In this episode of The Acquirer’s Podcast, Tobias Carlisle chats with Adam Butler, CIO of ReSolve Asset Management. In this podcast Adam provided some great insights into:
- How To Win A National Trading Competition
- Why Do Investors That Are Given Systematic Models That Work, Still Underperform
- How Investors Can Develop An “All Weather” Risk Parity Portfolio That Will Outperform
- What Is Mean Variance Optimal?
- How Philip Tetlock Changed My Life
- What Is Samuelson’s Dictum And How Does It Apply To Markets
- Is Trend The Most Important Factor In Investing?
- How Investors Can Bundle Multiple Trend Strategies To Smooth Out Returns And Reduce Volatility
- Why Current Typical Retirement Models Are Flawed
Tobias Carlisle: Hi, I’m Tobias Carlisle. This is The Acquirers Podcast. My special guest today is Adam Butler of ReSolve Asset Management. He’s the author of Adaptive Asset Allocation and he’s one of the deepest thinkers on investing. We’re going to talk to him right after this.
Speaker 2: Tobias Carlisle is the founder and principal of Acquirers Funds. For regulatory reasons, he will not discuss any of the Acquirers Funds on this podcast. All opinions expressed by podcast participants are solely their own and do not reflect the opinions of Acquirers Funds or affiliates. For more information, visit acquirersfunds.com.
Adam Butler: Yeah. So, like you say, I started out thinking that I was going to solve this market problem by outwitting the market, by understanding the prevailing macro forces and just by having a better grasp of the underlying dynamics and the major players and their incentives and sort of wargaming it, that I was going to be able to be in the right sectors, the right geographies, the right asset classes, the right time without any sense of risk management or diversification or most of the principles that I now hold most dear. And it took a frying pan of the face after the tech bubble and another one in 2008 to really cement the idea that I wasn’t going to be able to do this on my own because I’m smarter, faster, braver, whatever.
Adam Butler: And I almost left the business after 2008 because when you get into investing because you believe that the value that you have to add is in generating alpha, and then the whole framework that you built to create alpha was proven to be a dangerous façade, then you’re left going, “Well, I probably don’t really have any right to this.” And so I had a bit of an existential crisis and really explored other potential directions.
Adam Butler: I went into university, I wanted to be a psychiatrist and eventually I became disillusioned with the medical approach to brain science and wanted to go into psychology, and I didn’t discover markets really until my third year. I’d never been exposed to markets. I grew up in a medical family, so that was just never part of my conversation or my world growing up, and I kind of stumbled into it in third year and so I went back to that. Do I want to go back to medical school? Do I want to go into law school?
Tobias Carlisle: What didn’t you like about the medical approach to science?
Adam Butler: Well, it was specifically the medical approach to-
Tobias Carlisle: Brain science.
Adam Butler: … brain health, right? And I’m not necessarily talking about neuropharmacology so much as … well, it was the emphasis on neuropharmacology rather than cognitive behavioral therapy and other more positive treatment plans or approaches to emotional and brain health that I feel and felt at the time that people should focus on and give a much broader opportunity to work before resorting to pharmacological and surgical solutions. So, I felt at the time, and I think that this dynamic has probably moderated over the last decade or so, but I felt at the time that the go-to in psychiatry was almost immediately to pharmacological intervention and there were just a wide variety of steps that we should take before we go there.
Adam Butler: And I think the psychiatric profession has come a long way in embracing some of the empirical findings out of the psychology literature and I think there’s a meeting of minds there.
Tobias Carlisle: How did you get exposed to the markets? When did that actually happen?
Adam Butler: I actually can’t point to a catalyst. I can point to exactly what it was. I remember having a conversation with a family friend who is a broker at a big Canadian wealth management shop over Christmas, and this was probably Christmas in my second year of university. And I came back and was just really curious about how this whole thing worked, and I started reading the business pages, and looking at the stock tables.
Adam Butler: And I don’t know how many other people follow this learning trajectory. I think is the hardest learning trajectory, but I started watching the moves and I started noticing that there were these small cap stocks that they would double and in a week or two weeks, and I started zeroing in on the fact that these were typically kind of little mining stocks or little venture stocks. And so then I started doing a little poking around on those. And where did you start poking around on those in those days? Well the stock bulletin boards. And so I started trading based on some of the rumors and comments and stock bulletin boards. And I ended up making a little money only because I was only going along and it was a junior mining bull market, so you could throw darts and make money.
Adam Butler: And then because I made a little bit of money, I mean, like a few hundred bucks, I got really cocky, because I get cocky really quickly, and I went to my friend’s parents and I said, “You should try this. There’s lots of really cool stuff going on.” And so they ended up giving me a few thousand bucks, this is my roommates in my house at university. And then I ended up giving them back about half of what they had given me about six months later. So, that was my first. But I was far from learning any meaningful lessons just from that outcome.
Tobias Carlisle: So, at some stage, you enter into a trading competition run?
Adam Butler: I did, yeah. That was in my third year. There was a national trading competition that TD Bank had put on there. They had just launched their online discount brokerage. And so they had set up this competition where you are granted this illusory million dollars and you can trade stocks and options long, short, using exactly the same kind of commission structure and borrowing and margin structure that you would use on their platform and you had to see how far you could grow this in six months. And so I basically just traded far out of the money tech options, calls and puts for six months and ended up turning a million bucks into like 1.6 or something. I actually can’t remember. Actually now that I think about it, I think we started with half a million, it wasn’t even a million, it was a half a million and I think I turned it into 1.2 or something in the first competition. It was just you take on as much leverage as you can and hope for some luck, and I did have a lot of luck.
