The Exchanges

Q talked a lot about indexing and you've made the very astute observation that if Indexing means that those weaker players are coming out of the market. It could get even more difficult. Do you have any feeling for what the tipping point was over the last, it's really three, four, five years where all of a sudden massive flows, you know, Vanguard's raising a billion dollars a day. Why now?

A Well, I think there are a few things and, and, you know, when you see this thing really kick into gear was probably On the heels of the financial crisis. So let's call that roughly a decade, a little less than a decade ago. And that's where I think you start to see a little bit of the acceleration. And if you look, if you, and you really said the last, you know, 12, 24 months, I think it's the Department of Labor rule on fiduciary responsibility. In our piece where we wrote about active versus passive, we try to take a very long-term view and look at the role of regulation and how regulations encourage not only mutual funds in general, so sort of more professional investing, But also indexing as well. So I think the DOL rule itself was probably a pretty important specific catalyst for acceleration of indexing. And it makes sense, right? Because if you have a fiduciary responsibility, and you put someone into a product that competes with the S&P, and they do substantially worse, you don't want to expose yourself to any sort of liability in that regard.

AI assessment note: “I think it's the Department of Labor rule on fiduciary responsibility.”

Q What are the implications as you get your head around the distribution of outcomes in public markets for the optimal way to think about participating in public markets?

A Boils down to two things, seems to me. One would be just cast a wide net. We said, 60% are going to be bad, 40% are going to be good, two percent are going to be spectacular. Just index. And if you index, you're going to get the bad, but you're going to get the good, and you're going to capture that skew, and that would be, again, if you take the public market equivalent returns for venture capital, it's obviously very, very cyclical, but it's been pretty good over very long periods of time, and this would be essentially saying you're going to own every venture fund, you're going to participate in all the goods and bads, but over time, it's going to do pretty well for you. So indexing would be answer number one. The second thing to say as a fundamental investor is, might I be able to identify Those companies that are likely to do this over time, and it's a probabilistic assessment, but can I find those stars and put in my portfolio and hope to get much better returns? So that would be the second approach. And so that also automatically then says, all right, well, what are the characteristics of those kinds of companies? Can we identify them? Again, a priori pattern recognition, right? Going back to pattern recognition, are there things that we should be looking for? So that would be the second strategy. And I think that's what probably most active managers, fundamental managers…

AI assessment note: “Boils down to two things, seems to me. One would be just cast a wide net.”

Q As the public markets, I guess you could say get more concentrated, and we certainly know that from the Mag-Seven, but also this washout of microcaps. What have you seen over the years in how public markets have generated returns?

A It's interesting because venture capital is almost a completely shrunken time horizon of what happens in public markets. And this is another area I find fascinating. This is all worked by Hendrik Bessenbinder at Arizona State University, Professor of Finance, and he's got these jaw-dropping statistics. He studied every public company that's been around since the 19 twenties. Again, since the CRISP database was established, there have been a little over 20,000 companies. And what he found is Just under 60% earned returns less than treasury bills on a monthly basis. So just take a moment to take that in. Of all the companies that have ever been public in the United States since the 19 twenties, let's call it a century, just round it. Almost 60% have failed to earn treasury bill returns, and they destroyed an aggregate of nine trillion dollars of wealth by his calculation. Okay, so the other 40% plus have obviously created value, and they've created 64 trillion dollars of wealth. So the aggregate wealth creation, and this is through 2022, we'll have to update for 2023, was 55 trillion dollars. What's remarkable is if you distill that even one step further, the top two percent of that 28,000 created 50 trillion dollars. Of the 55 trillion in total. So just two percent of all these companies are essentially 90% of the total wealth creation. So again, it's worth taking a moment to le…

AI assessment note: “just two percent of all these companies are essentially 90% of the total wealth creation”

Q So to put color on that in the public markets, what does it take to own some of these, say, two percent winners for the long term?

A Besson Binder wrote a series of papers, and I would recommend people go check them out. If you go to SSRN. You'll find these papers and they're accessible for the public to download. And he found a couple of things. One of the things that not shocking to people would be they performed well. So they had good sales growth and profits and cash flows and so forth. So all the fundamental stuff was fine. The other thing he pointed out technology companies, they're not overrepresented actually. And so this is something we forget about this a lot in life, but everything's a distribution. And so even if a right tail is fat, doesn't mean the left tail isn't fat as well. So a lot of technology companies do well, but a lot of them fail. Same, we were talking about biotechnology, same thing. So you have to look at the fullness of the distribution. But I think the one that was most striking was that he documented that almost every one of these great compounders had massive drawdowns at some point. And it was common to have drawdowns of 75% or more, but some were 80, 85% drawdowns. And to state the obvious, it shakes out all but the very hardiest of shareholders. You buy XYZ, you think it's great, you think it's gonna be a compounder, and it goes down for a whole series of reasons, could be market generator, some specific stuff, goes down to 80%. That's just a hard one to overcome that and si…

