Mike: 01:20We thought about going out with, hey, we’re going to have this discussion maybe, and it might be hot off the press.

Adam: 01:26Yeah, that’s true.

Mike: 01:30 We’re going to have a Riffs, maybe, and things will happen on that thing. Anyway, I think it was kind of interesting.

Rodrigo: 01:33We hear you. Yeah, go ahead.

Adam: 01:35 We hear you big guy.

Mike: 01:38Yeah, last night was interesting too. So, we have up and coming guests that are that are going to help us understand how power grids and electrons are formed and how worm their way across the grids and how traders and power traders operate in that environment, I think. That’s coming up in the next few weeks. So, we were sort of pre-gaming that a little bit yesterday evening with these gentlemen and I am super excited about that one. Not to put a little teaser in for some upcoming guests, since we’ve been on a couple of times in a row, but when you think about what keeps your light on and your alarm clock going and your fridge going and the intricacies of turning, whether it’s natural gas or waterfalls into electrons and making sure that’s reliably up 99.9% of the time. Wow. That was fascinating. But yeah, as per the usual, yeah, it’s going to be a good one. So we’ve got some really good guests coming up, so just bear with us as you have to struggle through with the three of us on on a Friday afternoon.

And always, as a reminder, from a compliance perspective, this isn’t investment advice. In this particular case, we’re actually reviewing a research paper that Adam has spearheaded and is literally hot off the press. Just went to press, and if you’re looking for it, you’ll see it in the YouTube channel right below. Right there there’s, Peering Around Corners: How to Replicate Trend Following Managed Futures. You can download it there. You can also download it on Twitter. We have it there. So it is literally Hot, Hot, like just released five minutes ago. Ten minutes ago, maybe.

So we do realize you haven’t had a chance to look at it. We’ll be circling back with this. We’ve got lots more conversations, but we thought, you know what, we’d kick it off with a little bit of discussion. And as I said, not investment advice, do your own research. But we’re going to have a good, far-ranging conversation and dig into how you might think about replication of an index or a strategy like manage futures trend following, which is trickier than you might think. And there’s some things that you have to think about in your prioritization and how you might create that thing. And I think Adam has done an absolutely spectacular job trying to think through that problem and convey it in very approachable terms.

Replicating the Magic

Adam: 03:58Yes. Just to be also crystal clear, here Michael Harris is asking, are we marketing a product? This is absolutely not a product discussion. This is the discussion of a research paper that explores different methodologies that people might consider using for replicating, really, any strategy. And we just applied this in particular to CTA, trend following CTAs. And there are maybe some unique features of trend following CTAs that might make it a little bit easier to reverse engineer the underlying holdings or exposures of CTA managers in aggregate, than certain other types of indices which might hold a wide range of different products. And we may not know what the underlying products are that they might be holding. Right? Whereas with typically managed futures trend following, we’ve got a good sense of the general markets that these managers are trading on a regular basis, which gives us a real leg up. And we also have some decent intuition about the underlying strategies that they might employ. Right. So those are two important tools that we bring to the analysis that sometimes we aren’t able to bring to a replication analysis. So made it a little easier in this case.

Rodrigo: 05:39Yeah, it was a very interesting research cycle and the paper was quite enlightening because, what you normally hear with a replication paper is, at least in the past, it has been about trying to identify the factors that are driving returns in other mutual funds and strategies. Right? So if we think back at the original papers that Fama/French might have put out, and even AQR, trying to replicate what Warren Buffett invests in. You know, that’s just a way of peering through the data to see how well or how it is that they can replicate the magic of Warren Buffett in a unique way. And I think that that paper was really interesting because it was, correct me if I’m wrong, but I believe it was a levered portfolio of quality stocks 1.6 times, or something to that effect. Right?

Adam: 06:32Yeah, there was a quality exposure and then there was some inexpensive leverage that Buffett is able to employ by taking advantage of the insurance structure that he uses for investment purposes. And yeah, the authors definitely demonstrated that you could ex post right, giving Buffett full credit for having recognized this decades ago. But in the decades since, academics have identified certain factors that explain the variation in returns across different equity portfolios. And so the AQR paper was able to show that we could indeed explain the excess returns that Buffett has produced above just a cap weighted stock index, by just generating exposures to a small set of different factors like value and quality and also adding some leverage. Right? So that’s an example of something called returns based style analysis, which has been around since at least the early ninety’s, and probably long before. And the idea there is to use linear models, typically regression models, to identify the underlying exposures, and in some cases the underlying holdings of mutual funds or hedge funds or what have you.

Oftentimes it’s used to – so for example, if you’ve got a mutual fund and you’re evaluating whether the manager has skill, then you might run a returns based style analysis trying to explain the returns of that mutual fund using things like a long/short value model, or momentum, or the market portfolio or different capitalization stocks, small versus large, et cetera. And if you can explain the returns of that portfolio really well with those simple factor models, then some people might argue that that’s not really the manager adding his own skill because an investor can simply allocate to passive indices that give exposure to those underlying factor strategies and replicate that manager’s performance with very low fees. Right? So they can sort of circumvent the need to allocate to the manager with the manager’s higher fees for active management, by simply allocating to a combination of other index funds that would otherwise replicate the performance of the manager. Right? So the approach that we took in this case was at least partially informed by this returns based style analysis approach.

Rodrigo: 09:29Yeah, but before we get into that, I think Alpha Architect has a portal where you can go in, drop in data series and figure out how much of that manager has unique alpha, versus can be explained by a series of factors that you can by yourself. I think also from the, is it Two Sigma that has another portal that you can sign up for and see what can be explained, and if you find that it has many explanatory variables that are easily accessible and cheap. You might want to replicate it. And even if you’re not perfect, you’ll, you know, without the 2 and 20, you might be able to pull it off, right? And then if you don’t, if you if the thing is that you try to put the data through, see if you can replicate it and you find that 80% of it is unexplainable, then then that’s how you kind of are able to know whether there’s alpha there or not. So these are tools that are available now more than ever, that have, and one of the key things here is what is alpha and what is beta these days, right? It’s an ever moving target that managers and allocators need to be aware of because they might be paying more than what they need to.

Top Down vs Bottom Up

 

Mike: 10:27On that note, Rod, I think one of the things that I really found fascinating in the journey along this paper was sort of the I don’t know, maybe everybody else knew it, but for me, it hit home that when you’re replicating from a top down perspective and you’re trying to pick up what the positions are, you’re not necessarily concerned with how they’re creating the positions, which allows you to be participating in the innovation under the surface that’s going on with the various managers. Right? Whereas when you’re creating from a bottom up perspective and you’re saying, well, what are the general strategies that are used in this are, and do an ensembleing of those. So you’re sort of capturing best areas and trying to get the, maximize your signal to noise ratio. You’re still not sure where the whole market itself lies with respect to its innovation on trend and how different managers might be doing very different things.

