During their recent episode, Taylor, Carlisle, and Luca Dell’Anna discussed:
- Performance Is Subordinate to Survival
- Irreversible Losses Absorb Future Gains
- Reproducible Success vs. Luck-Driven Wins
- Give Yourself a Longer Time Horizon—or Only Bad Options Remain
- Why Short-Term Optimization Destroys Long-Term Growth
- Trust Is a Compounding Asset
- AI, Probabilities, and the Blind Spot of Ergodicity
- The Kelly Criterion: Mathematically Optimal, Practically Dangerous
- Failing Well: How to Lose and Still Win Long-Term
- The Rule of 63%: Luck, Risk, and Inevitability
You can find out more about the VALUE: After Hours Podcast here – VALUE: After Hours Podcast. You can also listen to the podcast on your favorite podcast platforms here:
Breaker
Overcast
RadioPublic
Stitcher
TRANSCRIPT
Tobias: I think we’re live. This is Value: After Hours. I am Tobias Carlisle, joined as always by my cohost, Jake Taylor. Our special guest today, Luca Dell’Anna. One of our favorites. He’s back again.
Jake: Luca. He’s in the house.
Luca: Thank you so much for having me again.
Tobias: Our absolute pleasure, Luca. Thanks for coming back on. Luca’s got lots and lots of really great ideas. Two of my favorite books Luca has written, one is Ergodicity. Incredibly powerful idea. Have very, very hard idea to explain, but Luca does a phenomenal-
Jake: Impossible.
Tobias: -job articulating it. And then, Winning Long-Term Games.
We’re just going to talk very briefly about Ergodicity, then Winning Long-Term Games. Luca’s always got lots of stuff going on, so we’re going to talk about that as well. Luca Dell’Anna, welcome.
Luca: Thank you so much.
Tobias: And give us the little taste of what is Ergodicity.
Luca: So, I always explain it with a story, because it gets the concept much more powerfully true. And that’s the story of my cousin, who was a great skier since a very early age. He made it even to the world championship for his age bracket. But then, one leg injury after another, he had to quit professional skiing. And from him, I’ve learned the lesson that performance is subordinate to survival, that it’s not the fastest skier who wins the race, but the fastest one amongst those who make it to the finish line.
Performance Is Subordinate to Survival
And here, I’m not just making the banal point that survival matters to performance. We all know that, but I’m saying that it matters more than performance over the long-term. And to drive this point, I always make a numerical example. So, imagine a ski championship consisting of 10 races. And my cousin who is a very good skier, he has a 20% chance of winning each race, but also 20% chance of breaking his leg. And the question is, how many races is he expected to win over a championship of 10 races? The naive answer would be two races. Because we think 10 races, 20% chance of winning each, that makes 2.
Irreversible Losses Absorb Future Gains
However, if you crunch the numbers, the real expected number of wins is only 0.64. And the reason why of this big difference, 2 versus 0.64, is survival, because if he breaks his leg in the first race, not only he loses that race, but he also loses all following ones. This is the importance of survival. It’s because irreversible losses absorb future gains. And this phenomenon is what we call ergodicity.
Basically, the definition is that when losses are not irreversible, you are in an ergodic environment, and you can extrapolate averages. But in the real world, all losses are irreversible, they all absorb future gains, and so you cannot extrapolate averages. And that’s the basic principle of ergodicity.
Jake: And that would then be non-ergodic for everybody.
Luca: Exactly. That would be non-ergodic. Yeah.
Tobias: I love the way you explain it. I also love the fact that book wasn’t written for a finance audience. But I think you’ve been embraced by the finance community who [Luca laughs] understands this idea. That’s been a powerful idea in finance.
Jake: I’m not sure they do understand it though, TC. I think it’s one of the least understood, but most important things that’s missed.
Tobias: I would say the finance community as a whole. There’s a cycle in finance where there’s a new crop of people who assault the shore, and wind up broken on the shore.
