not necessarily the best for a multi-asset class portfolio.  In addition, different lookbacks and parameterizations can lead to very different conclusions.

41:48 – Adam explains why he believes there is more opportunity for alpha in asset allocation than security selection.

CH: I was really excited to ask Adam this question, because I think he offers a really provocative argument.

NF: The risk factors that drive individual securities are often limited in number with limited true diversification opportunities. Using very different assets provides more risk factors to diversify over.  The fact that many asset allocators are constrained limits the number of investors who try to capitalize on these opportunities in asset allocation.

45:13 – Adam discusses the importance of the implementation in his theory of Adaptive Asset Allocation and the importance of appreciating uncertainty and the role of randomness.

CH: I thought Adam drew out a powerful point here worth re-emphasizing.  In academia, there is a snowball effect where later research builds upon early research.  This means that certain parameter choices made by early researchers can find there way to becoming the de facto standard decades later.  Yet, while Jegadeesh and Titman may have found that a particular measure of momentum worked best, the broader conclusion for practitioners should be that momentum works.  I think this is a really powerful distinction that practitioners  should be aware of when reading research.

NF: Academic papers often settle upon a certain parameter even if they evaluate its sensitivity. However, subsequent research often uses only this parameter when the underlying sensitivity should be reevaluated in the context of the new feature being investigated.

JS: Great reading from AQR on this topic.

49:14 – Adam explains why embracing diversification is the ultimate gift.

CH: If diversification is the only free lunch in the market, then Adam makes a very convincing argument that most investors are foregoing the benefits of process diversification.  And you can hear Adam’s passion for the topic in his voice.

As a practitioner, Adam’s use of randomness was particularly interesting.  When he begins to explain the many variations of how a momentum strategy can be constructed, my head begins to spin with the massively dimensionality of the problem.  Trying to diversify across all these dimensions can become computationally intractable very quickly.  Adam discusses sampling randomly; in the past, we’ve had success employing Sobol sequences in these scenarios.

NF: Diversifying your model is diversification that is often left on the table. We are trained to think in terms of asset class diversification and need to retrain ourselves to think of all the axes of diversification.

JS: The number of different processes that can be used, even for a relatively simple investment approach, are truly staggering.  Which measure(s) should we use?  What lookback(s) should we use?  If we use multiple measures/lookbacks, how do we combine the measures?  How do we translate from signals to weights?  Etc.

56:31 – Adam talks about why overly embracing a KISS – keep it simple, stupid – mentality can lead to fragility.

CH: Quite often KISS is presented as a robust decision-making tool.  After all, if you keep it simple it is hard to overfit.  Yet, as Adam points out, by being too simple, you actually run the risk of overfitting.  Indeed, while diversifying across hundreds of implementations may seem more complex, in reality it makes fewer embedded assumptions than a single implementation does.

NF: A model should be simple to understand, but a simple model can still have parameter uncertainty that leads to unintended risks.

JS: Totally agree with Adam on this point. However, when using some of these “more complex” approaches, it’s important in my opinion to maintain transparency.

58:14 – Adam posits on how he would think about building a portfolio of individual securities instead of asset classes.

CH: This was a fun question for me to ask, only because I knew Adam’s answer would be so highly colored by the way he thinks about investing at the asset class level.

NF: Taking a model of models approach using different factors and specifications for those factors is a great way to diversify model risk and express views that we do not always know what will be the best over the short run.

JS: Sounds like Adaptive Stock Allocation to me.

1:00:31 – If you were an investment strategy, what would you be and why?

CH: Adam’s choice of deep value / contrarian fits perfectly with the investment thesis he outlined in our discussion.  His true north is, ultimately, is a healthy skepticism.

JS: In a sense, momentum/trend-following is itself a very skeptical approach to investing.  You forgo trying to understand exactly why something is happening with security prices or what is going to happen in the future. Instead, you simply go with the flow.

NF: Pair investing in Adam as a strategy with Adam’s philosophy as an investor, and you’ve got yourself some good diversification!

Corey is co-founder and Chief Investment Officer of Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Corey is responsible for portfolio management, investment research, strategy development, and communication of the firm's views to clients. Prior to offering asset management services, Newfound licensed research from the quantitative investment models developed by Corey. At peak, this research helped steer the tactical allocation decisions for upwards of $10bn. Corey holds a Master of Science in Computational Finance from Carnegie Mellon University and a Bachelor of Science in Computer Science, cum laude, from Cornell University. You can connect with Corey on LinkedIn or Twitter.