In early May 2019, I did a series of interviews with Clifford Sosin of CAS Investment Partners in Westport. In this part of the interview, we talked about idea generation in investing. You can find all content with Cliff here.
- 1We got to know each other
- 2Since when have you launched your investment partnership?
- 3What did you do before your own partnership?
- 4We talk about idea generation or the organic process
- 5We chat about mental models and the application of them
- 6He explains the reason why he passes an idea
- 7We get to know what he likes about businesses
- 8What talk about “sensible people” and good management.
- 9Well- reasoned and sensible business answers to business questions: What does that mean?
We got to know each other
[00:00:09] Tilman Versch: Hello Cliff, welcome to our YouTube channel. It’s nice to meet you here in Westport Connecticut, on our way to Omaha. Who are you?
[00:00:16] Clifford Sosin: My name is Cliff Sosin, I’m the Founder and Manager of CAS Investment Partners.
Since when have you launched your investment partnership?
[00:00:23] Tilman Versch: Since when have you launched your investment partnership?
[00:00:26] Clifford Sosin: I launched CAS Investment Partners in… I started putting it together in July of 2012, but we actually started operating on October 9th of 2012.
What did you do before your own partnership?
[00:00:42] Tilman Versch: That’s nice. What did you do before that? Before your own partnership?
[00:00:45] Clifford Sosin: Prior to that, I spent five years at EBS, where I worked in something called the Fundamental Investment Group, which was essentially an equity long-short investing business within the EBS Investment Bank.
We talk about idea generation or the organic process
[00:00:59] Tilman Versch: During that time, you developed your own philosophy, and we want to take this interview at the chance to do a deep dive into this philosophy. My idea is to start with the process, like from ideas generation to buy or sell a certain stock. So, my first question is, where do you get your ideas from?
[00:01:25] Clifford Sosin: You know, it’s funny. I think there’s a desire amongst people to have a simply explained regimented process. It feels like turning a crank and result in good stock ideas that make a lot of money. My impression is that the way it really works is far more organic than that, especially as sort of the top of the funnel. So, the way that I start thinking about a particular idea, is essentially random. Organic is a nice word, random is true. As I develop an idea, as I look at it initially, what I’m trying to do is draw from, sort of a set, my understanding of the world, which is really you can think of a set of mental models that explain parts of the world, and apply it to that particular circumstance and see how well it fits. And if it’s, well, you start to develop a working hypothesis about something.
So, the way that I start thinking about a particular idea, is essentially random. Organic is a nice word, random is true.
And so, what I’m describing is maybe a process where you start, almost start randomly at the top of the funnel, and then you become more regimented as he tries to apply these mental models. And then you work your way down into, once you have a hypothesis, it really becomes more a process of attempting to invalidate your hypothesis, analogous to the scientific method or whatnot. And so, that would be a process where you think you understand the dynamics that are at play, that allow a company to succeed. And then you are thinking about what you should be able to observe in the world, in the ecosystem that would be consistent with that hypothesis. And you then go into the world or the ecosystem, you do due diligence and you try to figure out whether or not what you’re seeing is actually consistent or inconsistent with your hypothesis. And if it’s inconsistent, then you would [more or less] change your hypothesis or throw it all together. If it’s consistent, then you aren’t sure, right, but you’re incrementally more confident in being right. And if you can develop a theory for why a company is successful based on mental models that are well-founded and reasonably reliable, and then you can make a lot of predictions based on that, you start to develop a pretty reasonable chance at correctly forecasting the future of a business within a narrow enough margin that you can then make a good bet.
And so then, the last piece really is that you sort of comparing what you think might happen based on your understanding with what’s implied by the price or what you need to have happened to have an investment be profitable. Then that’s it. I mean, it sounds so simple, but just that’s what we do.
[00:04:01] Tilman Versch: But there’s the first step where you get to hear first about a certain name…. Maybe.