AI assessment note: “it shakes out all but the very hardiest of shareholders”

Q investment management tech, request a demo at ridgeline.ai. And now back to the show. Okay. So hopefully this part of this conversation get a little bit of a grounding and how to think through how people spend their time now. I want to get away from that a little bit and ask you the question I always love asking you, which is what research project are you working on now?

A So the big thing we're working on, I'll mention two or three things. The first is we have a piece that's going to be out shortly that picks up a lot on this work by Danny Kahneman on the idea of noise. And in particular, there are a number of colleagues at the University of Pennsylvania wrote a paper on what they call the BIN model, B-I-N. B stands for bias, I stands for information, N stands for noise. And they did something really interesting as they went back into the Good Judgment project and looked at the difference between the quality predictions of the super forecasters. These are these elite forecasters and regular forecasters. And by the way, I should say these regular forecasters aren't that regular, right? Because whoever's nerdy enough to go into a casting tournament is not a normal person to begin with. We'll set that aside. So the question was, so we could say there are three possible reasons they're not as good as the super forecasters. Bias would say they're systematically biased, right? So they're, the bias is getting in their way. Information just means they're not as informed as the super forecasters. Okay, possibly. And noise is this idea, just to define noise specifically, noise is this idea of a non-systematic departure from the right answer. The punchline of this, which was, I think, really the key lightning bolt, was that when they decompose this, they f…

AI assessment note: “The first is we have a piece that's going to be out shortly”

Q How different is it when you have, say, a fund where there's really a, a sole PM compared to there may be a PM, but they have more of a group process?

A So there's interesting research on this, and I think this is relatively new findings last two or three years. What it was found was that the funds that delivered the most Alpha were actually funds with three portfolio managers and better than single managers, and by the way, better than two and four. So what's interesting about three is odd number, right? So you don't need to have a consensus. You can have two versus one and make decisions. So that's an interesting question is whether, I mean, the whole purpose of things like committees or groups is to try to find that center ground and avoid extremes that an individual may be more subject to. I think the challenge in any sort of team setting or committee setting is that you really have to be extremely mindful about the process in that case. And so the first thing is to have a team of people, if you're going to say, let's say, do three, four portfolio managers, that you really want to make sure you have cognitive diversity. So people that really do think about the world, maybe different sets of skills, maybe different personalities. And then the second thing is you really want to manage the process effectively. Right. Which is the key to why you have group decision making is that it allows you to surface all alternatives, vet them appropriately, and then decide. And so the point of a committee or a team is to allow you to do th…

AI assessment note: “funds that delivered the most Alpha were actually funds with three portfolio managers and better than single managers”

Q you know, a couple of weeks ago, I interviewed Paul Rabel. I know you're a lacrosse player in college. So I actually, relative to other sports, know very little about lacrosse. So if someone is going to tune in in June and July, whatever it is this year, and watch the PLL on television, what are the key? So if you're doing data analytics on lacrosse, what matters for success?

A You know, so lacrosse is actually Not dissimilar conceptually, I think, to basketball. It's more complex, right, because there are more players, but it's conceptually very similar to basketball. There are a couple interesting differences, but it boils down to possession and possession efficiency. So how many possessions? Now, the reason it's different than basketball is after scoring a goal, you have a contested possession in the form of a face-off. And, you know, hockey face-offs, but lacrosse is like a, it's a real specialization. There are guys that just take face-offs. So if you have an amazing face-off guy, you essentially make it and take it, right? You score a goal, and then you can get the ball back and just do it over and over.

AI assessment note: “it boils down to possession and possession efficiency. So how many possessions?”

Q an awakening. If that's not how you would describe your investment management tech, request a demo at ridgeline.ai. And now back to the show. Have you spent time either on boards or with the allocator community that I spent a lot of time with? And through that, what are some of the lenses that you use, whether it's base rates, behavioral biases, luck and skill as it applies to allocators?

