The Exchanges

Q Are your clients, you, you mentioned 86%, is that primarily long short equity managers? And what, what's in the other 14%, and where else could you apply this tool?

A Long only managers. It's just that our background is, I came out of a hedge fund, a lot of my friends worked for hedge funds, and so our initials were clientele, were those folks, and it's a network effect. They tell other people, and they happen to be hedge fund managers, but I actually think that this is, what we're doing a lot of times is more applicable to a long-only manager, because they end up having a larger stable of analysts working for them, and they have less of a conversation sometimes between them. Just imagine, uh, Neuberger, for example. They have a large number of analysts, and they have portfolio managers that don't directly control those analysts, unlike, uh, Hedge fund. And so the information flow and making things explicit is so much more important in that process. And they end up having a lot more rules as well. And so there's a, there's a good fit for that. It's just haven't had as many conversations with mutual fund managers.

AI assessment note: “Long only managers. It's just that our background is, I came out of a hedge fund”

Q If you turn the Table a little bit, and think of this from an allocator's perspective. What questions should allocators be asking of managers they're interviewing that would indicate whether they're navigating this well or not?

A You were an allocator for a long time, and you asked a lot of these questions, and the question is, how do you size positions? Every allocator asks that question, and portfolio managers are well trained to say that what we do is we're constantly looking at risk reward, we're, we're evaluating that, we're making sure that our best ideas are our biggest positions, we want to evaluate management teams, and constantly looking at risk and, and other portfolio factors, and And trying to make sure that we're not in too crowded abets and looking at exposure analysis. And they just, when you, when you walk through it that way, it sounds great. But how do you do it? And really the question that an allocator could ask is say, hey, what's your sixth best idea? And pause. And wait for them to tell you what their sixth best idea is. Now, if they can turn to a sheet, or a dashboard, or something like that, that says, that's my sixth best idea, adjusted for all the things that I care about, which are going to be risk-reward, drawdown risk, liquidity, all those kind of things, I can look down my list and say, this is my sixth best idea right now, because things change. If they can tell you that, they can manage their portfolio effectively, because they can make sure that their best ideas are their biggest positions, their 16th best idea is their 16th largest position. But if they can't quickly …

AI assessment note: “the question that an allocator could ask is say, hey, what's your sixth best idea?”

Q Yeah, it's a tricky one because it's easy to embrace the notion that trading creates transactions costs, and therefore people should trade less, and this is just Movements around stock prices on a path. What other research have you been able to do based on the data that you're receiving from doing this a long time, and what has that shown you?

A One of the things that we did is we wanted to see if our clients that used our system more frequently were better performers. One of the simple things that we did was just say, alright, Let's carve out the positions of all of our clients and put them into two buckets. One bucket's gonna be positions where they took the time to come up with a price target. The other bucket's gonna be positions they didn't take the time to come up with a price target. Well, guess what? The ones with price targets outperformed the ones without price targets. And it was a huge margin. It's basically seven percent was the return on invested capital for positions with price targets. One percent was the return on invested capital for positions without price targets.

AI assessment note: “The ones with price targets outperformed the ones without price targets.”

Q And again, that's, that's, those are alpha percentages?

A No, it's just pure return on invested capital. Yeah. It's such a simple rule that a firm can implement where we have enough Data that should say, if you don't take the time to put in a price target, don't put the position on. Then we went to and said, okay, well, let's think about that a little bit further. We have a little bit of other data for our clients where we can break them down into the people, like I mentioned before, most correlated. So how closely do they follow the model? What's the percentage of coverage of price targets in their portfolio and how fresh are those price targets? So if they haven't updated in 90 days, it starts to get To the point where we called it, ah, you know, stale. What percentage of their portfolio has been updated in the last 90 days? So we take those three metrics. Correlation, coverage, and freshness. And we scored them equally and quartiled our clients and said the best actors, second best actors, third and worst. And then we just did return and invested capital for each of those. And it was I'm going to get the numbers wrong, but it's going to be directionally accurate. The top was around nine percent return on invested capital. The second was like seven. The next was like three. The bottom was like zero. And there is a distinct correlation, at least in our small sample size. Once again, keeping in mind the small sample size, a distinct c…

AI assessment note: “No, it's just pure return on invested capital.”

Q And so let's dive in just a little bit there. So what's implicit and what's explicit in that definition?