Jake: And make the mistake of ergodicity and then– [laughs]
Tobias: Yeah, just running, leave it all– Or, running a little bit fragile. It’s a phenomenon we’ve seen– We saw it in 2020. I think we’re in the middle of it again now. I think that one of the nice things, is that you had Nassim Taleb, who’s probably the Black Swan. He has embraced the idea. Have you had any discussions with Taleb? What did he think about the idea?
Jake: Did he yell at you or anything?
[laughter]Luca: I didn’t actually have any conversation with him. We only talked very briefly, because twice, we happened to be in the same city but never managed to actually meet. Once, I just arrived in Istanbul and he was taking a plane out of it and something like that. So, I never managed to really talk to him.
But it’s from him that I first heard about ergodicity. He mentioned this briefly in his book, Antifragile– Was it Skin in the Game? Anyway, he mentions it in the book, then I’ve read Ole Peters writes the most theoretical about ergodicity, and then I wrote, let’s say, the practical part about it. Yeah.
Tobias: So, the book that you brought out, I don’t know, so recently but it’s in the last few years–
Luca: It’s going to be the five-year anniversary in next year.
Tobias: Of Ergodicity?
Luca: Yeah.
Tobias: I actually was talking about Winning Long-Term Games, which is another book that I love, which is that I think an extension of the idea of ergodicity. It’s a more practical application of the idea, and you’re talking about surviving. So, perhaps, you can just walk us through, because we haven’t had you on to discuss Winning Long-Term Games. So, walk us through the idea for the book and the ideas in it.
Why Short-Term Optimization Destroys Long-Term Growth
Luca: Yeah, so, the book is centered around fighting the false belief that if you get the most out of every day, you will also get the most out of the year. And that’s false, because if you want to get the most out of the year, you also need to do things which are not optimal for the day. Sometimes they even don’t bring any returns for the day. But that’s what allows you to grow beyond a certain level over the year. For example, it applies for investing in the sense that survival is one thing that doesn’t bring you any return today or tomorrow, but it’s necessary to make sure that your returns keep growing over a certain amount of time. But it also applies to personal life, and to professional life.
I make the example of lying. For example, lying can be a great short-term tactic. Imagine that you are a very good liar and you have a 99% chance of getting away with lying. That sounds like a perfect tactic. However, if you lie once a week, that 1% chance of getting caught means that you have a 99.5% chance of getting caught over the next 10 years. So, what looks like a great short-term tactic becomes a terrible long-term strategy. There are so many things that work this way.
Another example is with sales. So, for example, I have a newsletter, and there are two ways in which you can run a newsletter. You can build trust, or you can consume trust. There are so many newsletters which, for example, they consume the trust of the readers, every email, they’re trying to sell something. There are other people like me who instead try to build trust. I don’t write the email to sell something, but I try to write the email, so that people are more likely to read my next email.
Trust Is a Compounding Asset
Now, the crucial thing, is what’s the best strategy. It depends by the time horizon. Because if you consider a six-month time horizon, the first strategy is much better. They will make more sales. But if you consider a 10 years-time horizon, my strategy is better. So, we go back towards extrapolation. Just like ergodicity was about survival, doesn’t allow you to extrapolate expected averages.
The topic of Winning Long-Term Game, is that the effect of skills, the effect of trust and the effect of relationships do not allow you to extrapolate the return of short-term tactics to long-term. And if you want to win over the long-term, you need to accept that you will have to run strategies that are suboptimal in the short-term.
AI, Probabilities, and the Blind Spot of Ergodicity
Jake: Luca, how does AI do when it comes to– It’s using these probability weightings. Is it thinking more in averages that are more like ensemble averages, or does ergodicity factor into AI for you at all? How do you think about that?
Luca: Well, so, to answer the first part of your question, whether it talks about ensemble averages, for me, it really depends by how you prompt it.
Jake: Okay.