A You know, the closest thing, Ted, and I don't know if it counts, but the, you know, I have done a fair bit with investment committees, which is probably related. And by the way, some of the work on investment committees has really encouraged me to do a lot of work just on teams in general, but let's call them committees in general. And so I, I think there are three, there are really three things that I've drawn from that literature that animate my thinking about this. The first is the size of a committee or the size of a team. And, you know, there's been a lot of work on this. The main guy that comes to mind is Richard Hackman as a professor at Harvard. And, you know, he studied teams across all different disciplines and found that the optimal tie size was four to six. And there was actually a really interesting survey of investment committees in particular. And they, and basically every person who's on a committee of seven or more says we would be more effective if we were smaller. Or we would not be more effective, but we are bigger, right? So that's a really interesting thought is just how many people are on your committee. And by the way, I'm chairman of the board of the Santa Fe Institute. We probably have 25 or, you know, twenty-five-ish people on our board. Nothing's happening at a board meeting.

AI assessment note: “I have done a fair bit with investment committees, which is probably related.”

Q Are there other variables in that network theory? You can imagine density being sort of an obvious one that may not be as obvious in how something like this spreads.

A I mean, I think the one that I always found surprising is that most of these networks work in clusters. So you have your clusters of friends and I have my clusters and friends. We have overlap in our clusters and so forth. But they found that what was often essential, this is very intuitive if you think about it in terms of how this thing has spread, is there are often these edges, these people that connect different clusters, and they become sort of the central carrier. So even though the clusters themselves may almost be immunized from one another, just one person going from one cluster to another propagates. And by the way, we're talking about a virus, obviously, or some sort of illness. This is also how ideas spread. We talk about this really much in the context of markets. What is an information cascade? It's an idea propagating across a network. Why do we all come to uniform beliefs and markets from time to time? Same basic. It's a mind virus versus a physical virus, but basically it's the same kind of concept. So that to me was always a surprising feature. It's just adding just a few of these sort of linking people from one cluster to another really made the world much smaller than it would otherwise appear to be.

AI assessment note: “the one that I always found surprising is that most of these networks work in clusters.”

Q What does it take to be well positioned to take advantage of?

A I think the big thing is just having access to capital, and that's something that runs through all these themes, right? Because you mentioned briefly this sort of stuff on pro cyclicality is, again, when things are going well, people feel good about the world, and they want to do more of what made them feel good, and When markets have done poorly, they want to do less of what feels bad. And so, yeah, it's access to capital. The question is, how do you, as if you're an investor, how do you make sure that you always have access to capital? Now, you know, you worked for a very famous endowment. I think some of these endowments certainly should be in a position to do that. You think about guys like Warren Buffett, you know, one of the advantages is he's constantly has flowed. He always has money coming in the door and That really helps for you to be able to behave in a way that counters this pro cyclicality.

AI assessment note: “I think the big thing is just having access to capital”

Q the way that someone thinks about it? So if somebody thinks about an idea as just, these are the key Lynchpin issues, does it become easier for them to move on? Because they haven't done all this work to understand all AD issues, they just really understand the three things, and it's It's easier to understand the three things at 10 different companies than 80 things at 10 different companies.

A Super interesting, yeah, and I think, well, there's a thread of research, and I'm sure you've seen this, but it's really interesting, and it also goes to our analyst portfolio manager differences, and the research shows that as you give people additional pieces of information about something, the accuracy of their forecasts don't improve at all, or their bets don't improve at all, but their confidence tends to soar. And, you know, part of that is the original study was done back in the 19 seventies, and it was with handicappers. So if you're only allowed to get whatever, five or 10 bits of information about a horse, you're going to probably pick the stuff that's most important first, right? So, and I think, you know, handicapped, that might be relatively self-evident. So I think the key task here is to figure out what matters first, and I think that's where PMs, so they're dwelling on that left-hand side of that graph, which is Not that much information is accurate is a lot more fresh and then not so confident, which means that the view is sort of lightly held and more malleable when confronted with new evidence. And so I think that is probably the sweet spot on this thing. And I think as an analyst, as you gather all this more information, you tend to think, you know, more than you become more confident in your scenario. And that may be a mechanism that locks you in more than …

AI assessment note: “view is sort of lightly held and more malleable when confronted with new evidence”

Q So I want to start on that cashflow side. What does it look like just a back of the envelope if you do cut out those first years of cash flows? And I imagine there's a wide range, a normalized discount rate. There's some range of a market like a GDP like grower. How much of the present value gets chopped out