Tobias Carlisle: If you were given the same competition parameters now, would you do exactly the same thing?
Adam Butler: No. I think I’d probably go the other way and I would probably seek to short as much fall as possible. Because shortfall is enormously profitable most of the time, so if you just happened to get lucky and be in a period where you don’t get your head handed to you by being a shortfall, then it’s a really nice surefire way to deliver really high returns in a short period of time. I mean, the problem is obviously that there’s unlimited downside, so when you do get your head handed to you, then it’s a life altering event unless your position size is managed properly. But yeah, I mean, just off the top of my head, I think I might lean in that direction. How would you do it?
Tobias Carlisle: Well, I was thinking I’d probably do the same thing that you would do. I mean, you know that it’s going to be a gamble. It’s pure luck so you need to maximize the movement of your portfolio, and that means probably you’ve got to maximize your chance of going up or down and just flip the coin and hope that it’s up.
Adam Butler: Yeah.
Tobias Carlisle: And that’s what I’d be doing. I’d probably find something that’s whipping around and then find out of the money calls or something like that. Like you know that you’re gambling, but that’s your best chance of actually winning. Because there’s no point putting it into a whole lot of big cap value names or something like that. You can’t win if you do that. The only way you can win is if you risk everything that you’re putting on run.
Adam Butler: Exactly. I mean, the winner is going to come from a tail event, whether it’s an accumulation of small tails or one large positive tail event, that person is going to win. So, you need to put yourself in a position to-
Tobias Carlisle: To win.
Adam Butler: … yeah, to be in the tail somehow, right? So, whether or not that’s shorting straddles or something on really high volatility stocks or like we were saying, just longing calls and puts, some combination maybe.
Tobias Carlisle: So, did you win the competition?
Adam Butler: Yeah, I won the first one. I entered it again and I came third. So really, that was just an enormous tailwind for the start of my career. It was entirely due to luck and maybe a little bit of chutzpah and then parlayed that into a front page in the business section in my university paper, which I got photocopies of and sent around to trading desks as I was looking for work.
Tobias Carlisle: That’s right.
Adam Butler: And I got written up in my hometown paper as well and sent that around. I got a letter from the broker that I had been chatting with at TD because I was also creating my own account at the same time. And so he wrote this glowing review of how I approached the problem. And anyways, it ended up just being a really good resume building exercise in the end, but most of it, as we both know, it was just completely luck.
Tobias Carlisle: So then you start on the trading floor, and what happens then?
Adam Butler: Well, I think I’m God’s gift to trading. I got my CSI or my securities course halfway through my first year on the desk. I think deep down I felt this is all garbage. Who needs this training? It’s an easy game. So, the desk I was on was a market making desk and we were also responsible for creating trading ideas for the retail brokers. So, I spent a lot of my time kind of on Bloomberg, just looking for ideas or getting ideas from the senior members of my team and doing some research on them. But then also in order to stay in the flow of the market, I was allowed to trade this Omnibus account.
Adam Butler: And so we were able to hold positions in the Omnibus account. And at the time, you were allowed to hold positions in the Omnibus account and then you were allowed to offload them to brokers after a little while and sort of say, “Hey, listen, I own some stock A, a little below the market here. You want to be a hero you can take this down, call your clients and put it on the books at this lower price and you’re already giving your clients and instant profit.” So, that was allowed at the time, it’s obviously no longer allowed today. But also we were trading our own book at the Omnibus account. And there was almost no latitude to trade on the Snow Capital allocation in the beginning but it grew because I was trading well because it was a bull market in high volatility tech stocks.
Adam Butler: And so I basically was just translating my experience from trading mining stocks and trading high out of the money calls and puts for my training competition, just doing the same thing in this Omnibus account, and it was working out great. And eventually they gave me more and more risk budget. I managed to accumulate a very large profit in the Omnibus account coming in right into the teeth of the Thai Baht Russian default long term capital management crisis in the fall of 1998 and ended up losing it all and more and getting escorted from the trading floor. So yeah, that was sort of the third. It was the first, but it was the first of the three major emotionally salient crises that finally led to me learning some lessons and changing my approach.
Tobias Carlisle: You’re skipping forward a little bit too far in time because you take a seven year sabbatical.
Adam Butler: I took a lot of time off, and my wife and I went to Thailand for a few years, and I taught math and physics at this huge boys’ school in Thailand, which were absolutely some of the best years of my life. Absolutely love Southeast Asia. We actually took our kids over there, we have three children, we took our kids over there for a month, a couple of March’sago and can’t wait to go back. But I learned computer programming. I worked at IBM for a few years. I worked at an Internet startup, and yeah. So I mean, it turned out that my university degree was in psychology, and then just by virtue of making big mistakes, ended up doing a lot of work in programming. And then psychology really is a degree in applied statistics. So, you’ve got an applied statistics combined with programming ended up being a really perfect foundation for the sort of next phase of my career after 2008.
Tobias Carlisle: So, we have the 2008 crisis, and I’ve heard you say previously that it’s a huge event because it changes the way that you think about everything. So, what was the process there?
Adam Butler: Yeah. I mean, this is a lot of deep work just sort of thinking about my own motivations, and how I perceived, the world, and my general social context, and values context at the time. But I think that I was persuaded as I grew up to believe that the world was a meritocracy and that anybody with a work ethic and a little bit of smarts could make anything happen. And I guess I had never really been face to face with the realities of the hierarchy of wealth and the fact that there are existing social structures and wealth structures that really don’t want to be disrupted and will do anything in their power to preserve the sanctity of their status. And so I just wasn’t prepared for the kind of interventions and the way that those power structures were able to get away with perpetrating what I felt to be just horrible crimes against the integrity of the financial structure.