Luca: I’ve noticed a super big difference in how AI answers the same question, depending on how you prompt it, depending on how much details you give about your goals, about your background, your situation, what you are trying to optimize for. And then, of course, if you don’t specify any of these, it will make some assumptions about you which might be wrong. And then, of course– [crosstalk]
Jake: I should have probably prefaced a little bit in saying that the AI being trained on humans, and their thought and their words is going to exhibit human like traits. Humans are typically not very good at the intuition around ergodicity. So, that’s what I was trying to drive at.
Luca: Yeah.
Jake: So, the AI has a blind spot as well, potentially.
Luca: I think that it’s smart enough– At least, the most recent models, they’re smart enough not to have the blind spot if you remind them of that. So, it’s as you said, like, their default behavior mirrors the above average, but still relatively average human. But if you prompt them correctly, if you remind them that I think that they are relatively good, that they can be relatively good. And of course, you should never trust them. You should always use a grain of salt. But it’s really a lot about how you prompt them, about what you tell them that matters to you as well.
The Kelly Criterion: Mathematically Optimal, Practically Dangerous
Tobias: Luca, have you ever looked at anything like the Kelly criterion and the concept? How do you think about the Kelly criterion?
Luca: Yeah, so, the Kelly criterion, just for the people who don’t know about it, it’s a very simple algorithm, let’s say, to decide how much to bet. The basic idea is the better your odds, the higher the fraction of your wealth you should put on any single bet. And the interesting thing, is that mathematically, the Kelly criterion is optimal. But in the real world, any real world, sports bettor or investor, they will tell you that the Kelly criterion is too aggressive.
The reason is because the Kelly criterion is a formula. And like all formulas, they’re only as good as the variables that you plug into them. And the thing is, when we plug variables into formula, we often do not know their true value. This is even true for variables that represent historical data.
So, for example, imagine that I ask you, how much did the standard and poor 500 grew? What’s the value of the increment of the standards and poor last year? You might check the data, and maybe the data says that it was plus 25%. And you plug that in the formula and you think it’s the real value of the variable. It’s not. Why? Because what happened last year, is that the conditions of the market produce the distribution of possible outcomes from which random process picked 25%. But that’s not the real value, even if it’s the historical value.
And if you plug the historical value where there should be another value, of course, all formulas, even if they are mathematically optimal, they will produce a suboptimal result and they will lead you to mis-underestimate your risk or to over allocate on your bets. And that’s the problem with the Kelly criterion. That’s why serious bettors, they use a fractional Kelly criterion, which means you had a margin of safety.
Jake: Yeah. Not just for– Sorry, TC, I was just going to say, not just for the potential of being wrong about the inputs, but also Kelly gives you the median optimized outcome. There’s a whole bunch of outcomes below that are very bad for you potentially, and so you want to take those out of the distribution as much as you can. Sorry, TC. Go ahead.
Tobias: Conceptually, it’s a good idea, because it says the objective of Kelly is to avoid ruin. So, that would accord with both of our philosophies, you want to avoid ruin. And then, it says you want to size your bet to the frequency and magnitude of the opportunity that you have, and you can see mathematically that you want to lean more heavily on the frequency than you do on the magnitude, because the bet size is more closely associated with the frequency, even though both contribute.
But I think the main problem, is that Kelly is conceived as something that you play in series, like sitting at a blackjack table, where you have a better idea of the fixed mathematics of what you’re doing, rather than most investments and most bets and most people. We live in a world where there is an abundance of opportunity to take bets on and properly understood, Kelly should include all of those other bets, which might mean that you have to size down–
I don’t know how you do it mathematically, but mechanically, you have to size down for every opportunity out there, which means that you’re not putting 40% of your book into a single position, because it ignores all of these other high or low, but positive expected value opportunities that are out there. But I just like it. Rather than the actual use of it, I just like it as an idea that it avoids ruin and sizes to the quality of the opportunity, which in rough terms is what most investors are trying to do. I just wondered if you’d encountered, because it’s very analogous to what you talk about.
Luca: Yeah, exactly. In the book Ergodicity, I actually mentioned Kelly criterion and then I also mentioned– I cite a fantastic article by Harry Crane. I think if you Google Harry Crane St. Petersburg paradox or Harry Crane Kelly criterion, it should pop up. Basically, what he’s saying is the Kelly criterion, when you’re using it, it implies that you have an edge, so that the bet is better for you than-
Jake: The house.
