Transcript
Podcast Announcer (0:00)
You're listening to tip.
Kyle Grieve (0:02)
Did you know that over long periods, just a handful of stocks will account for the vast majority of a portfolio's return? And that's even if half of your investments fail, the winners can still make up for the losses and then some. Now that asymmetry results from power laws and convex compounding. And once you truly understand how these mathematical forces work inside of real world systems, it will completely change the way that you think about investing. In today's episode, we're exploring the mental models from systems thinking and mathematics that have had the biggest impact on my own personal investing approach. We'll look at things such as feedback loops, kill criteria and the cone of uncertainty, and how these systems can be used to improve your thinking process. We'll examine how scale changes a business as it grows, how algorithms help you make more certain decisions, and how critical mass can propel a company into beneficial self sustaining mode. Then we'll shift over to the mathematical side of things and don't worry, you won't need to pull out a calculator to follow along. We'll look at concepts like hidden compounding, power laws, randomness and regression to the mean. And then we'll tie them all together so you can really understand just how these forces quietly shape your portfolio's long term performance. This episode is for investors who want to think more clearly. Whether you're trying to expand your mental toolbox, aiming to avoid common pitfalls, or looking for new ways to stress test your own reasoning, if you're someone who really values long term thinking, likes understanding just why things work the way they do, and wants an edge that isn't based on things like predictions or noise, then this episode is for you. Let's get right into it.
Podcast Announcer (1:38)
Since 2014 and through more than 180 million downloads, we've studied the financial markets and read the books that influence self made billionaires the most, we keep you informed and prepared for the unexpected. Now for your host, Kyle Grieve.
Kyle Grieve (2:03)
Welcome to the Investors Podcast. I'm your host Kyle Grieve and today we're going to discuss mental models from two very distinct areas, systems and mathematics. So my first introduction to systems was by reading Thinking in a Primer by Donello Meadows. That book really helped me develop a basic understanding of just how systems work. My biggest takeaway was just, you know, how these systems tend to work together and how small changes to one part of the system could cause a very massive change to outputs in other part of the system. And obviously this could be a desirable or undesirable outcome. As I began to think more and more about systems, I came across another excellent book which was the Great Metal Models, Volume 3, Systems and mathematics by Farnam Street. This book compiles numerous great mental models from these two broad areas of learning. While preparing for this episode, I also couldn't help but add yet another metal model that I use very extensively, which I learned from Annie Duke in her book Quit. I'll be discussing that metal model in a lot more detail as well as I think it synthesizes very well with systems. So in this episode I'm just going to share some of my favorite mental models from systems and mathematics, apply them more specifically to the investing landscape, and hopefully add a tool or two to your toolbox to maybe help you think differently or in a clearer way. So the first time that I ever heard of feedback loops was when I read them in Thinking in Systems by Meadows. She defined it as a closed chain of causal connections from a stock through a set of decisions, rules, physical laws or actions that are dependent on the level of the stock and back again through a flow to change the stock. Or to just put that a little more simply, the outputs of a system affect its own behaviors. Note the word stock here has nothing to do with the stock market, but as an abstraction used to determine the stock or, you know, amounts inside of a system. So let's use a real world example of an interest bearing savings account as a feedback loop. So let's say you have money in a savings account. Obviously it's going to increase if you do nothing as interest accrues and creates new deposits of those interest payments into the account. Now if you choose to keep the account at the same number, you may withdraw those interest payments over time to use for, you know, daily purposes. And that obviously is going to create an outflow. Or if you want to increase the size of your savings account, you can do so by just not touching it or contributing to it. That's really what a feedback loop is, something that just creates feedback. But we can further break down feedback into two separate types. So the first one is a stabilizing or balance feedback loop. Basically this is an equilibriating structure in a system that generates stability and resistance to change. And the second is reinforcing feedback loops which are self enhancing feedback which creates exponential growth or collapse. Now a great example of a balancing feedback loop would be when an investor has a certain amount of capital that they want to allocate inside of a specific asset class. For instance, today I have approximately 7% of my assets in crypto, 88% in public equities or stocks, and about 5% in cash. Now, let's say I want to maintain this allocation between public equities, crypto and cash. So what are the flows that would change the stock? Obviously, I'm