Adam Butler: And so it took me a long time to get past all that and just realize, “Look, you’re deciding to play in this sandbox. These are the rules of the sandbox. If you don’t like the rules, go find a new sandbox,” right? And just move on and find some methods to both leave that behind me and then figure out how to move forward in a constructive way to add value within my chosen domain.
Tobias Carlisle: So, I’ve heard you say that your approach to that point had been you thought that the machine could be understood and that if you worked hard enough and if you thought about it hard enough, you could at some stage figure out the right place to invest, the right asset class. But instead, that sort of changes your thinking on the whole process and you start thinking that’s an impossible task, so you need to be less certain, more agnostic to what the market could possibly do. And so part of that is you read Tetlock as many of us have, Dan Rasmussen had the same story, and I have it on my bookshelf behind there. So, talk us through that process.
Adam Butler: Well, it was at exactly the point when I was doubting whether or not I could add value within the investment domain, whether or not I could be a successful investor on behalf of clients that I stumbled on, initially, a presentation that Philip Tetlock had given to the Long Now Foundation, of which I was a member. And it was on his findings from the 15 year experiment that he’d been running and that led to the publication of his book, Expert Political Judgment, wherein he sort of described the methodology of the experiment that he had conducted over many years, the motivation, so what led him to want to do the study that he did, and then what he had concluded. And I found that to be just an absolute revelation. And we can certainly get into that.
Tobias Carlisle: Yes, let’s do that. He was in the CIA, is that right?
Adam Butler: Yeah. Well, he was certainly part of the intelligence community. He graduated with a degree, and I’m going to get a few of these facts wrong, but I think he graduated with a degree in applied psychology or operational psych or something like that, and he went to work in the intelligence community in Washington as a junior analyst. I guess he was responsible for documenting the remarks, taking minutes at these intelligence meetings, and they would come about on a quarterly basis or semi annual basis, and all the big wigs in the room, the top generals and the senior intelligence advisors would provide their opinion on what was going happen in the Russian Politburo or whatever other dominant political dynamic that was prevailing at the time.
Adam Butler: And so he’d write down their opinion and what this senior official thought would happen. And then at the next meeting, he would have to read out the minutes about we’re starting from here and here’s where everyone thought that the world was going to be at this time. And then what he noticed very quickly is that there was almost no relationship between what all these senior officials had predicted would happen and what had actually happened in the intervening period. And so that motivated him to go back and structure this long-term study on expert judgment. And so he recruited something on the order of 200 experts. And again, it’s been a while since I’ve looked through this, but I want to say the average education level, master’s degree, average level of experience, kind of 15 years in their fields. So, these are senior intelligence advisors.
Tobias Carlisle: Experts.
Adam Butler: Yeah, experts. Economists, journalists, etc, right? And then he asked them each, I think about 100 questions. We had about 18,000 forecasts that he had accumulated over about a 15 year period on a variety of different topics. And so when he went back and evaluated the results, he came to a few conclusions. Number one, experts were less well calibrated. So, what does calibrated mean? It means that if they say that they are 60% confident that something will happen, that on average when they say they’re 60% competent, those things happen on average about 60% of the time, right? So, that experts are less well calibrated than you would get from random guessing, that there were no outliers among the 200 odd experts, not one of them exhibited a calibration that was better than random guesses, in fact, most of them were substantially worse.
Adam Butler: Those that expressed the highest level of average confidence ended up having the least accuracy in their predictions. Some other stuff which I found even more interesting, the experts that were cited most often in future literature, or by the news media, or appeared most often on radio or television, were meaningfully less accurate and less calibrated than those who toil in obscurity. And then alongside these experts, he also ran these very simple algorithms. Like the prevailing trend will continue, or the dynamic will revert to their long term mean. And so of course, in the short term, what he found is that the trend continuing algorithm did very well, and then for longer term predictions, reversion to the mean did very, very well. And both of those systematic methods vastly outperformed the predictions of the experts.
Adam Butler: So, that coincided with some of the other systematic literature that I’ve been reading and so there was a positive feedback effect there I think.
Tobias Carlisle: That’s very interesting because there’s an area of study that finds the same thing, and Paul Mead I think is the sort of founding father most representative. And his is that, uh, a simple statistical models outperform experts, including when experts get the benefit of the output of the simple statistical models, the model acts as the ceiling from which you detract …which you build on, which is fascinating and incredibly salient for investors, particularly quantitative investors.
Adam Butler: Absolutely. And I just love the story of … Oh, what was his name? Greenblatt, right? So, Greenblatt writes this book. What’s it called?
Tobias Carlisle: The Little Book That Beats The Market.
Adam Butler: The Little Book That Beats The Market, right? And he’ll know what the exact specifications are for his model, but it’s got something to do with ROE and some value metric.
Tobias Carlisle: Right. EV, earnings yield.
Adam Butler: Yeah, there you go. Yeah, yeah, yeah. So, he’s been very successfully using this strategy to manage his hedge fund, which has done spectacularly well for over a decade. We’re talking what? 25, 30% annualized returns. Just something just absurd. And so then he shares this formula with the world and then he creates this service where he allows people to invest in this model so they give them money to this service, and there’s two ways that you can take advantage of this service. Either you can just let the service run the model for you systematically. So obviously, you’re taking all buys and sells that are spit out by the model, or the service will give you the stocks that you need to buy and sell and then you can go and buy and sell them on your own.