[00:04:04] Clifford Sosin: Yeah. I mean, that’s totally random. You know, I have invested in all manner of situations. I’ve run screens, I wouldn’t say running screens versus any better ideas or worse ideas. I’ve ended up looking at things for one reason or another and passing on them and coming back to them years later. I’ve had ideas that smart friends found and recommended. The key on sort of the ideas front… I mean, I don’t have a way to understand whether an idea is a good investment prior to understanding it to any important degree. So, in terms of what to look at next, the process is essentially random.
There are a few things that I like. We’re looking for very few, very extraordinary investments. So you tend not to get very extraordinary investments in situations where, instead of normal times. In situations where the company looks like a lot of other companies. You’re unlikely…. you know, not saying it can’t happen, it’s just less likely. So a unique business, it’s probably going to be more interesting to me than a not unique business.
There are a few things that I like. We’re looking for very few, very extraordinary investments. So you tend not to get very extraordinary investments in situations where, instead of normal times. In situations where the company looks like a lot of other companies. You’re unlikely…. you know, not saying it can’t happen, it’s just less likely. So a unique business, it’s probably going to be more interesting to me than a not unique business. And trust another bank, versus something else. But businesses can be unique in several ways. I mean, just another bank, might seem like just another bank but it may have some attributes over there that are quite unique.
And so, anyway, something unique is a helpful first step. But really, I would say, what really happened is I start looking at it and I start to make progress in that, as I start to understand what the business is doing, what the people… you know, from what management says, and when they describe the business and in public or whatever, and in the public materials. It starts to indicate to me various mental models that can apply to the situation. I start to think that, “Oh that makes sense. Maybe what’s happening is this. I’ve seen something like this before”. And that’s where you start to decide that something might be interesting.
We chat about mental models and the application of them
[00:06:05] Tilman Versch: Can you name me some examples for those mental models?
[00:06:10] Clifford Sosin: Yeah. Well, sure. You mean the mental models or the application? You mean, the mental models.
[00:06:16] Tilman Versch: Both is fine.
[00:06:18] Clifford Sosin: Yeah, well, I mean, a simple one that… I’ll give you two, and they’re both related to Carvana, which I think we’ll talk later. But you know, a simple one is that one thing that’s really important for Carvana over time, and then the benefits from, is what I call ‘economies of trust’. And this is something that everyone says no, but the idea being that if you can become a trusted counterpart in a system, there’s just an enormous amount of value that’s unlocked for everybody. And Carvana, they are competing with car dealerships that are by and large not trusted. And they are earning people’s trust in this industry, and that produces two big benefits.
The first is it makes it easier for people to do business with you, so they do more business with you. Then the other is it’s a huge competitive bench because people… why would anyone do business with someone they don’t trust, when they can do business with someone they do trust? And it’s very hard to duplicate, because trust is earned over a long period of time. And so, that would be a simple one.
A more nuanced, complicated mathematical one, is that if you look at industries where people sell different goods, unique goods, and there’s a small number competitors. There is some game theory that describes how the competitors will price those goods. And what I’m referring to here, is what would be called by microeconomics, a single state hoteling game. So here, I mean, the classic example of our a hoteling game, is you have two ice cream vendors on a boardwalk and people pick ice cream vendors based on the price that they charge. And the question is, what price should an ice cream vendor charge, given where the other ice cream vendor is? And it turns out that, there’s…. You don’t actually just charge…. In a perfectly competitive market, you would just charge marginal cost, there would be no profits. But if you imagine two vendors want 33, the other one 66 on a 100 unit long boardwalk, the vendor that’s 33 or 66 has… If both charge marginal costs, they split the market equally, but neither makes any money. So both vendors have an incentive to raise their price somewhat and they’ll give up some market share if the other guy doesn’t raise price, but they’ll make more on the market share they get. And so they’ll make more money doing that, independent of the other person’s decision. And since they both actually make that decision at the same time, what happens is they both raise price a bit, and they actually sort of re-divide the market share more evenly, and then you keep doing this. And you end up getting into an equilibrium, where there are economic profits for the competitors.