And even though they’re a small portion of the overall group of managers, you’re still picking up that innovation when you’re top down. Whereas when you’re bottom up, you’re responsible for that innovation. You’re responsible for understanding those in the index and how they might be managing money and understanding the strategies they might be using. And so that was something to me, that was a really interesting realization for me, along this particular journey.

Adam 11:58Yeah. I think we’re getting a little ahead of ourselves.

Mike: 12:02Yeah, probably.

Adam: 12:04It’s a good teaser, though, for sure. Right. So we do begin the paper with a discussion of why someone might consider managed futures alongside a more traditional portfolio. Right? So, Rodrigo, I don’t know if, do you want to share your screen? Do you want me to share my screen? I got it up. And we just illustrate the performance of a global equity portfolio price in the US dollars, a bond benchmark, the Barclays, now the Bloomberg Aggregate and the Barclay BTOP50 CTA Index. We chose the BTOP50 because it just has the longest history of CTA performance. Right?

So each year, Barclays, going back to the 80s, Barclays selects the largest managers and puts them together into an index. They reconstitute that yearly, based on changes in larger managers and aggregate the performance of those managers. They report the performance of this index monthly, going back to 1987. And it was neat to see, obviously, the blue line there is the BTOP50 CTA Index. The black line is the S&P 500. Sorry is ACWI, rather is … Cap World Index. So a global cap-weighted index proxy. And the yellow line is the Barclays Bloomberg Aggregate Bond Index. Right. So what’s interesting to see is that they all sort of end up in the same place over, what is that now, 35 years? And obviously since bonds got there with much less volatility, they had a much higher Sharpe ratio than equities. The BTOP50 CTA index has a Sharpe ratio kind of in between bonds and stocks, but they generally have a long-term average correlation that is about zero.

Now, we all know right from 2022 that the long-term average correlation doesn’t mean much in terms of day to day, month to month and year to year. We went through a very long period where stocks and bonds, several decades anyways, where stocks and bonds effectively had a zero correlation and for the last decade, a negative correlation.

And then along comes 2022 with a major inflationary impulse. All of a sudden, stocks and bonds are reacting to the exact same macroeconomic variable, this inflation shock. And so their correlations converge towards one, right? So, they become very highly correlated. Now, fortunately, in the recent episode, the CTA index was actually negatively correlated to both stocks and bonds. So in 2022, the CTAs in general acted as a really nice ballast for both stocks and bonds, as stocks and bonds correlated and began to move in the same direction at the same time for a similar reason.

Rodrigo: 15:19Yeah, yeah. I think it’s important to just talk a little bit about the like trend following as a concept. I mean, you just are showing three equity lines and they’re different. This is what we all look for, right? Things that make money over time, but do it with taking a different path. You put them all together, you get a better result. But why is it that trend following managers have to have generally provided some much needed support in periods when equities and bonds do poorly. Can anybody tell me that? We’ll lay up for you guys…

Mike: 16:00The ability to go short?

Rod: 16:03Well, okay, Mike, I’ll take it if you need to. But I think the key difference here, the key value of managed futures is that you’re actually getting exposure to asset classes, that are number one, super liquid across equities bonds, commodities, currencies rates, right? And you’re able to short. Right? So what are the two blind spots that the 60/40 portfolio, 50/50 portfolio, equities and bonds, tend to have? Well, certainly anything that has equities dominating the allocation of the portfolio, like a 60/40 really has 90% of its risk embedded in it.

And so the direction of your 60/40 portfolio will be largely dominated by whatever equities want to do, right? So if we go to a prolonged multi year bear market, then that 60/40, 50/50 portfolio is going to be dragged down. What makes it a structural expectation for managed future strategies and trend to do, well they’re following trends. So if the trends are negative and they are prolonged, they’re long enough to be able to capture them, you should be able to structurally find non-correlation to that equity market. And then what we see in 2022 is another structural reason why, you know, having something like managed futures trend in your portfolio is that when inflation goes up and rates go up, right, you’re able to, generally speaking, bonds will go down and equities will go down, especially long duration equities, like growth stocks.

And you have the opportunity to short bonds and go long energies in inflationary regimes. So you just have access to many more levers than you did before. And what’s interesting today versus 20 years ago is that 20 years ago, it was a very specialized group of individuals, very specialized group of custodians that would give access to a subset of asset classes with less liquidity, and it was seen as alpha, right? You need it. You need, it was a big lift to be able to provide this non correlation. Today, fast forward 20 plus years, we really have turned this alpha, you could say, into something that’s more like a risk premium, right?

Not necessarily an S&P 500, easily replicable, but it is getting closer and closer to something that the average investor and portfolio manager can actually create and replicate. And so it’s starting to become more transparent. It’s starting to become more accessible. There’s more information out there than ever. We’re starting to see the value of it. And I think it’s one of these things that, if you look through all the alternative options out there for you to think about replicating, what you really want is something that has a low correlation to both bonds and equities and has an impact in periods when you need it to have an impact. And out of all the categories that we see, trend following, systematic global macro, tend to be the two that make, that fit that bill, right? So that’s important. I think that was, replicating a new beta is important and at a time where people are more accepting of this particular strategy. So, anyway, that was just to set up the idea of why trend is an important thing for portfolio construction, in our view.

Adam: 19:07I think it’s worth also mentioning that the fact that we can effectively replicate a good portion of a strategy doesn’t mean that there’s no alpha. Right. It doesn’t mean that there’s no skill. There’s absolutely no doubt in my mind, knowing many of the managers of large CTA funds personally, having read their research, having spent the last decade or more doing our own internal research on managed futures, there is an enormous amount of skill that goes into running a managed futures portfolio.

I think the whole idea here is that there is an element, or there’s a portion of those returns that are able to be systematically replicated. And that’s what we’re seeking to do here now. Right? And if you can replicate a good chunk of what’s going on at those highly skilled CTA firms and you can do it with considerably lower fees than those firms charge, then maybe there’s a fee alchemy. Sorry. A fee alchemy. Michael just posted alchemy there. But a fee alpha, that can be..

Rodrigo: 20:31He totally incepted you. That’s hilarious.

Adam: 20:34He did, too. There’s a fee alpha that might be able to be harvested. And I want to talk a little bit about the idea of alpha and beta, right, because you’ve sort of mentioned this a few times, Rodrigo. And alpha and beta, they do have technical definitions, but I like to think about alpha as really being very unique for every investor, because what a large, sophisticated institution might perceive as alpha is probably different than what a typical retail investor might consider alpha, right? Alpha really is what you personally can’t access cheaply and easily, right? Or in other words, beta is something that you can access cheaply and easily, right?

A sophisticated institution may be able to replicate or run a high quality managed futures program internally and not need to outsource that to other external managers, right? So for them, managed features is purely beta. But for a retail investor, they may have difficulty accessing top managers. They definitely can’t access top managers for very low fees. And so, to whatever extent we can turn some portion of the skill that’s being applied in managed futures funds into to a cheap beta and offer it for lower fees, or one can, then that’s a form of alpha that we can or not we, but one might be able to offer retail investors that they didn’t have access to before.