A Somewhere between it can be 10 to 20%, but that's sort of the high end. And, you know, the other thing is weird, Ted, is that when discount rates come down, that means actually future cash flows are more valuable today than they were when discount rates were higher. Also means expected returns are lower, but that's an interesting sort of trade off. So yeah, it's the drawdown we've seen. You could argue that we had a very good 2019 equity markets. There are a lot of people have been concerned going into this whole thing, concerned about economic growth and so forth. So that would be the other argument you would make is, gee, we were overextended, and this part of this correction has been a function of just flushing out some of that excess. I'm not so much in that camp. I mean, I think if you looked at the kinds of things that I like to look at, like shareholder yield or free cash flow yields, those numbers, especially given prevailing risk-free rates, those numbers were sort of in line with or even attractive relative to history. And I do think there's some accounting stuff with companies and tangible investments versus fixed asset investments and so forth that make it a little trickier to compare things like tape ratios, Apple to Apple over time. So again, we don't know. I think you and I would agree. We have no idea what's going to happen the next week, month, few months going…

AI assessment note: “Somewhere between it can be 10 to 20%, but that's sort of the high end.”

Q Let's start on the behavioral side. What is it about behavior that drives someone's ability to have an edge?

A Just to clarify a specific point, a lot of what we talk about in the behavioral stuff are things like the heuristics and biases, you know, so these are the mistakes you and I make as people. So we're overconfident and we anchor and how things are framed affect how we decide and so forth. The point I try to stress is that it's not clear to me that those little things actually manifest in markets, right? Because They can cancel out, essentially. So the market is really, you're not competing. It's not me, Michael versus Ted. It's really Ted versus the market, and that's a different dynamic. So you're competing with this thing, a complex system called the market. So as a consequence, really the level you want to think about is how do people behave in groups, in group settings? How do they follow one another? So there are two or three things that I highlighted. The first is something that we see everywhere. There are probabilistic decisions being made, and that's over extrapolation. So people take the recent past and they think the recent past will continue going forward. And there's some very nice work on this. By the way, Andre Schleifer has a new book out where they talk about this. And basically here's a simple example. If the markets have been doing well, people think the markets will continue to do well. They've done poorly. They'll continue to do poorly. And so over extrapola…

AI assessment note: “really the level you want to think about is how do people behave in groups”

Q What, what does it come to? I know the numbers are really small.

A It adds well, you know, and over, so if you look at, if you just look unweighted numbers, we have these data back to probably the mid sixties, but unweighted numbers, about 40% of managers beat the market in an average year. And the standard deviation is high. It's like 17% standard deviation. So if you want to say what's the probability that this fund will beat it, it's a 40% with a 17 standard deviation. If you look at it on an asset weighted basis, the numbers go, they're not quite 50, but they get into the mid to high So it's a temp called a 10 or 15% uplift on that percentage basis. And that's not inconsequential. And of course, the gross profit thing is all pre fee. And that's a very important thing to bear and bear in mind. So if you are extracting value from the market, then the question becomes what's a fair and quotes fee allocation between the manager and the, and the client and so forth.

AI assessment note: “about 40% of managers beat the market in an average year”

Q On that point of experts, there's all kinds of research that shows, particularly in markets and complex adaptive systems, that experts don't do very well in their predictions. So just as that starting point, before we dive into when that works in pattern recognition, how do you decide upfront what constitutes an expert and not just someone with experience?

A One of the ways to think about this, and this actually frames pretty much the whole report as well, is that you're likely to see expertise work or intuition work, candidly, is when you have environments that are relatively stable, where the relationships are linear, so cause and effect are quite clear. The canonical example of this course is chess, which has been studied by cognitive psychologists for a very long time. If you show a chess master, the chess board, a game in progress, they are going to very quickly know which player has the advantage. They're going to very quickly look at either side of the board and tell you either an optimal or close to optimal move. And if you had three grandmasters lined up, they would pretty much tell you the same thing or something very similar. Those conditions are all very structured, stable, and linear. Now, if you came out and said, hey, instead of eight by eight, the board's going to be 12 by 12, and instead of the bishop moving this way or the queen moving that way, we're going to change how they all move, that chess expertise will go out the window. They're essentially back to square one. So what they've been able to do, chess masters, is they're essentially have learned deliberate practice. They've learned about this system in such a way that allows them to conceptualize the whole thing in a way that's very effective. Now, as you po…

AI assessment note: “expertise work or intuition work, candidly, is when you have environments that are relatively stable”

Q And we've seen, you know, if you go across markets, right, there are certain assets like distressed debt, where people have raised commitment funds. Private equity is probably the greatest example of you have multiple years. Have you thought about any ways to execute on that in the public markets?