A Yeah. So let me give you an example for investors. So imagine I'm a analyst walking to my portfolio manager's office and, and we're talking about investing in a security. And I believe that, uh, I tell him, you know, I think the stock has a lot of upside. It's got a great management team, not a lot of downside. He's kind of a liquid. So we have to be aware of that, you know, Hey, portfolio manager, try and size that position. There's a lot of implicit information that I've given him. And so if I could translate implicit information into explicit assumptions, Then we can be better off. So if I, instead of saying I have a lot of upside, I'd say I have 60% of upside, or I say I don't have a lot of downside. I say 20% of downside because my portfolio manager could have interpreted a lot of upside as 120% or 30%. There's a great quote by Richard Schuer who wrote, uh, The Psychology of Intelligence Analysis where he says that objectivity is gained by making assumptions explicit so that they can be examined And challenged. It's one of my favorite quotes because it is the essence of what we try and do. We try and help firms take a lot of the bias and emotion of the decision-making process out so that they can make better decisions.

AI assessment note: “translate implicit information into explicit assumptions, Then we can be better off.”

Q have a bias towards being right in their probability assessment, and can adjust that, they'd have a less risky portfolio. You've also written a piece called the Concentration Manifesto. That almost argues for the opposite of that in some sense, uh, depending on how you define risk in the portfolio. So why don't you talk about what your research has shown you and what the concentration manifesto is all about?

A Yeah. So at a high level, what the concentration manifesto is, is that active management is under immense pressure. And a big part of that is self-induced because either fees are too high to get over the hurdle of the performance that they generate. And, you know, they do generate alpha. It's just, The alpha's not high enough to get over the fees. It's also self-induced because we've been told by the people giving us money, either an institutional allocator or retail allocator, that we need to be diversified. We need to control risk as a portfolio manager. But once again, it goes back to what I was saying before. We're trained to be great analysts, and generally we can pick good stocks. So what we've found, going back to that example that I gave before, is that if you look at A fund's top 10 positions. Those, for our clients at least, and a lot of other academic studies, I know that Novus, um, did some research here as well. The top 10 positions outperformed the next 10, outperformed the next 10, and so for our clients, it ended up being, you know, I said the batting average for our clients was right around 51%. This is on an alpha adjusted basis, so the top 10 positions was 57%, next 10 was 55, next 10 Was 52, and then it went into, right after 30, it sort of went into randomness. Basically, it was a coin flip if you were picking a good stock or not. And so, my premise is this…

AI assessment note: “what the concentration manifesto is, is that active management is under immense pressure.”

Q How much does the optimization of the inputs of the tool reacting to the just dynamism of markets or stock price movements end up driving kind of the efficiency of portfolio construction? So you have your assumptions. The assumptions might not change, but stock prices are moving, and Managers trade around a little bit every day, and the question is, are they making the right trades?

A We find managers should be trading around positions more. They're generally not, and the positions that's easiest to ignore is the one that's working and making you money, and so what happens in that situation is the price is going up, which means the position size is going up as a, as a percentage of your total assets under management, and so let's say a five percent position has now grown to a seven percent position, but what also happens to the expected return in that case? The expected return goes the opposite direction, all else being equal. Let's just say that our Price target was 100. The stock's gone from 50 to 70. So the expected upside has gone down. The expected downside has gone up. And clearly something's changed. Maybe the probabilities of upside or downside, but clearly you can reset those expectations, but the expected return in some ways has gone down while our position size is going up. Managers in general should be trading more because there's a big reversion to the mean factor in the market.

AI assessment note: “We find managers should be trading around positions more. They're generally not,”

Q What have you found? Changes in the sort of portfolio manager's implementation once they're using alpha theory compared to when they aren't.

A It runs the gambit. So we will walk in to a new fund that we start working with, and they already have a lot of this philosophy embedded in their, their mentality. So codifying that is relatively straightforward. Then we'll have others where a lot of this is greenfield for them. And so we're asking questions that As a firm, they know they should have been asking, but they haven't been, and so a lot of it is the big gains that they get from starting to use Alpha Theory is the, the initial phase where they try and come up with a similar vernacular and standardization for how they make decisions, and a lot of those questions are really enlightening for them, and so for both firms, once they get to the point where they have a process in place, it starts to become refinement over time. There's an evolution. Just like I said before, people like to override the system. Because we have empirical evidence that proves out if it works or not, we can then get them more comfortable with making decisions using the model they built in advance instead of overriding it.