Rodrigo: 22:18Yeah, go ahead, Mike, go ahead.

Mike: 22:20The other thing is, you think about the idea of, so I’m going to do some trend following and managed futures. Okay, so where should we start? Should we start with trying to pick a number of different managers where the dispersion in this area is very large, it is quite different? Or should we start with trying to capture the main portion of what people are trying to do and then build around that with managers who we may be paying higher fees to, in order for them to capture unique viewpoints on this trend? Or if your bias is, I actually want more short-term trend following and the major indexes skew away from that? The major sort of replication stuff when you’ll read in the paper is that the shorter term side of the trend following is from the index’s perspective, largely nonexistent.

And that may be for a few very good reasons that we could speculate on. But so what is the core to this type of thing? And then where might I want to explore? What are my biases? Do I think it’s really long trend following? Do I think it’s really short? Do I think I need a lot of mean reversion in my trend following? Like, what are the things that I might want to add? But you can have a core of this replication at a lower fee. So it’s not to say that, I don’t think it’s to say that you want to replace all of it. You still want to look for, as Adam alluded to, there’s some really thoughtful, cutting edge CTAs that are, you’re kind of looking for those edge cases where yeah, I’m replicating the market and I’m going to get a lot of that innovation. But where do I think the innovation might be real? And where would I want to place asset bets or exposures to capitalize on that, and see if I’m right to some degree, said in that way?

Rodrigo: 24:12So, for example, we know a lot of people love Warren Buffett, love the way he chooses stocks. And with a long enough timeframe, you’ll be able to make some returns, hopefully above the S& P 500, right? That’s the whole point. That’s the whole jam. He’s got some structural components in there with the insurance policy, what he can invest in, how long he can invest in, and what leverage he can utilize. And so there’s clearly some alpha there that people who can get access to it, got access to it early and are able to do better than the S&P, you can. And institutions also search for managers with cutting edge, all the time. They have a team that helps them do that and put things together, right? The average retail investor has largely decided that they’re not good at that, necessarily, so they buy passive indices, right?

Instead of trying to pick a handful of really thoughtful US equity managers across that whole spectrum, a lot of people have decided to just buy the S&P 500 very cheap. So reduce the fees, get the big muscle movement of the S&P, and forget about the the bells and whistles, right? In a similar way, trying to identify, and trying to get access to the top CTA managers, they may not even and likely don’t go to retail. Right. And so, you know, what is the best way of doing an S&P 500-like index? You know, a way of doing that could be to replicate the BTOP50, the S, and the SOCGEN Trend Index, the SOCGEN CTA Index, because it does, it is representative largely of that kind of big muscle movement. And also, you know, if you’re replicating that, you’re going to get it at a level of risk that is common, you know, across all these managers and what they offer, generally. Other bells and whistles you can add or take away.

And as an institution, a lot of them prefer this is, they’re looking for 20 to 25 volatility targeted CTAs because when they put them together, that volatility could collapse and go way down. Right. And so this is about giving the best for the most, right? Also, if you’re going to be building something for the masses, you want to make sure that it’s as liquid as possible. You’re probably going to want to take away some of the smaller markets out there to ensure that you get that consistency and that liquidity across the board.

So these are just things to contemplate and understand the difference between what you can do with the Replication Strategy. But if you do have the resources and you do want to go out there and find the best, more power to you. That’s a way of going about it, right?

Adam: 26:40Rodrigo, can you share figure two?

Mike: 26:42Yeah. Or is the figure two the violin chart? Yeah, because I wanted a little dive on the violin chart because I think it’s relevant to that point, Adam, and it’s not a particularly regular chart that people might see, so might benefit from a little bit more explanation.

Adam 27:00Yeah, definitely. The point here is, too, that we talk about the CTA Index, but the reality is you can’t invest in, this is not an index you can invest in, right. Unless you go and buy all the managers in the index or buy units in the funds of all the managers in the index. So, people typically gain access to the, don’t bring that up. That’s not – that is false.

Mike: 27:23I totally agree. Yeah. If, like, you got to say, like, again, like, we have to drink every, like, time.

Rodrigo: 27:36For those listening in, Jason Buck has put in for every like in the video, ReSolve will reduce their fees by a bip, apparently. Drinking game. Let’s play a drinking game. And I’ll drink if someone says, like.

Adam: 27:50All right. Trying to stay on course, you can invest in the index, right? So people typically buy one or two or maybe a small basket of funds. Typically. I think a retail investor would just buy one, maybe. So that can be problematic because what goes on under the hood for any individual managed futures fund can mean that the return experience for that fund could deviate quite dramatically from the index, and from all the other funds in the index, over reasonably long periods of time.

So what we’re showing here, each of these yellow splurges here illustrates the dispersion of individual managed futures mutual funds in each calendar year, right? So the one on the far left is 2018, then 2019. And so, it just shows sort of the dispersion in the return experienced by all those different funds in that year. And you can see that, for example, in 2022, over on the far right, some managers of the mutual funds that we use, that were representative, that we just drew from the Morningstar Managed Futures category, actually had negative returns in 2022. And then there were other funds that delivered returns greater than 30%, right? So there’s a very large dispersion.

And you can see in each of the other years, there is still a fairly largish dispersion. On average, it’s probably five or 6% per year. And that’s on average. And if you you just happen to get unlucky or lucky, that could be very substantial and that may compound over several years. So just choosing an individual fund may not be an optimal way to gain general exposure to the category, right. Your experience may vary. And so this is sort of another motivation for, if you want general exposure to managed futures as a concept, maybe allocating to something that tries to replicate the index may be a better alternative than trying to choose any individual fund.

Mike: 30:18Yeah. And the other thing is the dispersion in the style and the approach in this space is quite large. And so you’re also faced with the challenge of understanding, was it skill or was it just that this is the repetitive process that they use, a process that they use, wherever you are in the world. So short-term trend followers had a lot of trouble for a long time. And maybe that’s why we don’t see a lot of that reflected in the index anymore. Because you go through a decade of really getting punched in the face and all of a sudden, the area dries up and they’re there’s not too many professionals even practicing that type of thing anymore. And so was it their skill or was it that they were unlucky in this particular set of circumstances didn’t manifest in their favor for a decade, and then what will happen over the coming decade? And so it’s an interesting, when you’re thinking about trying to disentangle the features of somebody’s sort of ethos, or what they feel is in their domain of expertise and how they’ve determined that they would look at the lens of trend following, is it just that that particular thing laid over top of a really opportune period of time? Or is it that they have skill and they actually are changing these types of environments? So just to say that it’s really hard to try and figure out who the best trend followers within the trend following space might be, and is it skill or is it a little bit of luck, as we’ve had many discussions on previously.