That’s a single-stage game. And you see that all over the place, not think that’s geographic space, but if we think about brand space, you have two types of pizza that are different, that have different preferences. Or in the case of used cars, I have a unique used car. It happens to be a particular color and style and trim and everything else. And you have a unique used car, but they are different and there are different… there are consumers who would be willing, at the same price, who would prefer mine over yours. And as they change, increase the price, the differential between the two. There’s an elasticity of demand effect, but that means you’re playing a hoteling game. By the way, the multi-stage hoteling game, is if you could locate your ice cream vendor, where would you put it? And the funny outcome there is that both of them end up putting it right in the middle and they make no money. So that’s why if you’re Ford, you make a car that’s just like Toyota and you try to make them as similar as possible.
So, the second mental model though, in the case of, that I’m pointing out, is also applicable to Carvana, because it deals with the possibility that there is a competitor right now, that I think there’s a reasonable chance there won’t be a big viable competitor. But if there is, they’re not… even if they’re competing in a non-cooperative way, the non-cooperative equilibrium should be one where both of them make a lot of money per unit and it’s a very big market. And so that’s why it’s another good reason that I like the business.
So, two very different mental models. One drew, that you could find in a microeconomics textbook that has like math, and we could keep going. And the other that’s squishier, that you would read about like a Buffet letter or something, where it’s like, okay, if people trust each other, a lot of good things happen and being part of an access of trust or better control in access of trust, is really, really valuable. And you know, that isn’t… I don’t think you find that that often in microeconomics seminars, but they’re both true. So I try to draw them all and then apply them to the situations to understand them.
He explains the reason why he passes an idea
[00:10:57] Tilman Versch: What makes you passed an idea? You said before, that you sometimes pass an idea and come back years later.
[00:11:03] Clifford Sosin: Well, I mean, there are lots of reasons why I pass on ideas. The most common reason is as a reasonably comprehensible business and it’s priced about right. You know, the world is competitive. And that obviously happens most of the time. Another reason is that I either don’t have the mental models to make good predictions about the business over time. Or I do, and those predictions are that it’s actually very hard to predict.
So, I think of fashion as sort of intrinsically unstable, because when people…. It’s not necessarily always the case, but you can think about it, there are certain elements of fashion where it is important to me to wear something different. People buy fashionable items to signal that they have money and taste. And it’s the taste element that’s complicated, they complicate things because it can be the case that buying something different than everybody else shows that you have taste. And in that sense, the thing that used to be a way to signal to have taste can actually become a way of signaling that you don’t have taste anymore because you’re not buying the new thing. And you know, this isn’t clearly true in every fashion category, and I don’t actually know all the answers there, but it’s one of those things where you can look at businesses where there’s a lot of trend risk. And I find them to be fairly difficult to predict. And maybe I just don’t have the tools. Maybe someone else has a mental model to understand it and I don’t have it. Or maybe it can be developed. Or maybe it’s one of these things, you know, the thing about quantum mechanics, but there are some things that are just not knowable. And so, that’s one.
There’s a book I read that described an experiment that someone did, which I thought was great. It was before the days of downloading music. And these researchers, what they did, was they got a hundred unique pieces of music by artists. And they then got teenagers and they created a web… so novel. They created a website and they allowed the teenagers to download the music and listen to the website and also to comment on the website about the music, and to rank the music. And they then had the teenagers engage in this, and what happened over time, was there was a ranking of popularity of the music in terms of frequency of download and review and whatever else.
What they didn’t tell the teenagers, was that they actually divided them into separate, distinct sets of worlds. So you would be originally randomly assigned to one world or the other or the other, the other. And they all started with the exact same set of music. But what happened was, as those worlds evolved, the music that rose to the top and the music that fell to the bottom were actually totally different. And there was a very weak correlation. Like if you were the best in one universe, so to speak, you were probably slightly higher ranked on average in the other universes, but it was a very weak effect. And the dominant effect was chance. And so, if you think about how certain works of art have become… Why the Mona Lisa is the Mona Lisa and other works that are just works of art. There is probably a meaningful, kind of self-reinforcing chance-based phenomenon that’s happening. And my point on that is that there are just some things that are intrinsically unpredictable. And so sometimes you either know some things are intrinsically unpredictable or you just don’t know how to think about it and, or frankly, it’s just priced about right.