Adam Butler: And then a few years later, he examined the performance of the accounts that had just allowed the service to run the trades versus the accounts of the investors who had been given the trades and had to execute on their own. And the accounts that had systematically executed the trades had just vastly outperformed the accounts that had discretionarily executed the trades, presumably because the people did not take all of the recommended trades, right? They looked at some of the names on the list and were like, “Wow, these are getting a lot of garbage coverage in the press or, “Hey, they say I should sell this stock but look at all the great things it’s doing, look at it’s earnings growth, whatever,” right? So, they’re overlaying their own bias.
Adam Butler: So, this is a perfect example of that the system sets the upper limit and all you can do with your discretion for the most part is detract from the results.
Tobias Carlisle: He wrote it up as an article in Morningstar called Adding Your Two Cents Will Cost You A Lot. And what he found was the model over the two years that he tracked it returned 84%, the market did 65%, so it outperformed the market materially. And then the average of the accounts that cherry picked, the discretionary accounts was like 54%. So, they actually underperformed the market and the model.
Adam Butler: Love it. Yeah.
Tobias Carlisle: The best discretionary account just bought all of the stocks in the model at the start and then didn’t trade for two years and it outperformed.
Adam Butler: Oh, that’s great. Yeah. I mean there’s just so many examples across so many disciplines, whether you’re looking at recidivism or selecting medical students, or-
Tobias Carlisle: Wine vintages.
Adam Butler: … MBA medical students that’ll complete programs. It’s overwhelming.
Tobias Carlisle: Bordeaux wine vintages.
Adam Butler: That’s right. Yeah, wine vintages.
Tobias Carlisle: Countless different things. So, let’s move into what you’re doing now. Just describe the firm for me, if you would, and the strategies that you guys offer.
Adam Butler: So, the firm is ReSolve Asset Management. We were founded and September, 2015. We launched on the back largely of a strategy called adaptive asset allocation. And the thinking that went into that was we realized soon after the financial crisis that the … everybody says during financial crises, the correlations go to one, and we discovered that that’s just unequivocally not true.
Tobias Carlisle: Right. I would have said that.
Adam Butler: Right. Yeah. Yeah. And I mean, you probably know this because I know you, but we ask this question in conferences all the time. We sort of say, “What was the best performing asset class in 2008?” And we got all kinds of guesses. Cash, short funds-
Tobias Carlisle: Gold.
Adam Butler: … gold. Yeah, right.
Tobias Carlisle: Volatility.
Adam Butler: And gold was up a little bit. But yeah, treasuries.
Tobias Carlisle: Treasuries?
Adam Butler: I mean, long treasuries. Yeah, long treasuries were up 30% in 2008, 30%. And this is a TLP, all right? And what’s interesting is of course for Australian, or Canadian, or many other foreign investors, you don’t just get the benefit of the move in treasuries, you also get the benefit of the move in the dollar. So, for example, for Canadians, a Canadian who held the TLT/ETF long term treasuries in 2008, would have earned over 60% on that position. So, the idea is that there are almost always places to hide and ride out a market crisis, right? It’s not always treasuries as we know, right? The 1970s bonds were highly correlated with stocks and bonds and stocks both did poorly over that decade. But at the same time, commodities and gold both put in a double digit compound returns.
Adam Butler: So, that sort of understanding crystallized over several years. It wasn’t sort of an overnight epiphany, but as we started to learn more about the precepts of risk parity and the all weather concept and what drives asset pricing, then this whole idea took shape. But the initial focus was on asset allocation because that was where we discovered that if you were properly diversified, you were least susceptible to the vagaries of just being in the market at the wrong time.
Tobias Carlisle: So, just to talk to us a little bit about risk parity. You’re looking at as many asset classes as you possibly can, and you’re allocating to those asset classes based on … you’re trying to equilibrate the risk for each, defined as volatility in that sense.
Adam Butler: Yeah. Just at the highest level of abstraction, the idea is that assets are predominantly sensitive to two dynamics, and those are changes in inflation expectations and changes in growth expectations. And so if you have an unexpected change in either expectations for growth or inflation, then that change will impact different asset classes in different ways for different reasons, right. And so you’ve got traditional kind of growth assets like equities and real estate, and sometimes commodities, but then equities do particularly well, especially sort of developed equities do particularly well during deflationary growth environments, whereas commodity oriented economies or stock markets, like emerging markets, tend to do well in inflationary growth environments. Stagflation where you’ve got high inflation and low growth, you typically get really great performance in commodities and gold and TIPs.
Adam Butler: And then you’ve got deflationary bursts like the Great Depression in 2008 where you want to rely on things like long term treasury bonds, cash and gold, right? So, you don’t know what environment you are going to be going into next or be facing next, and so you just want to have access to all of these different asset classes that are fundamentally designed to do well and in each of these different quadrants. So now that’s diversity, you want to have sufficient diversity, and then to your point, you also want to have balance. Because the problem is you’ve got emerging market bonds, and you’ve got intermediate treasuries, and so one of them is going up and down at one or 2% a day and the other is going up and down at a half percent a day, and so as you accumulate the returns over time for these different assets, the returns in emerging markets are going to completely swamp the returns to intermediate treasuries. So, the treasuries don’t have an opportunity to diversify the stocks.
Adam Butler: So, if you want all of these different asset classes to be able to diversify one another, then you’ve got to scale them to contribute the same amount of risk to the portfolio. And how you define risk and how you equalize risk is a whole very deep conversation. But at the highest level of abstraction, you just want diversity across all these different macro environments and you want balance across each of these different markets, and that gives you a risk parity.