Adam: 31:48Yeah, there’s a lot of discussion in the chat about what the Lost Decade, which the managed futures category did have a rough period in the 20-teens. The question is, well, if a strategy can go through that kind of Lost Decade, then is it worth allocating to?

So, a few thoughts. First of all, the S&P had a Lost Decade from 2000 until 2012 in nominal terms, and a little longer than that in real terms, right? The S&P has also had very negative real returns through several other decades, over the last hundred or more. Bonds have also gone through very long, 40, 45 year periods of negative real returns. So I don’t think anyone’s making the case that trend following managed futures are a panacea. I think we have often made the case that you want to have a core exposure to global equities, to global fixed income, to global commodities, to trend following managed futures, to other types of global macro strategies and other diversifiers, if you can, where you have faith and you can get access to them. And yeah, we should, absolutely. You should expect that each one of them are going to go through lost decades. That’s the nature of the game. Right. The point is, I think this actually makes the case for us, right?

So, Michael Harris put in the chat that, you know, there was a 2% per year, or there was a decade where the managed futures index compounded at 2% a year, which was slightly negative after inflation. At the same time the S&P went up 11% a year. That’s the point, right? That’s why you want to own both managed futures and stocks and bonds in the portfolio, because you never know which one of them are going to go on to deliver the performance that you need over your investment horizon, right? And the more bets you have in the portfolio, the more likely you are to generate the returns that you need to support whatever your financial objectives are over your unique, personal, limited, finite investment horizon. Even if your horizon is, you know, several decades long, you know, global equities have gone through multi decade bear markets, especially in after inflation terms. So, yeah, diversification, man, get after it.

Rodrigo: 34:38And it’s also like what I was looking at this table, and Table 1 in the paper and just observing, you got from 1990 to today, you got MSCI at Barclays AG and the BTOP50, all roughly returning the same, and it’s six. These are total returns, right? These aren’t excess returns, Adam. Those are total returns. Yeah, yeah. So they may seem a little anemic, right? Six, five, 5%. But the benefit, like with the whole concept that we’ve been talking about, return stacking.

You know, depending on what your risk tolerance is, if you’re stacking six on top of five, on top of five, if that’s, if you have a high risk tolerance. And remember, these are non correlated return streams, so it’s not like you’re stacking, you’re stacking those returns, but you will not necessarily be stacking the same amount of risk to the portfolio, right? The volatility will go up, but it won’t necessarily go up by the same level as your return, your expected future return, right?

So again, you go through a decade where after inflation, trend does zero and the S&P does eleven, and you’re stacking one on top of the other, well, then you got your eleven. Then you go through a decade like the 2000 and 2010 period where the S&P does zero and trend does double digit, well now you stack double digit. And then there’s everything in between when they’re both making money, right? So I think again, I think a lot of times we have these discussions, we’re still stuck in this whole paradigm of the canvas being 100%, where you have to give up some of your favorite things in order to get some of the diversification.

I think today, in today’s environment and with the options available out there, you can do, you can have your cake and eat it too in terms of diversity and stacking. So something to consider there.

Mike: 36:33It’s also interesting, that period of time has one predominant direction of interest rates. Pretty sure that acts as a tailwind in the bond market, definitionally. That’s something that might account for that slightly higher Sharpe ratio from bonds over that period of time.

Rodrigo: 36:50What do we think about this comment from Sean? “Non correlation of managed futures to bonds and stocks may not be inherent to managed futures, but in fact is selection bias”. It’s possible.

Adam: 37:01He’s going to have to clarify.

Rodrigo: 37:04I think maybe Sean, you can describe this, but I think what he’s saying is we saw an equity line we liked ex post. And then we put it in because we hope that it does the same in the future, without really understanding what’s underneath. Whereas I think we can make a case, and I think I did earlier, as to why it is likely a structural reality that managed futures will continue to provide conditional non correlation, right, conditional non correlation.

Mike: 37:41Rod, do you mean trend managed futures or …

Rodrigo: 37:44Yeah, we’re talking about trend managed futures here. So I’m actually talking about trend managed futures. So specifically, you can create a base case for why the correlation needs to be there. If you have multiple time frames and look-back periods for trend, where you’re going to be long and short these different things, right? So it may not give you a high level of confidence that there might be a return there, although we can make a case on that, for that, with regard to Shannon’s Demon and Rebalancing Premium, where that’s maybe where the returns are coming from, regardless of whether trend provides positive outcomes. But one thing you can kind of intuit is that it will be –  the correlation or the lack of correlation is likely to continue over time.

Adam: 38:25Yeah, I don’t think we have conviction that the low correlation exists. Our conviction doesn’t come from the past history of low correlation. Our conviction comes from the mechanics of how the markets are traded, right, correct. First of all, there’s probably, over half the markets traded are not even related to stocks and bonds, right. They’re commodities. And commodities in general have approximately zero long-term correlation to both stocks and bonds. So there’s that. The other thing is they, managed features trade long and short, without bias. So, definitionally they’re going to be uncorrelated. We’re not suggesting that they’re going to be negatively correlated. There’s going to be periods where they’re going to be positively correlated, and then there’s going to be other periods where they’re going to be negatively correlated, and it’ll be largely random, the times when they’re positively and negatively correlated which means, you know, you probably shouldn’t try to time your allocation. But over time, structurally and mechanically, they’re going to be uncorrelated just because of the way that they trade without bias, long and short and because half the markets they are allocated to are uncorrelated stocks and bonds.

Universes

Mike: 39:35
It might be good to show the universe from the paper, maybe after whatever point you’re going to make here.

Rodrigo: 39:42Well, this is just kind of showing the condition, because one of the things is correlation. When we say correlation we’re talking about the average, right? But in reality, the correlation of managed futures to bonds or managed futures to equities is like having your head in the freezer and your feet in the fire. In the middle you have zero, but you have moments of high and low and negative correlations. This just shows kind of the trajectory of the SOCGEN Trend Index against ACWI and Bloomberg Global Aggregate Bond Index and you can see that it is, by virtue of being a trend follower, when the trends are strong on the upside, it will be correlated to equities.

When the trends are negatively to the downside, it’ll be negatively correlated and that’s kind of by design. So, from the fundamental understanding of what it’s doing underneath the hood, that’s kind of, also because a lot of times the other asset classes away from equities and bonds may be correlated to equities that you’re long, or you might be actually out of equities and might be long commodities, but commodities happen to be correlated to equities.

And so back to Mike’s point, what is it that managed futures trade in? Think the universe is here?

Adam: 40:40While you’re pulling that up, I want to address the capacity issue that’s been raised by Sean Wylan a couple times asserting that if too many people allocate to managed futures, then the edge will go away. I want to echo Brian Moriarty’s comment that that is true of every, of every strategy, including market beta, and also that there’s absolutely no evidence that that’s true. Right. So that the total CTA category is about $150,000,000,000 after you net out Bridgewater, which is not a trend following strategy at all.