AI assessment note: “get them more comfortable with making decisions using the model they built in advance”

Q Right. So when you take that framework and apply it to portfolio construction, what is the tool that Alpha Theory is providing to portfolio managers?

A Largely what we're doing is we're a toolkit. We're, we're trying to allow portfolio managers to do what they're doing already, but trying to take those implicit assumptions, We capture a lot of those implicit assumptions and allow them to make them explicit inside of our system. So those things can be something as simple as how much do I believe I can make if I'm right? How much I could lose if I'm wrong? What are the probabilities of those things occurring? And I think that your last couple of podcasts, you had a Mobison on, you had Annie Duke on. One of the things that they found of paramount importance and they kept coming back to was expected value or coming up with ways of using probabilistic expectations to come up with The decisions that people should make, and so what we do is we want to capture that and make the implicit explicit, and then allow them to find other things that they deem important. It can be a checklist of items, so how good is the management, what's in the rate, how good the management team is, or how good is the balance sheet, and for each of our clients it's different, but taking those implicit assumptions, making them explicit, and then allowing a portfolio manager to set rules For how they want to translate explicit information into decisions. That decision inside of a portfolio is going to be, what's a position size? A zero percent position size is…

AI assessment note: “taking assumptions that are implicit, making them explicit inside of our system, setting rules”

Q What's been the biggest challenge in running this business?

A Changing human behavior. I don't think that we go into too many meetings where people sit across the table from us and they say, this thing that you're proposing that we do is a bad idea and not something we should do. Almost uniformly, they all agree that this is something that they should probably do. With Alpha Theory or not, they should definitely create some more process around how they make decisions in general. But I sort of equate it to a gym membership. Everyone knows that going to the gym or working out or whatever you want to, whatever way you want to analogize it, it is important to our health and our well-being and our happiness and all of these things, and we have three different groups of people. There's the people that say, yep, that's a great point. I'm going to sign up for your gym, and they're going to go. Great. Those are the people we're looking for. Then there's going to be the other people that are aspirational, and they're going to say, yeah, that's a great idea. I'm going to sign up for your gym. They go. They try it out. Doesn't quite stick. It's behavior. They stop going, and they cancel their membership. And then there's the third group of people that just know up front, I'm not gonna do this. And even though they know they should, they still don't. Most of us overeat at Thanksgiving and Christmas. Why is that? We know we shouldn't, but we still do i…

AI assessment note: “Changing human behavior.”

Q Have you had any conversations with credit managers and, and distressed debt managers?

A Yeah. I mean, this is, this is decision-making one-on-one. It's what we're doing. You could apply to, uh, trying to buy a house or, you know, should I take a new job? This is all applicable. And yes, we have had some high yield stress guys that have said this is something that we're working with a few right now to vet that idea, but it's the same. You're still coming up with, you know, the upside may be par instead of something. Actually, in some ways, it's better because your expectations are a lot more discreet. You know, you kind of know what your upside is. You kind of know what your downside is, and the probabilities are still a little fuzzy, but you know, the assumptions going in are a lot better.

AI assessment note: “yes, we have had some high yield stress guys that have said this is something”

Q And what are the mistakes that you see portfolio managers make?

A Step one is, is really that sort of implicit, explicit translation. Then, um, setting rules around those things. So if I, um, Set rules for how I want to size a position, how I want to think about liquidity, how I want to think about risk reward. I need them to be explicit so that I can actually follow them. But even if they are explicit, what we find is a lot of times people override the rules they've set in advance. And I think that there's enough cognitive science studies, psychological studies that show that experts really aren't good at making decisions in general. Political pundits on who's going to be elected or Wine experts on, uh, what's a great vintage, or an oncologist on who is most likely to get cancer, and so these are people that are experts that spent their whole life studying these things, and so as an example, let's say the oncologist study. The oncologist were asked to tell the psychologist, what's important to you when you try to determine how long someone's going to live with cancer, and they will say, you know what I care about, I care about, you know, if I'm looking at the x-ray, I want to see the, The tumor, the perimeter, how far it's spread, size, the age of the patient, T cell count, and I'm not a doctor, so I can't give you the litany of things that they care about, but they're going to give you this litany of things that they care about, and what th…

AI assessment note: “what we find is a lot of times people override the rules they've set”