Luca: -the house. Exactly.
Tobias: The other side. Yeah.
Luca: The thing is that you have uncertainty on the edge. Like, you think that you have an edge, but you don’t know whether you really have it and how big it is. And so, you need to start betting small, so that if you discover that you do not have the edge or that the edge is smaller than you think, you discover before you lost too much money. I also like this interpretation, because it really shows that the point is that there is this aspect of seriality that you mentioned about trying to understand whether the information you have and whether your assumptions are correct before you bet too much.
Jake: Yeah, that’s a good point. The information gathering there is important.
Luca: Yeah.
Tobias: It’s also clear that over betting is a big issue. So, if you have any of that lack of clarity about how good your data is, then you definitely need to be sizing down for that.
Jake: Toby, you only live once. Let’s go get in the game. [laughs]
Reproducible Success vs. Luck-Driven Wins
Tobias: That’s what’s been holding me back so far. [chuckles] This is a related idea. But one of the ideas from Winning Long-Term Games, is this reproducible success strategies, which I love. I like the way you frame it up that if it’s adopted by 100 people, it leads to success for most or all of them, rather than requiring luck or leaving many casualties behind. [chuckles] Can you perhaps expand on that a little bit?
Luca: Yeah. So, this is something that the reproducible success strategies. It’s something that came up to me talking with investors and talking with entrepreneurs who are always saying, “I’m doing okay. I’m doing good, but there is someone else that is doing much better.” I’m always telling them, “Are they doing better in a way that’s reproducible or not?” Because if they’re doing better in a way that’s reproducible, which means if they do it again and again and again, they always get great results, then you should learn from them. But if instead they’re doing better than you in a way that’s not reproducible, then you should absolutely not learn from them. You should ignore them.
Tobias: Do you have practical examples of that?
Luca: Yeah. So, for example, imagine that– I think that in the book I make the example of, imagine that the three of us, we go to the casino and we all play the roulette. But we play with some weird strategies. So, for example, Tobias, you always play on the red, and Jake only plays on the even numbers, and I only play on the 36, for example.
And then, let’s imagine that maybe Luca and Jake will lose a lot of money, and Tobias makes a lot of money. Does it mean that Tobias strategy was good? Absolutely not. It was not a reproducible strategy. And if Tobias, we went the second night, maybe the results would be different.
Now, the thing is that this is banal, because we know about the rules of roulette. It’s clear that our three strategies were stupid or random and non-reproducible. However, imagine that we are talking about stocks. Imagine that we all invest randomly on stocks and we have weird rules such as Tobias only invests in-
Jake: Chico’s.
Tobias: Companies that go down. [laughs]
Luca: Yeah. In young entrepreneurs that have a background in food and beverage, whatever. And Jake, you only invest in, I don’t know, in Norwegian entrepreneurs. And I, I only invest in entrepreneurs that never listen to earning calls. And after one year, it turns out that Tobias strategy, for example, was good.
Now, this example is analogous to the example of the casino roulette. They were random strategies, but somehow when we’re dealing with stocks, it’s a bit more plausible that Tobias strategy of only investing in entrepreneur that come from a food and beverage background, whatever, was much better. But we should recognize that again, it was not reproducible.
Jake: Luca, do you have any– Oh, sorry. Go ahead, finish your point before I– [crosstalk]
Luca: No, no, I was just going into the importance of asking yourself whether strategies are reproducible, and then how to answer that.
Jake: I was going to ask you if you had any heuristics or shorthand’s for us on, across a lot of different domains on how much data do you need to see to understand if something is reproducible or not.
Luca: Yeah. So, for me, more than data, it’s about asking yourself good questions. And so, one good question is, if you ask yourself like in 20 parallel universes, in how many will this strategy succeed? It’s a great question. And in general, you want to shift your thinking from maximizing the expected average to maximizing the distribution of averages. I don’t want to be as rich as possible to have the highest expected wealth. But for example, I just want to be wealthy in 20 out of 20 of the universes. And for me, that’s much more important. So, that’s one way of thinking.