So it’s $150,000,000,000. And it hasn’t really changed much. It doesn’t change much. Ah, you know, went up by like a couple, $10 billion after the 2008 financial crisis. We haven’t really seen much of a change in flows despite its fantastic performance in 2022.
The fact is it’s, just it’s actually a hard sleeve to allocate to. It does go through multi year periods where it underperforms stocks. Most people have very strong risk aversion to tracking error of domestic equity indices and just can’t stand to not be participating in the stock market when all their buddies are participating in it, and seem to be getting rich around them. It’s just a really hard thing to allocate to. So from that perspective, there is a little bit of a risk premium built into it. Michael Harris asking, futures a zero sum? Yes, definitionally futures are a zero sum.

So who’s on the other side? Well, typically, I’ve been trying to find this paper, but the IMF put out a paper in 2011 that demonstrated that large institutions, they examined insurance companies, public pensions, private pensions, endowments, and foundations.
And what they discovered is that they typically chase performance, but at exactly the wrong time horizon. And they showed evidence that that may be one of the contributors to why trends exist. If you look at major institutions, if you look at retail, at the money- weighted returns of institutions and of retail investments in mutual funds, etc., the money-weighted returns show that most people are really bad timers. They chase returns over two or three year periods in both directions and so they over-chase after a period of strong performance and they over-sell during periods of poor performance, and that drives markets in the underlying, drives strategies in the underlying markets and instruments that they hold, to trend at different horizons, right? I mean, there’s lots of …

Mike:44:03There’s also. There’s also the outside forces, right? The whole idea of carry and the way the term structure might be forming in the particular contract has a return to it, and that return is a return of preference and all kinds of things like storage and whatnot, and getting things through bottlenecks, and getting things through periods of time, or crop sessions, that outside of this system, where it appears that there’s two people that are on each side of every futures contract, that’s true, but there are external forces where a producer of goods and a consumer of goods will want to have price certainty over a time frame, and they meet at a price which gives a risk premium to those speculators that are offering liquidity to that market.

So, whilst it appears to be zero sum inside the game, there is a broader, outside of the game, where, that there’s preferences that are not just tied to the actual price. There’s actually, we could call them willing losers, but they’re just trying to hedge off some other thing that they’ve got on their balance sheet to worry about. And so, they would rather price certainty and someone else’s taking on that price uncertainty in order to garner a return. So it’s not zero, it’s not quite a zero sum thing.

Rodrigo: 45:32No. It’s like only seeing one leg of the trade, right. If a hedger comes in and is willing to, even though they may believe that the market is going to go up in price, they need to reduce the volatility of their cash flows so that they can go out to the capital markets and raise new IPO or go out to the bank and raise some cash. And they want to have, that bank is going to want to have some certainty of cash flows. And so as an enterprise, we’re seeing one leg of the trade where we’re looking at them and maybe calling them willing losers, but in fact they’re winning and the whole is greater than sum of the parts, right?

Rodrigo: 46:00So these are the elements that kind of are explanatory variables as to why we’ve seen a positive outcome from speculating and providing the other side of the bet. Another interesting one that I really liked back in the day when we had Chris Schindler in the podcast, is the idea of Queueing Theory, right?

Mike alluded to bottlenecks as a thing, and you should listen to the podcast because he’ll do a better job than I did. But the idea of Queueing Theory, which I think is an operations research kind of idea, is that you have a bunch of tellers that can take X amount of customers at a set amount of time. Like they can take one customer per minute and people, as long as people line up and one customer comes in per minute and they’re able to clear those tellers easily, there is no issue, there is no bottlenecks.

But the moment that you start randomizing which customer goes to which cashier, you all of a sudden start getting these bottlenecks and start in the, think about commodities. Commodities don’t clear easily, right? We have five long year, five long year cycles where or more, depending on the commodity, where you might have demand and there’s enough production in order to be able to say shit. That demand, the demand dries up, people stop producing. And then it’s, these bottlenecks and Queueing Theory that may create may make commodities, in particular, more susceptible to trending, right? So these are all things that you guys can …

Mike: 47:38Think of oil being at negative, what was it, $30 in March of 2020 or whatever it was.

Adam 47:44Yeah. Because they couldn’t fit any more crude oil at Cushing. Exactly. And that’s the bottleneck and that’s the problem. And so then you get creative solutions on that of where that’s going to go and what the risk premia, what the cost is for that to go to those other places and what people are willing to pay for it. In fact, there’s a wonderful story. As part of a hazing when you come into, a former colleague of ours told us this story about, when you come into the futures trading desks and some of the high frequency shops, what they will do is tell you that they’ve actually been assigned several hundred thousand head of cattle and it’s being delivered. So, you got to find a place for it. To teach them how you don’t want to be assigned any of these contracts. And they run these juniors around to try and find places, which by the way is literally impossible to do.

And if it’s possible, it’s extremely expensive to do. And so they just want to emphasize that you don’t want to get caught in these situations. We’ve tricked you. It’s fine. You don’t have to find a place for for 2000 head of cattle. But if you did, you now know that the saliency of why you’re going to want to close that contract before you take delivery.

The Nitty Gritty of Replication

Rodrigo: 48:51Okay, let’s shift away from the the trend side of things and go back to the replication side of things, and maybe we can get into the nitty gritty of replication, Adam? And what’s unique about replicating a fast moving trend strategy versus other things, for example. Why don’t we get into that?

Adam: 49:12Yeah, absolutely. So we approached it from two directions, right? So, by the way, we’re not the first ones to to talk about hedge fund replication. Obviously, there was a product launched a couple of years ago that’s been very successful and done a good job of tracking the CTA index. And there have been plenty of papers written about hedge fund replication. There’s a particularly good one by Lars Kesner on CTA Replication. So, we drew inspiration from the literature and from successful products as well. My understanding is that some of the more successful products that do replicate the hedge fund index use what we term to be a top down approach, right?

So with the top down approach, you’re sort of looking back over the last 20, 30, 40 days of returns of the underlying index. And you’re using regression analysis, a robust regression analysis called ridge regression to or elastic net to figure out what combination of assets in a portfolio, if you were to hold those assets over that same period, would best track the returns of the underlying index, right?

This is the same concept, exactly the same as a returns based style analysis. And by the way, there are more sophisticated ways to do this, right? You could use RNN, you could use a LSTM, long/short term memory type, neural network model, which would account for the fact that the holdings auto correlate through time, et cetera, et cetera.

In the paper, we kept it pretty simple. We just used an ensemble of rolling regressions where so, for example, look back over the past 20 days. We’ve got a basket of instruments that we’re going to use to explain the movements of the underlying index, and then we’re going to try to create a portfolio that best mimics those movements.