We get to know what he likes about businesses
[00:14:41] Tilman Versch: You already mentioned two criteria you like on businesses. Like predictability and price. What other criterion makes you like a business?
I’m looking for businesses that are going to do very well for a very long time, that I can get at a price that doesn’t reflect that, that are run by sensible people.
[00:14:51] Clifford Sosin: I’m looking for businesses that are going to do very well for a very long time, that I can get at a price that doesn’t reflect that, that are run by sensible people. I don’t think there’s anything more to it than that.
Predictability is an interesting term. It depends on what you mean by predictability. The ability to predict the performance of a business every day, every week, every month, every year, can be nice. It can simplify your life if you’re analyzing a company. But there are plenty of businesses that can be great businesses that maybe are predictable over an average over 10 years, but maybe not predictable every day, because maybe they serve a volatile end market.
And think of someone who’s trying to value a company that sells parts into the new car space. It may very well be that like you, for some reason, technological, manufacturing scale, whatever it is, are very likely to make $10 a car no matter what, for as long as cars exist. But every year the number of cars made might vary substantially, so it might be very hard to predict how much money they can make every year. But a sensible guess of the number of cars made over the next 10 years is easy enough to do. And so, the predictability…. You can take predictability too far, but predictable… but the key there, would be the predictability of the 10 $ per car. That would be the bid; that would be what you’d be focused on. So, it depends on what I try to think about.
What talk about “sensible people” and good management.
[00:16:30] Tilman Versch: You mentioned the sensible people. What importance does management play for you?
[00:16:38] Clifford Sosin: This has been there, has evolved a lot over time. When I started, I probably would have told you, look, I generally screen for bad management. You know, but my general sense is that managements on average, roughly average, and I can find good businesses at cheap prices and you know, with average management it will be fine.
Over time, I have started to develop some real perspective on that. It’s arbitrary whether or not this is right, but in general, there’s a lot of ways to think about this. But the way I’ve tried to put it is, I want management teams who provide well-reasoned and sensible business answers to business questions.
I want mangement teams who provide well-reasoned and sensible business answers to business questions.
Well- reasoned and sensible business answers to business questions: What does that mean?
[00:17:29] Tilman Versch: What does that mean?
[00:17:31] Clifford Sosin: Why did you decide to invest in this market versus this market? Why did you decide to buy this supplier? Why did you do…? Ask why they made their decisions. And then listen to the answers. And if the why is a reason that makes a lot of sense from the perspective of someone who was going to own this business for a hundred years and is trying to maximize his profits. Then, it’s well- reasoned and they’re thinking from kind of the right principles, then that’s probably indicative of someone making good decisions, running the company.
If the why is sort of poorly thought out or not thought out or driven by factors that aren’t consistent with what you want, then that’s a problem. Like, why? Well, I thought shareholders would value a higher growth, lower CapEx Company. Like, really? That’s a terrible reason.
And it’s important to not judge by the outcome, so, why did you make that terrible decision that ripped up a lot of money? Well, you know, here was our thinking about it, here is what I thought was a reasonable set of outcomes, here’s what I thought. These were going to be better outcomes and then we did it. We learned this and it was a disaster. You know, that can be a great answer and that can be indicative of a great management team. So, I’d say that’s one piece; well-reasoned, sensible business decisions, and it’s also kind of the measure of intelligence. I mean, if they’re just making dumb decisions, that’ll come through.
And then, there’s an element… I think management teams are a product of the culture they’re brought up in, and, or brought to. Or whatever. They’re not totally… It’s not as though someone sits on top of the company and like makes the company what it is. It’s much more organic than that. Companies are living things, with a lot of relationships between people.
[00:19:24] Tilman Versch: Companies shape management, management shape the company.
[00:19:26] Clifford Sosin: That’s right. There’s a relationship between them. That’s right. A couple of observations here.
One, the right management depends on the situation. Think about companies as a little society, they all have different norms and values and expectations.