Tobias Carlisle: And so the philosophy that you describe in your book, Adaptive Asset Allocation, that’s what you’re driving at and that’s what you’re trying to incorporate in the firm?
Adam Butler: Yeah, I mean, that’s sort of the place to start, which is you want to have access to a universe of assets where at least one of them will be positively exposed to whatever economic environment we are going to face in the future. So, that’s sort of step one, and we use risk parity to help to develop that universe. And then we describe the optimality conditions for risk parity, in other words, under what conditions about relationships between risk and return is the risk parity portfolio mean variance optimal?
Tobias Carlisle: What does mean variance optimal mean?
Adam Butler: Does a risk parity portfolio give you the maximum amount of returns for a target level of risk? And so there’s a set of conditions under which that is true. So, that’s a really good starting point. If you believe that those conditions are a reasonable … in markets, then that’s a good place to start. And then if you want to deviate from that, then you better have identified an edge that gives you confidence in systematically deviating from that risk parity portfolio with an expectation of increasing your risk adjusted performance.
Tobias Carlisle: Do you deviate? And what are the conditions under which you do that?
Adam Butler: Yeah. So, adaptive asset allocation takes the view that markets trend and that they exhibit momentum. And so the idea is to take this diverse universe of asset classes and emphasize those markets with the most positive trends and de-emphasize those markets with the least positive or negative trends. The way we describe it in the book is those markets that are in the top half of the momentum distribution, we just take the minimum variance portfolio of those, which is a sort of simple proxy for the overall process. So, in other words, you just sort of take all those markets that have positive momentum and you find that combination, the weights of those markets that minimize total portfolio risk. And that works reasonably well, and then there’s other ways that you can use these momentum signals and optimizations, and then there’s lots of ways that you can make the process a lot more, stable. But that’s the general approach.
Tobias Carlisle: What is Samuelson’s Dictum and what does it imply?
Adam Butler: Yeah. I mean, Samuelson’s Dictum it really is a foundational concept for what we do at ReSolve. And this comes from a letter that Paul Samuelson, famous economist wrote to Robert Schiller, so two Nobel Laureates, where he made the assertion that he felt that markets are macro, inefficient, and micro, efficient.
Tobias Carlisle: What does that mean?
Adam Butler: So, it means that at the individual security level that there are actors with sufficient capital relative to the size of the market caps of the securities that they are seeking to arbitrage, that they can drive the prices of those securities towards equilibrium. And that there is sufficient cooperation of agents that are sufficiently capitalized to create a regular state of quasi equilibrium. So, individual securities are probably efficiently priced most of the time.
Tobias Carlisle: But then why does that break down at a macro level? Why is it macro, inefficient?
Adam Butler: Well, he didn’t say so now I’m off-roading here, right? But we believe it comes down to two reasons that generally you’re broken into portfolio agility and just flexibility of mandate, right? Mandate flexibility. So, these large institutions that are generally in charge of deploying capital, let’s think about how they’re structured, like a CalPERS or any major pension plan or endowment. So typically, you have an investment committee at the very top that is responsible for a policy portfolio, a strategic asset allocation. And in many cases, that strategic asset allocation has to go through an actuary. The actuary has to say, “Yeah, based on the historical performance of markets, this portfolio is likely to meet our actuarial objectives,” right? And once they put a stamp on that, that portfolio is largely static. And the ability for the investor committee to deviate meaningfully to take active risk against that policy portfolio is extremely constrained.
Adam Butler: So, you’ve got this long term strategic set of weights that you’re going to give to equities, to fixed income, to real estate infrastructure, private investments, etc. So then that money goes to the equity team, and maybe this large equity team, maybe some of them are picking individual stocks, but largely they’re finding managers that are going to pick individual stocks. There’s a credit team that finds managers who are picking great credit managers or picking managers that they think can time duration or whatever, right? And they do the same thing with private equity, infrastructure, whatever. But there’s nobody at the macro level who’s even empowered to take active risk across the different asset classes.
Adam Butler: So, you have a situation for a very long period of time where even if the equity sleeve is becoming more and more and more overvalued, there is no facility for the institution to act against that. So, that’s one challenge, right? Just the way they’re structured, they just don’t have the mandate flexibility, and it’s not just a large institution. Think about the way that a typical retail client portfolio is structured, right? You set a strategic asset allocation that’s informed by either your emotional risk tolerance and your financial risk tolerance and your required return, and then largely that stays put, and if you want to deviate from that, you got to go through your compliance department to deviate meaningfully from that, right? So, you have similar kind of constraints.
Adam Butler: So, there are very few actors in markets that are well capitalized and that also have the mandate flexibility to arbitrage across markets, right? And then the other challenge is that to sufficiently arbitrage an entire asset class requires far more capital than any institution has on its own. So, there needs to be a globally coordinated effort to equilibrate the valuations of different asset classes in a way that there doesn’t need to be at the individual security level. And so in our opinion, those barriers to arbitrage are sufficient to preserve very large inefficiencies at the asset class level that may not exist to the same extent at the individual security level.
Tobias Carlisle: Right. So, how should we think about the problem of getting the most information from market edges?
Adam Butler: Well, I’ll tell you what, look, we started out, even once we sort of decided that we were going to move into the systematic space, once we had discovered the power of diversity and balance, we zeroed in on one major edge, and that was trend, right? I mean, we have a small loading on momentum, but it is primarily trend that informed our adaptive asset allocation strategies for many years. So, how did that happen? Well, because we examined the extensive data on all of the known edges in markets, and of all of the edges, the one with the largest, most persistent edge has been trend, and we’ve got very detailed data that goes back to the 1970s on actual futures markets trading, then you’ve got index extensions that go back to the early 1900s, and then you’ve got other market extensions that go back … you know, … and Kaminski and their book go back 800 years on trend.