Right? So it’s called top down. So we just use a 20 day, 25 day, 30 day, 35 day, and 40 day. And then the question is, well, how do you weight the shorter-term models versus the longer-term models? And we just used kind of a metamodel where we regressed the underlying tracking error between the different models and, to find whether we want to give the longer-term models higher weight or the shorter-term models higher weight, at each individual point in time, right? And then we just took the weighted average of the weights that came out of each of those different models, and that became the sort of top down explanatory portfolio on that day, for what is probably close to the underlying holdings in aggregate of managers in the index.

Rodrigo: 52:13So, this would be, again, going back to Buffett, just to keep it simple, this would be what the AQR approach would have been, as to trying to identify the weights and then replicate it.

Adam: 52:25It’s a little different, I think, because I think that the Buffett one is a little closer to the, a little closer to the bottom up, I think, right? Because the idea is you’re trying to use the same process to select stock that Buffett used, right? Whereas in this case, we’re literally just trying to figure out what are the underlying holdings.

So this has some benefits. As Mike mentioned earlier, the benefit of this type of top down replication is that if you don’t have a good understanding of the likely methods that the underlying managers are using to inform what goes into the portfolios, then you could just, you know, you could just see or try to look through to approximate what their holdings are, through time. And therefore you’re going to be able to take advantage of any innovation, as Mike mentioned, that might be going on in the space, right.

Please keep in mind these trend following managers, many of them have been around for decades. They’re constantly innovating. And so this top down method may allow for the potential to better take advantage of some of those innovations through time. On the downside, you’re trying to mimic what’s in the portfolio over the last kind of 20 to 40 days. A regression analysis, even robust regression analyses, can come up with relatively noisy approximations when the number of observations you’re using to fit the model is relatively small, relative to the number of variables.

Rodrigo: 54:01So, for example, even in our small universe that we’re going to – so we should probably go through the universes.

Adam: 54:10Yeah, let’s go through the universes. Maybe actually, just can you pull up the universe thing again? So, for the top down modeling process, we use both a small universe, which is nine markets, and a larger universe, which is 20 or 27 markets. And you can see although it’s not super clear in this image, but you can see that the small universe is a subset of the larger universe, and the holdings in the small universe have a red square around them.

So, for example, FV, TY, TU, US. Yeah, if you zoom in, you can see right, so there’s nine markets. It’s just meant to be sort of a small representative basket. You’ve got some bonds, you’ve got some equity markets, you’ve got oil, CL, you’ve got gold, right? And that’s pretty well, you’ve got a couple of currencies, euro and yen and that’s pretty well it, whereas the medium sized universe, we just add more markets in each of the different categories, right. For the purpose of creating an index that can be easily scaled to many hundreds or of millions or even billions of dollars, we left out any illiquid markets where investors might face some constraints in terms of CFTC limits, et cetera, right?

So for the top down approach, we ran this rolling regression, trying to approximate the holdings of the CTA index, using both the small index and the medium sized index of markets. And for the reasons I discussed, you’ve got nine markets that you’re using to explain returns over 20, 30, 40 days.

That regression is going to be noisy. The models are going to be noisy, even with the regularization techniques that we use. And so, that is definitely a downside of top down models. And also what’s interesting to see is that, maybe just go on to the next one, if you don’t mind, is that the top down model and the bottom up sorry, the top down small universe model and the top down medium sized universe model had slightly different return trajectories, right? They’re reasonably highly correlated – correlation of about 0.7 between them, but they outperform and underperform and track the index in a relatively better or worse way at different points in time, right? So what’s great about that is that you can combine them and get sort of the best of both worlds there. And so this is what you see. The yellow line is the top down medium universe. Black line is the top down small universe, and then the blue line obviously is a combination of the two, right? They all kind of meet at the same, in the same place. The combination does a little bit better on a risk-adjusted returns basis because you’re getting the the diversification benefits between the two types of models.

And the combination strategy also tracks the underlying SOCGEN Trend Index a bit more closely than either of them on its own, right. So a perfectly reasonable approach to use, my understanding is, that this is a closer approximation to what some of the more popular CTA tracking ETFs use under the hood and it does reasonably well, right. But it’s not the only way to go.

Rodrigo: 57:51Let’s talk about the pros of that. Once again, just to summarize, I know we’ve already addressed them. So what are the pros of an approach like this from a replication perspective?

Adam 58:05The pros are that you are directly trying to mimic what’s going on in the underlying manager portfolios. So you need to make fewer assumptions about the strategies that the managers are employing in order to create the portfolios, right? You’re literally just trying to mimic what they’re doing. You’re not making any assumptions. You’re making some assumptions about the markets that they’re probably trading within them, right? But we know in general that most CTAs try to trade across a broad basket of markets.

A bunch of equity indices, bond indices, commodities, currencies, some of them trade more exotic indices, calendar spreads, et cetera. But sort of broadly, that is a representative basket of the exposures that most of the managers are employing, right? The drawbacks of using this kind of approach are that you may be out of sync in terms of your modeling with changes in the portfolio, if they’re happening relatively quickly.

So, you know, we’re looking at the performance of the portfolio over the last 20 to 40 days. There’s obviously a lag between our models and what our models are representing that the underlying funds are holding, and the changes that might be going on within the underlying funds, based on their unique mechanics that they’re using to drive their portfolio selection, right? So pros and cons that the bottom up method can fill in some of the gaps on.

Rodrigo: 59:35 And so the detraction is that it might be a little delayed. Now how often in the paper are you refreshing the portfolio or rebalancing the portfolio?

Adam: 59:46 Every day.

Rodrigo: 59:49So this is a daily rebalance. Right, okay, so that’s top down approach. Now let’s get into the bottom up approach and why that was a separate useful tool.

Adam: 01:00:01Yeah, sounds good. So the great thing about managed futures is that we have a general intuition about the strategies, the general strategies that people are using, that these managers use to identify trends, right? Some managers use moving averages, moving average cross, double moving average cross, breakout strategies, et cetera. It may seem like these strategies would produce very different profiles.

In reality, they overlap by quite a bit. Shorter-term breakout strategies have a lot in common with shorter-term time series momentum strategies, and certain types of moving average strategies, and double moving average strategies, et cetera. And we can see that as we sort of look through the results of our bottom up analysis. But the idea here, is we have a general intuition, they’re using, you know, past returns over look-back horizon, somewhere in the neighborhood of, call it, 20 to 300 days to inform whether a market is in an uptrend or a downtrend, right?

So if the past 20 day returns were positive, then, you know, the shorter-term trend portion of the manager’s strategy would obviously be looking to hold a long position in that market and vice versa for negative returns and short positions, right? But what we don’t know is what markets are they trading and in what proportion, right? And what strategies are they employing? Are they skewing towards shorter-term strategies or are they skewing towards longer-term strategies, right? So what we do is we take a representative basket of all of these different markets, the 27 markets in our medium universe, and we run simple trend following strategies on each of the markets independently, right?