going to have inflows that would include things such as adding cash, my brokerage account, maybe getting some dividends, or just the increases in prices in my crypto or public equities. Outflows would include things such as selling stocks, withdrawing money, or if the prices of my crypto or equities tend to fall. Now, these inflows and outflows will obviously adjust my allocation of each asset class, which provides me with feedback on whether I want to maintain those same levels of stock in each category. Now, here's where balancing comes in. So over a year's time, if I don't touch my portfolio at all, you know, don't buy or sell anything, the stock of each of these categories is 100% going to change. This year, my portfolio has done okay, and it's gone up a little bit. So my stock of public equities by the end of December here might go up from, you know, 88% to maybe 90%, although in the market we're at now, that could easily just go down. So that's my feedback. That's right there. Remember, a balancing feedback loop creates equilibrium. So we're going to imagine that in this example, I'm not adding cash at regular intervals, which isn't what I really do in real life. So if my public equities go up to, let's say, 90% by the end of the year, I would need to rebalance by increasing the outflow from my public equities. That could be done by selling some stock and withdrawing it from my portfolio. If we go back to a worse performing year, such as 2022, when my portfolio went down, I would actually need to buy more stocks with my cash and maybe sell some crypto to main, maintain my desired asset allocation numbers. So as you can see here, each action of observing, buying and selling works to restore my balance in my portfolio. You observe the discrepancy, then you take an action to minimize it. So where I really get excited about feedback loops, though, is in the reinforcing feedback loop. This is the one that has the word exponential in its definition, and that's why I like it so much. So where balancing feedback seeks to maintain balance in a system, the reinforcing feedback loop creates exponential growth. Or in the Worst case, exponential damage. So, going back to the same savings account analogy, let's say that we wanted to take advantage of a reinforcing feedback instead of just making withdrawals on interest from that account. So in that case, we just leave the interest inside of the account and not touch it. That way, the interest continues to compound as the account grows, simply by allowing interest to accrue. And if we want it to grow even faster, we just regularly deposit into that account. So where I like to use this more in real life is to model just how big my portfolio can grow simply by compounding over time. If I continue to achieve my goal of 15% interest on my investments, then my money doubles about every five years. It doesn't take many doubles for me to reach a point where I'd be financially independent, which is my ultimate goal. And I can achieve my end goal faster if I regularly make deposits into my brokerage account, allowing me to compound my returns even more quickly. But as Munger says, the key to compounding is to never interrupt it unnecessarily. The ways we interrupt it unnecessarily are pretty innumerable. But a few ways off the top of my head about how I could interrupt it would be, you know, selling a stock that has continued to compound in value for multiple years after I sell it, withdrawing cash from my portfolio for use in daily spending or emergency spending, medical emergencies or, you know, maybe buying a house. So you can argue that some of these aren't unnecessary, such as a house or medical emergencies. However, I think, you know, points two through four can be managed by setting aside specific finances so I don't have to interrupt the compounding. So where I like to think about reinforcing systems is usually in relation to the upside and the downside of my decision making on a specific business. When I'm thinking about a company, I must focus not only on what I can make from owning it, but also on what I can lose from owning it. After all, a reinforcing feedback loop can also work against us. If you require a business that has to invest capital each year just to maintain its ability to function, I think you're looking at a pretty risky business. Because what happens if that capital is no longer available? In that case, the company won't be able to operate and could theoretically be worth a zero to equity holders. And this can happen very, very quickly. I'm fortunate enough so far in my investing to have never had this happen to me. But I'm not naive enough to think that it may never become a Reality for me in the future. So when I look specifically at businesses, I'm always using feedback loops. I use these most often creating my investing theses and my maintenance due diligence process. So once I have a thesis, you can argue that I also have a framework for a system in mind. For my thesis to unfold, it requires the correct inputs. For instance, if I look at a business like Sezzle, which is a business that specializes in buy now, pay later, which I don't own, there are a few inputs that I think are required for it to succeed. So the first one is that it must increase its gross merchandise volume or just the amount of products that are being purchased with its services. It needs to increase the number of monthly on demand subscribers, and then it needs to maintain or improve the quality of the credit of its customers. Now any of these three inputs can easily change the company's