Adam Butler: And so I mean, this is just an omnipresent phenomenon and now you’ve got literature coming out that trend also provides an edge even on factors in alternative premium, which we haven’t personally investigated, so I can’t vouch for it. But there was this overwhelming amount of evidence that trend is omnipresent and it just vastly outperforms all of the other edges in sample. I mean, literally, my plotting app won’t even plot trend against the factor is unless I adjust the returns to trend down to incorporate two and 20 fees because the returns are just so extraordinary. I mean, it’s like an 18% annualized return at a 10% ball going back to 1900, and 10% after two and 20 fees at the same ball. And it works across asset classes and we were looking for a factor that works really well across asset classes. So we said, “Well, this is clearly the winner. We’re going to just make as much use of this single factor as possible.”
Tobias Carlisle: Can I just ask. You just said before, are you distinguishing trend and momentum?
Adam Butler: Well, yeah, but I mean, there are ways that you can measure-
Tobias Carlisle: Are you saying trend is in a time series in momentum is in cross sectional? Is that how you’re thinking about it?
Adam Butler: Yeah. So classically, that’s how it’s measured. But let’s imagine this, we’ve got five different measures of trend. Let’s say we’ve got one month, three, six, nine and 12 month measures of trend. And for each time horizon or each trend horizon where a market is in a positive trend, we’re going to give it a score of one, for every horizon that’s in a negative trend, we give it a score of negative one. And at the end, we’re going to add up the scores, right? So, any market’s going to have a maximum score of five and a minimum score of negative five, and we’re going to take that score and we’re going to feed it into an optimizer, or we’re going to hold the markets in proportion to their score. Is it a trend strategy or is it a momentum strategy?
Adam Butler: We’re kind of a ranking, but we’re ranking on the stability of trend. So, I don’t know, it’s kind of both, right? So, I will grant you that the classical definition for trend momentum is cross-sectional distribution of returns for momentum and time series, positive or negative for trend, but there’s lots of ways that you can mix and match those and we mix and match a lot, as I think you know. So, it’s hard to say that we use either trend or momentum because we kind of just use a combination of both. But either way, we did really focus in on this one factor, this trend factor, and we started really running with trend in late 2011. And if you look at the returns on trend from 2012 to 2019, it has been the worst period for trend in the history of trend, right? It’s just as bad a period for trend as it’s been for you on the value side, if not maybe a little worse.
Adam Butler: So, the major takeaway there was that as diversified as you think you are, we’ve got all these different markets, and they’re designed to do well in very different macroeconomic environments. We are allocating or emphasizing markets based on the most widely documented, omnipresent, strongest edge, and still, things can go very wrong for a very long period of time.
Tobias Carlisle: Why do you think that is? Why for trend?
Adam Butler: I don’t really know. I personally have a theory that trend is by design a function of investors overreacting to macroeconomic dynamics, at least at the global macro level, overreacting to macro economic dynamics and hurting into assets with strong returns because they’re return chasing and they’re benchmark focused. But the current period has been dominated by macro economic dynamics that have been unfavorable to powers that have the power to intervene against those trends. So, the dominant macro economic dynamic over the last decade has been deflationary stagnation and the actions of central banks have been to counteract that natural trend, right? So, you have this system where you’ve got the markets move in a direction of deflationary stagnation, the central banks observe that this is taking place, they intervene with massive stimulus, the macroeconomic trend reverses for a time, the central banks back away, the dominant macro economic environment prevails again. And this cycle has gone on over and over again, so you’ve got this situation where the trend begins to develop just as the central bankers intervene.
Adam Butler: This is purely me and I’m not making excuses. We have the responsibility to recognize that this is a risk to trend strategies. And so whatever your explanation is or whatever you want to blame for the effect, the reality is people have given me their money to compound and I’m making decisions on their behalf and I should be aware of this potentiality. So really, what I’ve learned again is the power of humility, right? That you can’t just rely on this one no matter how powerful it is. In fact, if anything, I think what I’ve learned is that the more statistically significant and economically significant a factor has been historically, and especially in the recent period, the more likely it will be to be recognized by a very substantial portion of agents who will begin to deploy to that factor and therefore compress the expected returns. So yeah, it’s again the power of diversity.
Tobias Carlisle: Well, let’s just talk very quick. One of the things that I’ve learnt most from you is this discussion about simplicity and complexity. And so can you just tell everybody how you think about that and what the implications for that are.
Adam Butler: Yeah, sure. So, a really simple example is trend. So, a classic factor trend strategy might own a market if it’s above the 10 month moving average or above the 200 day moving average, or has exhibited positive returns over the past 12 months, or the past 252 trading days. But the reality is there’s no magic to that, right? Sure, those filters have worked over time, but other ways of defining trend have also worked, like markets that have broken out above their 252 day high, or their six month high, or a double moving average cross, a 5,200 day moving average cross, or a price to moving average cross, or a triple moving average cross, or MACD or whatever.
Adam Butler: I mean, there’s an enormous variety of different ways that you can define trend. And if you examine the performance of all of these different trend strategies, you find that sometimes the 5,200 day outperforms, someday that 12 month trend outperforms, or time series trend outperforms, all of these different trends specifications outperform at different times, and this is enormously powerful because it means that you can allocate to all of these different specifications of trend, which will underperform and outperform at different times, and really have the opportunity to smooth out your expected return trajectory.