So, we’re running a short-term trend strategy on oil futures, a short-term trend strategy on S&P futures, et cetera, et cetera, right? Medium-term trend strategy on oil futures, medium-term on equity futures, across all the different futures markets and across a wide variety of different trend length specifications from about 20 days up to about 250 and 300 days, right? And then we use, again, ridge regression or robust regression in order to identify the market, the market strategy combination, that best explains the returns of the managed futures index, right?

Maybe you can put up, I think it’s Figure 3, Rodrigo, because I think it’s interesting to dig into the, specifically the trend lengths. So how the trend lengths were emphasized. Here we go, where the trend lengths were emphasized.

So obviously the managers in the CTA Trend Index, they don’t really trade short-term trends, right? We did allude to this earlier. Typically they skew towards longer-term trends. In fact, about, I think, 95% of the weight across all these different strategies went to strategies with greater than 90 day look backs, right? So, you can see this in the chart. This is Figure 7 for those listening, I’m going to follow along later. So not figure three. Yeah. Which makes me think that maybe the numbering may be off in the figures anyways.

But this figure shows that the 260 day return is most widely used by trend followers across all the different markets, right. Some people might also be wondering what markets were chosen. And it turns out that all of our 27 markets across equity indices, bond indices, commodities and currencies were selected for inclusion, though some of them received more weight than others, and some of them received different weightings in terms of what trend length exposures were allocated to within the model, right? But I think this is consistent with our intuition that most of the managers in the trend index, large managers, tend to trade based on longer-term trends for a few reasons, one of which is liquidity, right. Longer-term trends just incur a lot. You’re not trading nearly as much, right. So you’re incurring less trading costs, and you’re also not incurring a lot of trade frictions as you, as you’re going in and out of positions.

So just in general, as you listen to trend, big trend following managers describe their process, most of them will typically say, we allocate to sort of intermediate to long-term trends. And we do observe that in the data.

Mike: 01:05:23Yeah, and I think there’s a liquidity issue there, too, spreading the bets across multiple contracts, quarters, because you do run into a challenge with respect to capacity if you’re just trading the front month, and if you’re shorter-term, it becomes harder.

Rodrigo: 01:05:44And short-term managers exist out there, and they have their place, and they don’t take a lot of money. They provide their alpha for a small group of investors, and then that’s it. So, it’s just not, again, broadly something you can make broadly available to…

Mike: 01:06:00No, precisely. It’s really tough to index that or bring that to the masses in a cheap way.

Rodrigo: 01:06:09All right, so in contrast to top down, we just described bottom up. What are the pros of bottom up in contrast to top down?

Adam: 01:06:17So the bottom up empirically tracks the index better than the top down. It also has lower turnover than the top down, and it has just better long-term performance as well. So, but we’re making some assumptions, right? So, the drawback here is that we’re making an assumption about the underlying processes that managed futures managers are using to identify trends. And if the processes that we’re bringing to the table to model the returns don’t overlap to a meaningful extent with the strategies that are used in aggregate by the underlying funds, then we could be in error. There could be some more dispersion between what we hold in the Replication Portfolio and what the actual managers are holding in the index, right?

Rodrigo: 01:07:10Yeah. So in the one case you’re getting, the top down allows you to evolve faster as the manager evolving their strategies and their trend links and the like. So if the goal here is to replicate that index, you will be closer to that. Whereas the bottom up approach is, it’ll take a little while longer to reassess and re-mine the new types of strategies that they might be deploying. So a little bit less agile in terms of innovation, but likely more agile if things don’t largely change too much, right? More agile in that you are getting in and out of markets as they trigger, in contrast to looking back, having a bit of a lag and using the performance in order to inform whether it should be long or short.

Adam: 01:08:03Yeah. So actually we did some experiments that I think it is worth discussing. And I guess that would be Figure 8, Rodrigo. I’m hoping so. The question is, first of all, does our modeling work? Is that Figure 8? No. Then it’s the one before that. Okay, go up one more. Yeah. Okay, perfect. So it’s great to say we could look at all the long-term returns of the CTA index all the way to back to 2000 and say, well, yeah, we can model the underlying strategies and markets that go into that index relatively well.

But the question is, does, will that work out well, out of sample, right? So what we did to test that, is we used something called k-fold cross validation, which is a very common technique in data science, where you take 90% of the data and you fit a model. So in this case, we’re going to try and replicate the CT, the SG Trend Index, using our market strategy pairs as explanatory variables over 90% of the the term of the underlying index. And then we’re going to generate the weights that we should allocate to those market strategy pairs, and then we’re going to apply those weights to the 10% of the data that the models were not fit on and see how they do, right? So is the exposures that we’re  identifying that drive the portfolio changes in the SG Trend Index in one period, do they persist in another period, right? Are they relatively stable through time?

So what we did here, the blue line here, is actually where we took the first 90%, used the fit, applied it to the final 10%, and then we took the final 90%, made a fit, applied it to the first 10%. And so we did, we divided it up like that, in the ten different sections, and then we just stitched together all of the out-of-sample results. That’s the blue line. The yellow line is if we just did a full fit on the entire sample from 2000 and saw how that explained the underlying models, and you can see by virtue of the fact that the dark blue line and the yellow line overlap each other almost perfectly, they have a correlation of over .99,
that the styles or the strategies and markets that are identified by this underlying process are highly stable and consistent through time. And so that gives us the confidence to be able to use the full series to model the underlying processes using the full series, and that those underlying models are likely to be go on and perform similarly well in tracking the index in the future.

 Rodrigo: 01:11:22Yeah. And just, Mike says something that is very interesting, that that outlier in 2014 looks interesting. 2014, 2015, that was due largely to a 75% collapse in the energy markets and the commodity markets in that period. If you look at that, that is a clear defined trend that lasted long and went for 75%, rare events. But that’s the advantage of having that the ability to trend follow commodities when you have a bond and equity portfolio, right?

Mike: 01:11:52Well, I think that’s the whole point is, it’s kind of going to be like lightning. This is why trying to time the allocation from the perspective of picking it, with trying to think that something’s going to happen, is tricky. You have to kind of have the allocation for when the thing that creates the opportunity, happens.

Adam: 01:12:04Absolutely. And I like to some extent, how some of the trend followers, like Jerry Parker, describe trend following as outlier hunting, right? You’re really looking for these external, extremely large, persistent moves that you have no idea which market it’s going to happen to, you have no idea when. But we do know that market returns are highly non- normal. They’re highly …. many of them are highly skewed. And so trend following is able to pick up on some of those major outlier moves, and that’s what makes those big outlier years for some of them.

Mike: 01:12:59Cut your losses, ride your winners. Yeah. As trite as that is, it is the waiting for those lottos, or as Mike says, the lottos or those outlier events to occur, capitalizing on them and then waiting for the next one.

Rodrigo 01:13:14People often talk about the smile characteristic of a managed futures strategy, right. Where you get a lot of observations and a lot of data points and performance that are in like, the low returns, right in the middle. But then you have, whenever there’s an extreme positive or negative event, there’s convexity to the strategy that provides that diversity. So that, again, structural. We’ve seen it, we can explain it, we can understand it, and that gives us confidence that it’s likely to be there in the future.