Then another question is whether the strategy has a failure rate. Because there are some strategies which, the worst that can happen, is it’s a little bit bad, but you can keep carrying it on. And there are some strategies where instead you might be constrained to stop playing. And usually, the strategies are not reproducible. That’s the second question. And then, the third question is whether those are strategies in which you get closer to where you want to be even if you lose. And by this, I mean, that there are strategies which lose poorly and strategies which lose well.
So, for example, imagine in the context of entrepreneurship, you might do a cycle of entrepreneurship. You gather funding, you start a company, and then the company fails. But it can fail well or fail poorly, in the sense that you fail well, like you never over promised, you learned skills, you demonstrated good things, you learned things along the way. And your investors, even though of course they’re not happy that they lost their money, they would still want to invest with you again, because they saw that you demonstrated skill, that you learn things along the way. And if they bet on you three or four times, it will go well. Or, you can– [crosstalk]
Jake: I’m going to get Toby a T-shirt that says, “Value investing failing well since 2008.”
[laughter]Tobias: Ah, damn it.
Jake: Just kidding.
Failing Well: How to Lose and Still Win Long-Term
Luca: Yeah. But you get the way, or you can lose pool in the sense that you over promise to people, you take some bad bets and then they will never bet with for you again. So, that’s another telling point of whether strategies are reproducible.
Tobias: I think value investing failing well since 2009, because I think it failed along with the market in 2008. I think it’s been failing since then. But one of the questions that I wanted to ask you, Luca is this, you’ve always got this tension between playing the longer-term game and the shorter-term game. It’s not like it’s an intellectual choice that some people can make.
When you’re younger and poorer, you’re more focused on making rent or feeding yourself and you have no asset cushion. And as you get older, you should get a little bit more of an asset cushion. I found with a little bit more of an asset cushion, it’s easier to be a long-term thinker and not worry so much about what you’re doing with the money. Any money that you don’t need immediately, it’s easier to invest. So, how do you reconcile those two positions?
Luca: Well, so, for me, it’s a bit like the question of what’s the optimal time horizon for your life? The naive answer would be, like, life expectancy is whatever, 80 years old and I am 37, so my time horizon should be 43 years. But the problem, and we all know this instinctively, is that there is a high variance on the time horizon. Maybe I die next year, or maybe I live until 100 years. So, for me, it’s rather than optimizing for a single time horizon, you want to make sure that you will do well no matter what your time horizon or no matter what time horizon life gives you.
And so, for me, that means that every month you do a few things which are only good for the short-term and you do a few things which are only good for the long-term. And that means in the context of life, it means, for example, that you do a few things that are good for the long-term, like you build long-term relationships, you save a bit of money, whatever, and you do something that’s good for the short-term, like, I don’t know, you enjoy life, you do a few things like that.
In the context of business, that will mean that of course, you spend part or even a large part of your working week doing small-term or medium-term things to keep your investors happy rather than whatever. But the question is, do you also do something every week or every month, at least one thing that brings absolutely no return now, but is necessary to grow beyond a certain point over the long-term? And for me, that’s the question that you want to ask yourself.
And that, for example, could be, did you train your employees this month? You don’t need to do it a lot, but if you never do it, you will never go beyond a certain point. Did you build some relationship with some key people in your industry? Did you learn some skill yourself? That’s the kind of thing of thinking.
Tobias: Let me give a shoutout to everybody, and then JT, you want to do some veggies?
Jake: Yes, sir.
Tobias: I don’t want to miss the– Oh, we’ve got Petah Tikva in the house. What’s up? Tallahassee. Brandon, Mississippi. Breckenridge. Nashville, Tennessee. Valparaiso. What’s up, Mac? Tomball, Texas. Bologna, Italy. Lausanne, Switzerland. Mayfair, London. Boise. Casper, Wyoming. Toronto. Jupiter, Florida. Cromwell, New Zealand. Early stuff for you. Łódź, Poland. Bellevue. Mean streets of Gerrards Cross. I love it. Gokulam, India. Wiltshire, England. Tampa. Dubai. Torino, Italia.