fortunes. Points one and two are kind of on the inflow area of the system. And point three could be perceived as either an inflow or an outflow. However, if the credit quality of their customers were to deteriorate significantly, this could pose a very substantial risk to Sezzle, as they have lenders who fund their late fees and need to be repaid. Now all this discussion of feedback loops makes me think of one of my most used mental models, which is kill criteria. A kill criteria is a form of a pre commitment contract. It helps you commit to making a decision when noise might make making that decision a lot harder in real time. So here's what Annie Duke wrote about kill criteria. The best quitting criteria combine two things, a state and a date. A state is just what it sounds like, an object, measurable condition. You or your product is in a benchmark that you have hit or missed. And a date is simply when kill criteria generally both states and dates in the form of if I am in a particular state at a particular date at a particular time, then I have to quit, or if I haven't done X by Y time, I'll quit. Or if I haven't achieved X by the time I've spent Y, whether that's amount in money, effort, time or other resources, I should quit. So the reason that kill criteria are integrated into feedback loops is that kill criteria are a way to close the loop on feedback loops that might take a long time to actually close. So when investing, when we invest in a business, the business may be undergoing some positive things. Maybe they're transitioning to a higher margin product. And in order to sell this product to the market, it has to make long term decisions or actions such as advertising, increasing their sales staff, changing its manufacturing or R and D process, investing in new facilities, etc. And all of that costs money. So in the short term, a business that's trying to improve might have numbers that are unattractive. They may see some margin compression and decreased profits or cash flows. But if those investments have a good chance to produce earnings in the future, then they're an excellent decision for management to make. The problem with feedback loops like this is that you may not know if you are correct or not until a few years have elapsed. But over time I think there's going to be objective data points that things are going in the right direction. For instance, if a business has taken a hit to margins due to new investments, then you might say in a few years margins might improve from something like 5% to 8%. There's your state and there's your date. Your actions will be based on that trigger of the 8% margins. If margins are 8% or greater, then you take no action. And if they are less than 8%, you're probably going to sell out or if you can, maybe make some other decision whether it's a full or partial sale. Now I love this mental model because it really helps me fight the forces of complacency. I know for myself when I have an idea that I believe to be long term, I'll generally give a much longer leash to that business to allow me to keep them through a few bad quarters. But I also have ideas that get a much shorter leash. And these are businesses in my inflection point bucket. Firms like these are required to grow profits at about 25%. And if they can't keep up, then I will remove them from my portfolio. One example of how I use kill criteria in this exact scenario was with a business that I no longer own called Thermal Energy International. This business specializes in energy efficiency and emission reduction solutions for the industrial sector. So on September 30th of 2024, I wrote a journalytic entry that stated that they must meet or exceed the following criteria over the next year. The three criteria are about 37 to 40 paid development agreements, 35 to 37 million dollars in order intake, and 22 million to 24 million in backlog. If the business fails to achieve two of these three criteria, then sell. And unfortunately, all three of these numbers seemed incredibly out of reach. I didn't even need to wait a year to sell as it felt pointless as there just wasn't enough momentum in any of these KPIs to assume that they were going to meet these goals. So I ended up selling in February and March of 2025. While a business looks now like perhaps it's regaining a little bit of momentum, I think the opportunity cost of keeping my capital in that business was high and therefore I moved it elsewhere. Another sub segment of feedback loops that I like is the cone of uncertainty. This is one that I picked up from Nick Sleep and Kay Sakari after reading their shareholder letters. Here's what they wrote in the Nomad letters. What you're trying to do as an investor is exploit the fact that fewer things will happen than can happen. That is precisely what we are trying to do. We spend a considerable portion of our waking hours thinking about how company behavior can make the future more predictable and lower the risk of an investment. Costco's obsession with sharing scale benefits with customers makes that company's future much more predictable and less risky than the average business. And that is why it's our largest holding. Our smaller holdings are less predictable, but in circumstances could do much better as investments. We're just not sure that they will as their cone of uncertainty has a much greater radius than at Costco. Now this is obviously just an excellent framework for examining businesses through the lens of certainty. So you can imagine, you know, picking up a traffic cone that maybe has a 3 foot diameter on the