Adam Butler: Now, I think it’s reasonable to assume that all of these different specifications are equally valid. If you’ve got 1,000 years to invest, then you’re probably going to do about the same using any of these different strategies. Just like if you’re an expert blackjack player and you’re going to sit at a blackjack table and you’re going to play 10,000 games, then you’re probably going to do just fine, but if you’re going to sit at a table where you’re only going to have a chance to play 15 or 20 or 50 games like a typical investor’s got a limited time horizon, right? Then you’re better to have a team of players that are going and playing at a variety of different tables all at the same time using basically the same strategy, and the results are going to even out over time, right?
Adam Butler: This is all related to ergodicity economics, and geometric versus arithmetic returns, and we want to converge on the arithmetic return, which means you want to minimize your variants. Like there’s all kinds of stuff going on here, but the underlying principle is simply one of diversification again and being humble about whether or not we think we’ve got the perfect mousetrap for trend, or the perfect mousetrap for value, or the perfect mousetrap for whatever edge you think that you have in markets.
Tobias Carlisle: So then how do you then work out how much you allocate to each of these different trend models? So, you say the one that has performed the best gets a little bit more or do you just say we don’t know which one is going to be performing the best over this period of time so we equal weight, or do you use that risk parity approach?
Adam Butler: Well, it depends on what strategy. I mean, we’re launching a strategy, it’s a global equity momentum index that Corey at Newfound-
Tobias Carlisle: Corey Hoffstein
Adam Butler: … and ReSolve are co-launching, yeah, and we’re just sort of equal weighting a variety of different trend specifications. We’ve got models that allow us to determine analytically over a long horizon what the correlation should be, for example, between a one month time series trend specification and a 12 month time series trend specification, or a 12 month time series trend and a 10 month 200 moving average cross trend specification should be. These have actually well-defined analytical correlation relationships that you can use to maximize diversification across the trend specifications, and we do use those internally. But I mean, even just equal weighting them is more advantageous than having to look across all of the different trend strategies or trend specifications historically and say which one of them has outperformed in sample.
Adam Butler: I mean, we just wrote a paper on this global equity momentum concept, and you probably know Gary Antonacci, who I’m a big fan, and he’s a great guy and a smart guy. And I wrote a paper in 2012 on this idea of dual momentum. And one of the strategies was where you want to be invested in equities subject to the fact that equities are in a positive trend, and then if equities are on a positive trend, then you want to be in either US equities or international equities depending on which one of those has the highest momentum. And he defined trend and momentum based on the 12 month return, right? Which is a perfectly valid specification. And if you look back to 1950, that strategy has performed very, very well, right?
Adam Butler: But what’s interesting is that the original study that he did in 2012 used monthly data from 1974 to 2011, and he used the 12 month approach, and he showed the 12 month approach was good, and then he expanded it, he got new data and he expanded it back to 1950, and then from 2011 to, I don’t know, 2018 or ’17 or something. Right. Which is great because now we have an out of sample “sample”, right? And so what’s interesting is you’ve got this specification that worked the best from 1974 to 2011, and then if you look at the performance of that specification, that 12 month specification on the period from 1950 to 1973 and from 2012 to 2018, what you find is that that strategy performs at about the median of all of these other different trend specifications, which is exactly what you’d expect if your prior belief is that all of these different trend specifications are equally valid, right? If they’re all equally valid, then over the long run, they’re all going to converge to the median, right?
Adam Butler: But what’s great about it is that they all have the same expected return but they’re not perfectly correlated to one another. So when you put them all together, you get the same expected return, but you get a lower volatility. And so that means you’ve got, over a rolling five year period, the expected tail loss is much smaller if you use the ensemble of all of them than if you use any single specification. Your expected maximum draw down is smaller.
Tobias Carlisle: … goes up.
Adam Butler: The expected shortfall is smaller. Like there’s all these benefits just from ensembling all these different methods.
Tobias Carlisle: So, that leads to your portfolio construction.
Adam Butler: Sure. Yeah. So, there’s a couple of different steps, right? That one is, well, what edges are alpha sources or what have you, do you think have merit, and are sustainable, and have produced economically significant marginal sharps over some sufficient horizon and across different markets and over different time periods? And so now we’ve got this set of preferential edges we have some confidence in. And so how are you going to weight those edges? But what’s so interesting is that if you’ve got a strategy with a Sharpe Ratio of around … call it a Sharpe Ratio of one, which is very high, but a strategy with a Sharpe Ratio of one and you’ve got 30 years of data on that, the standard error of the Sharpe Ratio around one is about 0.25. So, your 95% confidence interval is somewhere around 0.6 and 1.4, right?
Adam Butler: And so if you’ve got a strategy with 30 years of history that’s got a Sharpe Ratio 0.7 and another strategy over 30 years of history that has a Sharpe Ratio at 1.3, well, we statistically can’t say that they are different, they come from different distributions, right? So statistically, we sort of have to say, well, they have the same expected Sharpe Ratio, so they’re equally legitimate.
Tobias Carlisle: Even though one is almost double the other.
Adam Butler: Yeah, exactly. Yeah.
Tobias Carlisle: Right.
Adam Butler: Because the error term is just very large, and even worse, even if they come from different distributions, that difference may not manifest over a time horizon that matters to investors. The DALBAR study of investor behavior has a lot of flaws, but I think one of the things that we can rely on is their data on the holding periods for different classes of investments. And I haven’t looked at it in a while, probably in three or four years, I don’t think these behaviors change very often by very much. But the average holding period for an investor in an equity mutual fund is, call it three and a half years, the average holding period for a bond fund is maybe four years, and for a multi-asset fund like a balanced fund is four and a half years. So, even if your time horizon is 30 years, most people’s emotional time horizon is four or five years. And I think if you were to look at it at alternative investments, that that time horizon is even shorter because people just don’t understand them and they don’t have the same level of confidence.