A No, I don't know a way to do that, and I've been thinking a lot about this topic of public versus private markets, and there are certain Both of those markets have their pros and cons, right? But that may be something that may be a check in the, a benefit for private markets that allows you to have these mechanisms in place. Now, I guess you could do something similar just to say public marketing, we're invested in some public vehicle, and if we have, again, preordained, whether it's some magnitude of some drawdown, or some valuation levels, or some combination of these things, that we'd go back to our pre-commitment contracts, but I'm not aware of anything. Like you said, you, you do see it much more in private stuff than you do.

AI assessment note: “No, I don't know a way to do that”

Q And is that, did Taylor come at that from that kind of principal agent lens?

A He didn't so much. Well, that came up in the conversation, which gave me the idea to try to really be more specific about it. One book I read recently, which I really enjoyed, it's called The MVP Machine, about player development in Major League Baseball, and so I was asking one of my football executive friends what's going on in the NFL, and he just said baseball's a really kind of a weird thing, because In a sense, we have control of our player for five to seven years. In fact, if they're good, we're playing them way below market for the last few years that they're in the organization. So it makes enormous amounts of sense to sort of get as much value from those players while they're around, and then you can let them, you sort of offload them and let someone else pay for essentially their economics. So in other sports like football, I think to a lesser degree, to some degree basketball as well, less of an incentive because they're just not as part of the organization. So it's this interesting balance because you want the players to be as good as they can be, but there's obviously a cost. So how do you balance those two things? So that's another really interesting sort of goes back to these time things, these time horizon things.

AI assessment note: “He didn't so much. Well, that came up in the conversation”

Q What are some of the ways you've seen where people use the label of pattern recognition or intuition, but the research would show you that there's no signal in that?

A I think this is very common. That's why I opened with a story. I was a junior analyst working for a senior analyst and came in and basically made a decision about a recommendation of a stock based on no analysis. But he knew a couple of things that were true for the most part historically, and these are imprinted on his own experience. And hence he made this decision. So just to be clear, we all have these sensations. Now, the other thing that's really challenging about this basic concept is we remember when pattern recognition led us to a good outcome, even if it was luck, and we forget about when it led to a bad outcome, even if it was bad skill. That's the other thing is we're very selective as to when we remember when it works and when we remember when it doesn't. We all have these sensations when we walk around about these patterns. The whole goal of this report is not to say this doesn't apply, it does, but just be careful about applying it too broadly or too boldly.

AI assessment note: “made a decision about a recommendation of a stock based on no analysis”

Q Yeah. What information do you read of the vast amounts that you do that you think others might not know about but would benefit from?

A Well, I don't know if it's, there's information's an interesting word in and of itself. I, for better, for worse, and many days I think it's for worse, but for better, for worse, I probably spend a lot of time thinking about more frameworks than I do, like mental models than I do nuts and bolts. And the reason I'm a big mental model fan is I do think that when ideas are thrown at you, if you have a framework, a lattice work to hang it on, you're going to be much more effective. One example I get, I mean, this is like a, this is a pretty nuts and bolts example, but You know, for example, we have a framework for thinking about mergers and acquisitions and how to evaluate the quality of an M and a deal. And I just found as an analyst over the years and a strategist and so forth that analysts, almost every deal, it's like a one-off right there. They're analyzing it without a framework. And if you have a framework for it, it just allows you to put things into context, analyze it quicker, to be more accurate and so forth. So probably is that it's less the what information, but more like I'm, I'm very committed For better, for worse, some days, like I said, for worse, to mental models and thinking about big mental models. By the way, I'm a Buffett fan. I know you're a Buffett fan. I, I, I've learned probably more from Munger, or I've taken, I've probably taken more from Charlie Munger…

AI assessment note: “probably is that it's less the what information, but more like I'm, I'm very committed”

Q How do you decide when you're going to read through the whole book?

A Yeah, I almost always read the whole book, for better or for worse. I have a large, for every book I read, I buy many more books, so I have a huge library of things I've not read, and I will typically flip through them and try to familiarize myself at least with that. So reading, but I would say probably actually a sports, and so I'm a big sports fan. I love watching almost any sport on TV, and I still have got a bad knee, but the one thing I'm allowed to do is play ice hockey, so I still play ice hockey in a winter club, and so that's good fun, and the other thing I'll say about that, besides it's good exercise, and it's a great sport, there's a lot for the camaraderie in, in these locker rooms, because hockey is, it happens to be a weird sport where you spend 20 minutes before, you play, and 20 minutes after, sort of changing and unchanging, And there's also a lot of folks that play that are not in the world of finance or business, which is great. So it's like getting access to different walks of life. So yeah, I would probably say reading and sporting, I like love to travel and all that kind of stuff, but those would probably be the big ones.

AI assessment note: “I almost always read the whole book, for better or for worse.”