wide end. If you look through this cone at the business landscape of a company, there's a lot of area where things can happen. But as a business such as Costco gets better and better, you can view its future using a cone with a much smaller diameter, let's call it 6 inches. In that case, the future can be looked at with a much higher degree of certainty. And when you have a higher certainty in a business, it means that your thesis is much more likely to play out the exact way that you think. And it also means that the company has fewer risks present that can derail it from success. So the simple framework for how I use the cone of uncertainty is you should have a picture of where the business is headed. And that picture should include certain KPIs or events that increase your certainty if they occur. And if certainty increases, your cone of uncertainty narrows. In the event that these KPIs do not occur, then your level of uncertainty might rise. In this case, your cone of uncertainty becomes larger. So the way that I like to use this is to make sure that the positions in my portfolio with the narrowest cone of uncertainty are also my largest positions. If I know that a position that I have has a very certain future, it means that I have an excellent grasp on the future cash flows of that business. And if the certainty of those class flows is getting more and more likely, then I want as much money in that position as possible because that will also mean that that it's above my hurdle rates. Now I'd also like to touch on a part of that excerpt that might be overlooked, which is our smallest holdings are less predictable, but in circumstances could do much better as investments. We are just not sure that they will as our cone of uncertainty has a much greater radius than Costco. So as a microcap investor, I agree entirely with this statement. While my biggest winner has come from the microcap world, I would not say that the cone of uncertainty is narrower than on some of my other positions. And because the cone of uncertainty has a pretty big radius. When I started buying it, it began as you know, just a paltry 1.5% position by cost basis in my portfolio. But as it grew, that cone of uncertainty began narrowing and I averaged up. But even as it's 10x since my initial purchase price, the cone of uncertainty on that position is not as narrow in my view as other businesses in my portfolio like Atopicus or Adino Polska. So you can think of the cone of uncertainty as a device which helps you determine how much conviction that you have in an idea. And for me, higher conviction ideas deserve more of my capital, even if the returns on my highest conviction ideas might be lower than the returns on the businesses in my portfolio with a wider cone of uncertainty. So one aspect of investing that just fascinates me is the ability for a business to scale successfully. But what exactly is scale? Scale refers to the size or magnitude of a system, entity or process and how that size changes things such as behavior, cost, complexity and dynamics of the system. When something scales up or down in size, associated parts do not necessarily increase or decrease proportionally. There are relationships and costs that can change in non linear ways that the brain has a very very tough time imagining. So the thing about scale is that it creates new problems and solutions that were non existent at smaller sizes. A micro cap business with a hundred million dollar market cap is going to have a lot different problems than Microsoft with, you know, a $3.75 trillion market cap. And even when Microsoft was a small startup, it probably had way more different issues than it does today. The things that Gates focused on when trying to grow Microsoft are now just a footnote in history when investors think of scale they usually imagine the good parts of scale, generally named economics of scale. This is when as you grow, you get access to new efficiencies. If you manufacture a product and you're growing, perhaps you start utilizing some form of automation. This automation allows you to increase output without adding new staff. So as you scale up, you sell more of that product but aren't incrementally increasing, increasing your labor expense. And this can obviously result in a lot of improved margin. This is the type of scale that benefits most investors. But obviously there's a downside to scale as well. So let's use the same manufacturing example. Let's say that you're now able to double your capacity without adding any manual labor. That's great, but now maybe there's new problems that come up. Perhaps your automation is a little more complicated than you initially thought and you need to hire a full time engineer just to oversee it. And since you're now producing double the trinkets that you used to, you have to go to your shipping partners and try to figure out how they're going to take on double the capacity in such a short period of time. These are both problems that were non existent when the business had not scaled. So while scale can create large amounts of shareholder value, it can also affect the robustness of a system. As a system scales becomes more complex and has more variables, the potential for failure may also increase. That complexity can create problems that were never conceived of or even strategized to deal with in the first place. If the proper people aren't put into place to deal with these new issues that scale will create, then the business can quickly crumble under its own growing weight. Let's take a quick break and hear from today's sponsors.