One, the right management depends on the situation. Think about companies as a little society, they all have different norms and values, and expectations. And the way you signal importance and relevance at Google, versus the way you signal importance and relevance at Procter and Gamble, is probably very different. And so the types of people and the behaviors that they undertake in order to be successful, vary a lot. And so the type of leadership that they’re going to want and need will vary a lot. But in general, I think you can try to sort of just ask yourself simply whether or not you want to work for this person and whether this person would inspire you to do your best. And whether you get a sense that people, they’re…. The term I’ve sort of come around to, is you want to invest in firms where the people feel energized. You know, I think energized, is like when you look at organizations that really accomplish a lot.
You want to invest in firms where the people feel energized.
If you look at Apple under Jobs or Pixar under [inaudible 00:20:30], or the Apollo Moon Landing Order. You know, the word that comes to mind, is ‘energized’. And so, if you can create a place where people feel energized and you won’t get to talking to the management team so much. First of all, you can almost just feel, if you walk into the office, you can almost just feel it. But you need to talk to former employees or talk to current employees and you’ll just get a sense that people are really excited to be a part of this team and anxious to help each other and equally important, the interest to ask for help.
What you want in an organization is a great deal of a free, willing, intermingling of ideas. And if you want them to be kind of creative and productive and whatnot.
On the asking for help thing, it’s totally intuitive that you know, there was a great… What you want in an organization is a great deal of a free, willing, intermingling of ideas. And if you want them to be kind of creative and productive and whatnot. And free, willing, and intermingling of ideas and a hand of information. And in order to achieve that, people need to trust each other. They need to be willing to look foolish. They need to be able to just generally, they need to ask for help when they need it. They need to offer help when it’s asked for. And they need to do the necessary reciprocity. So like, it can’t be that Bob helps Janet and Janet helps Bob, it has to be that Bob helps anybody and Janet helps anybody and so you get more transmission of help through the system.
One way to think about it, is that reciprocity…. In the economy, we solve the transaction for goods and services with money, right? But in an organization, you can’t transact for, like will you come help me with my Excel project and I’ll pay $30.00. And so, what do you do instead? So people transact, the problem is that you kind of left at barter. If you don’t have a culture where people just help everybody, you’re left at barter. So you get a whole lot fewer transactions. And so you can make the economy, so to speak, much better if you can add, infuse it with a sense, where people just sort of are… There’s no us in team, no I in team and I’m here to help and like anyone needs help. And the hardest part about it, and sometimes there’s people asking for help because you need to look vulnerable and stupid, like, I don’t know. I didn’t understand what we’re talking about. Like, can you explain this to me? There’s this reading on this dial that makes no sense to me, like, I’ve never seen it there before, but I forget. Like can you help me? If you do things like that, then good things happen. And there’s two pieces of that thing I read that really inspired my thinking along these lines. But “Give and Take” * by Adam Grant, is a book kind of about this.
I’m actually reading a book now, called “The Rainforest” *, which is sort of describing, using those same ways of thinking to describe why Silicon Valley is so productive. And then also there’s a great Harvard Business case about Shell. I think it was Shell, one of the big energy companies, and how they basically had all these really tough roughnecks, engage in trust-building exercises. And what they found was that by doing that, they radically reduced the frequency of accidents. And the theory that they had was that it was this idea that if people feel that they can be more vulnerable, and they’re more likely to ask for help and then they ask for help, information like, I don’t know what that dial means, but it seems to be at a level I haven’t seen before. That information transmits to the organization and kind of gets a response, versus someone says, oh gosh, that’s a problem. I need to prove to everyone I’m smart. Let me go ahead and in my room and read the manual, which could lead to the whole thing blowing up.
So long answer. But you want an organization where the people…. The organization itself seems to have the right norms and values, seems to be energized, and seems to be a place where people are collaborating well, where everyone’s kind of pulling well aligned. And then the leadership needs to kind of be the sort of people who foster that; not domineering, someone you want to work for. A simple test is, would you want to work for that person? Would that person inspire the best of you? And if the answer is no, then you just need to factor that into your range of expected outcomes.
A simple test is, would you want to work for that person? Would that person inspire the best of you? And if the answer is no, then you just need to factor that into your range of expected outcomes.
[00:24:37] Tilman Versch: Thank you very much for the first part of our interview. We continue the discussion in the second part.
[00:24:42] Clifford Sosin: Excellent.