Adam Butler: And so it’s not that your strategy needs to work over the very long term, it has to work over a time horizon that investors will stick with, right? So, really, you’ve got to kind of run bootstrap returns on combining these different edges over a three and five year horizons to see whether or not you should distinguish between them, right? And so that really makes you humble because you discover that there are a lot of edges that kind of have equal merit in over a three to five year horizons, you kind of want to own some portion of all of them because it just leads to a maximum probability of success over horizons that investors are going to stick with, right. And then there’s this whole portfolio optimization problem, which of course, we spent a lot of time on as well and is a whole different rabbit hole.
Tobias Carlisle: Well, let’s just talk very quickly about retirement analysis. You’ve identified some issues with the typical retirement models.
Adam Butler: Yeah, actually I spent a lot of time on retirement modeling earlier in my career back in sort of ’07, ’08 working on some of the models that Moshe Milevsky had proposed with his retirement safe withdrawal rates without re-sampling. And so he had this simple gamma distribution, you could sort of model it all analytically rather than having to run a bunch of simulations. But sadly what that missed is, is that the markets do not conform to the type of process that is well described by his analytical solutions. So, I just come back to this because I’ve been working on something with Andrew Miller, I think Andrew, who’s an adviser in Indianapolis, and who’s obviously thought a great deal about this problem. And so he was doing a bunch of modeling, and we started talking, and then I started thinking about this. And it’s good timing because over the last three, four years, we really started to incorporate a lot of machine learning thought process in how we think about the problem of portfolio construction. And part of that is this idea of ensembling or sort of blocked bootstrapping.
Adam Butler: And so when you come back to this process of retirement modeling and you think about it from the perspective of trying to capture all of the important dynamics and markets. So for example, a 60/40 portfolio, it’s not like a 60/40 portfolio has this completely random return trajectory through history. Rather, what happens is you’ve got kind of a 20 year horizon where 60/40 has zero real returns followed by a 20 year horizon where it has 16% real returns, followed by another 20 years of zero and another 20 years of 16, and so it’s really sort of a feast or famine type of process.
Adam Butler: And so any sort of Monte Carlo that assumes that those returns are randomly distributed, are independent and identically distributed through time are failing to capture that, and therefore it is overstating the probability of retirement success pretty substantially. And if you go back and apply, for example, a block bootstrap of 60/40 returns back to the 1920s, even for a US 60/40, which is biased massively upward … I mean, if you think about the US equity return premium is little over 6%, the equity premium XUS is about 3.5%, so the global equity return premium is kind of just a little over 4%, not including fees or taxes or anything like that, right? So, the US case massively overstates the expected premia in our opinion, but even using US 60/40, if you simulate the return process capturing the auto correlation dynamics of stocks and bonds through history, your save withdrawal rates actually come down quite a bit.
Adam Butler: So, just working on how to apply that new thinking or new way of thinking about retirement modeling, and then the impact of adding alternative return sources like trend or like diversified factor portfolios in terms of increasing safe withdrawal rates and using this new simulation process has been a lot of fun.
Tobias Carlisle: So, what does that indicate that the alternatives are? What do you do instead?
Adam Butler: Well, I mean, it can be as simple as adding trend, you can get a little bit more fancy and think about diversified style premia, I mean, like the AQR style premia fund or the BlackRock total factor fund, for example, these sort of diversified global style premia allocations. You can branch out and get even further into the weeds with hedge funds, like a Bridgewater type allocation or what have you. But I mean, the reality is the more diversified sources of returns that you add to the portfolio, the less likely you are to encounter a series of very negative contiguous returns. So, it’s these long periods of draw down that so derail your retirement outcomes, right? And so if you can just minimize the probability of those long U-shaped negative market environments, then you can very substantially increase your expected safe withdrawal rate. I mean, some of the numbers you get sort of a 25 to 50% improvement in safe withdrawal rates, which is pretty profound.
Tobias Carlisle: That’s coming up on time. If folks want to get in contact with you or follow what you do, how do they go about doing that, Adam?
Adam Butler: InvestReSolve.com, you’ll find our firm’s website. We’ve got a blog that we’ve been running for … well, I think I started that in ’09. When did you start Greenbackd?
Tobias Carlisle: December, 2008.
Adam Butler: Wow. Yes. So, I think I started a month after you, I think I started January, 2009. So, there’s lots of articles on there and it really traces the evolution of our thinking. And it’s not just my own thoughts, it’s the aggregated thinking and learning of our partners and guys like you who we have the opportunity to talk to all the time. So, definitely check out the blog. And then we’ve got a bunch of papers and podcasts and all that kind of stuff on the website too that you can poke through.
Tobias Carlisle: And you’re on Twitter @gestaltU, G-E-S-T-A-L-T-U.
Adam Butler: Yeah.
Tobias Carlisle: One of the better Twitter accounts to follow.
Adam Butler: Yeah. I’m feeling a little curmudgeonly but I’m trying to come off a little bit more friendly.
Tobias Carlisle: Trying to cut down.
Adam Butler: Yeah.
Tobias Carlisle: Adam Butler. Thank you very much.
Adam Butler: It’s a pleasure, Toby. I’m sad. Our guys are going to be out in California there in mid July, so I’m sad I’m going to miss that, but I’m looking forward to the next time we get together and have a pint.
Tobias Carlisle: Likewise. Thank you.