The whole point of this paper is to identify interesting ways. And I know you and Andrew Butler, the CIO at ReSolve Canada, worked really hard to kind of use the right techniques and put them together in a way that makes sense. So what is the best approach here from a ReSolve perspective? We pick one over the other?

Adam: 01:14:00Yeah. So, I mean, as usual, at least for the purpose of the paper, we just decided to give 50% of the weight to the top down approach and 50% to the bottom up because they complement one another’s strengths and weaknesses, right? The bottom up one suffers from the fact that we may not have a totally representative set of strategies to explain the returns of the underlying index, but it is a lot more consistent over time. It does a better job of tracking the index and has overall better performance, and a lot less turnover. The top down has the benefit of being able to react to whatever underlying innovations are happening in the funds because we’re just looking through and trying to capture the holdings of the funds from day to day. But we also may be wildly out of sync from time to time with portfolio holdings, because of the lag between the period that we’re assessing the fit over, and what may be happening in the underlying portfolios, from time to time.

So I think they’re just nicely complementary from a theoretical standpoint,. and empirically. When you put them together, they just do a better job of tracking the index, and they have demonstrated better long-term performance. Maybe you can put that up, Rodrigo.

Rodrigo: 01:15:36Yes. Looking for it? Yes. That’s Figure 8. That’s the one you’re looking for, right?

Adam: 01:15:50Yes, that’s the one. Yeah.

Rodrigo: 01:15:53This requires some explanation.

Adam: 01:15:57Yeah. So, first of all, this is all excess returns, okay? For futures strategies, you deposit some margin, and you’re able to get a lot of exposure for having very few dollars available to you, right? So, what I didn’t do here is include collateral yields. Typically with managed futures, you post collateral at an FCM, like a futures trader, and then you use that collateral, which is relatively small, relative to the size of the actual exposures you’re getting.

You’ll typically get a yield on that collateral. So they’ll invest that collateral, they’ll give you some, probably a discount, on T-bills or whatever. But then there’s a huge amount of cash. 50% to 70% of the portfolio is just held in cash. And that cash obviously, can be held in T-bills or high interest savings accounts or other types of strategies to generate a yield on it. So what I’m showing here for all of these different strategies is just the excess returns. So the excess returns, in excess of cash. We did the same thing for the SG Trend Index in blue. That SG Trend Index is the performance of the Trend Index in excess of what you would have gotten from investing in cash, all right? So that all of the lines are apples to apples for comparison purposes, okay? The light blue line at the top is the combo, the 50/50 combo replication strategy, net of our estimated fees, sorry, commissions and slippage, but no fees.

All right, now, we do make the case that very few people will be able to run this strategy on their own with no fees. They’ll probably have to allocate to some kind of product or something, in order to get exposure to it. So we just took a 1% fee, as maybe a standard fee for other types of funds like this that are out there. And so the black line is the replication strategy, net of expected slippage and commissions and net of a 1% fee.

The blue line is the excess return trajectory for the SG Trend Index. One of the complications here is that the index itself is a combination of funds. Each of those funds charge a slightly different fee, and they almost all charge performance fees. The performance fees are charged typically after they reach a high water mark. Different funds are going to reach high water marks at different times. Different investors who invested in those funds are going to hit high water marks at different times. And so we’re not, it’s not like you can just run a 2 and 20 overlay on the net replication strategy, and basically get the same returns as the, as the underlying index because all of the underlying funds have charged their own performance fees at very different times along the way.

So in order to illustrate how well the strategy tracks the underlying index, we just ran the Replication Strategy, but, where we set the Sharpe ratio of the Replication Strategy to be equal to the Sharpe ratio of the SG Trend Index, over all rolling three year periods, right? So the yellow line there is really just to illustrate how well the Strategy mimics the character of the SG Trend Index. It is not meant to sort of illustrate any sort of performance expectations.

Is that clear or did you guys want to have…

Rodrigo: 01:20:00No, that was clear. I think we were talking about this internally. When you don’t include a fee, it just seems like it’s really tough to tell how well we did, right, with this experiment. It’s just a matter of, okay, how do we get the results that we got and be able to visually show that indeed, we are doing a pretty good job using these techniques, of replicating the index that we want. It’s all it is.

Adam: 01:20:25Exactly. Maybe just scroll down to the performance table. So first of all, remember, the goal here is to track the index, right? So in reality, the actual performance is less important than how well the strategy correlates to the SG Trend Index. So you can see in fact that we get a correlation of about .86 which is pretty high. The, if you look at the live correlation for some of the other replication ETFs, their correlation is quite a bit lower than that. Right? And in fact, what we do observe, I don’t discuss this in the paper, but I can just discuss it live here, is that the correlation of the Replication Strategy has actually been higher in the recent period than in the distant, past period.

So in fact our tracking seems to get a little bit better over time, at least empirically in what we’re able to observe. I’m convinced that the majority of the improvement in annualized returns that we see, right, the difference between sort of 8, 9% excess returns versus just shy of 5% excess returns for the SG Trend index, is mostly just that fee delta, right? There’s no performance fees and even a 1% fee is lower than the typical 2% management fee that most futures funds charge. So that gap there, that, call it 3% gap, is approximately the fee alpha that you might be able to potentially capture if you were to allocate to this in a low fee tracking structure instead of allocating to a basket of funds that are all charging 2 and 20 performance fees.

Rodrigo: 01:22:20Very well said. Yeah, so that’s basically it on the paper. The link is included in the top of the chat and it’s included in the show notes and YouTube and in your favorite podcast listener. You can always go to Investresolve.com to download it as well. And yeah, we hope it was educational, useful, made you think about the concept of replication in a unique way. Is there anything else that we didn’t cover, Mike, Adam, that we think we need to cover here?

Mike: 01:22:54No, I’m not…

Adam: 01:22:57Someone asked the about the impact of slippage and commissions. I want to say that that was a little over 1%. About one and a half percent was what came out because of commissions and and slippage. Estimate. Estimates.

Rodrigo: 01:23:08If you really want to get your hands dirty at the end of the paper, there’s an appendix that goes into the different regression approaches that were taken, and get into more nitty gritty detail there.

Mike: 01:23:22Yeah, I would say, yeah. We’re welcome for more feedback. This was, as we mentioned at the outset, is hot, hot, hot off the press. And so we’re looking forward to more feedback and more discussions on the paper to dig into the areas where people might be unclear, want more details and those types of things. So it is launch day one, I guess, is it? Where yes…

Rodrigo: 01:23:47Do reach out to ask questions, please.

Adam: 01:23:52Like and Share the paper around. If you find it useful, please hit the Like button on this podcast. Please subscribe, if you haven’t already subscribed. And thank you very much for tuning in.

Rodrigo: 01:24:08Thank you, everybody.