Luca: Oh, wow. Who’s that? [crosstalk]
Tobias: Patrick Rossi.
Jake: [chuckles]
Luca: Wow.
Tobias: Says hello.
Jake: Small world.
Tobias: I think I got everybody. Apologies if I missed you. Jumps around a little bit. JT, you want to hit us with some vegetables?
Jake: I will. Before I’d be remiss if I didn’t– I’m in New York this week, and I happen to be at Santangel’s event, and so got to give a little shoutout to those guys, and Steve Friedman for getting me set up with my own little private podcast studio here, which [chuckles] I will tell you, is actually a little maternity closet for breastfeeding apparently, that’s where I’m at. [laughs] But I digress. But thanks to those guys for being so accommodating.
The Rule of 63%: Luck, Risk, and Inevitability
So, I’ve been saving this particular segment, knowing that Luca was coming on
the show– As I think he will enjoy it and also probably be able to add a lot more wisdom than I even am able to start with. So, you’ve no doubt heard of the rule of 72, right? A quick refresher. It is just a quick way to estimate how long something takes to double. So, divide 72 by the annual growth rate in percent, and that gives you how many years you would need. So, let’s say 8% return takes nine years then to double eight times nine, 72. Double in 10 years requires a 7% return, you get it.
Now, there’s another number that seems to be somehow created by the universe, and that is 63%. It’s hiding in the background of almost everything that feels like luck or fate or randomness. It’s not actually mystical. There’s something mathematical that’s happening here. It’s a probability that’s found in– If something has a one in chance of happening in one single try, and you try it N number of times, the probability that it happens at least once in those N number of tries is about 63%. It’s not 50, it’s not 90. It’s this strange middle number of 63.
That’s actually not a coincidence. Like, the shadow of the math behind it is this constant E, which is 2.718. You might remember that from calculus. I’m sure I’m more confusing than helping when it comes to that, and we’ll probably save more details about E for another veggie segment, and we can forget about all the math. But what makes the 63% rule so special, is that it connects all these things that seem very unrelated. Money marketing, risk failure, persistence habits, inevitabilities.
So, let’s just play with a few examples to help build our intuition around this. Let’s say in your job that you send out cold emails to potential clients, and each one has a one in 10 chance of getting a reply. So, if you just send one, odds are bad, 10%, right? You send 10, your chance of at least one person responding jumps to 65%. So, it’s around that 63% rule. Let’s say you drive a car every day. Your chance of an accident on any given trip might be one in 10,000. Basically, zero feels like. But if you drive 10,000 times once a day for 27 years, the probability of having at least one accident in your lifetime is about 63%.
Or, imagine an investor taking small independent bets every day, 1% chance of a catastrophic loss, sounds safe enough. It’s never going to happen. 1 in 100 that we as humans, we round that down to zero. But after 100 trading days, the chance of avoiding that big loss drops to just 37%, which of course is the flip side of 63% chance of something rare happening.
So, it’s almost like this little rule of the universe, almost like gravity’s 9.8 meters per second squared, but it’s that rule for randomness. So, you don’t really need to understand the formula. I just wanted to give you a little bit of intuition around, one over N and then times N number of entries gets you somewhere around two-thirds. That’s really the intuition that we’re trying to build here.
Tobias: Didn’t you say, Luca, that it was the expected 20% chance of blowing up, 20% chance of winning, your chance of winning one race was 0.64?
Luca: No. So, that was the expected amount of wins of [unintelligible 00:32:21]
Tobias: Why is that number? What’s the significance of E? How does E get to 0.63?
Luca: I’m not sure.
Jake: Toby, I told you, I don’t know the math.
[laughter]Jake: And I said I can’t explain it. Here’s what the equation is, okay, if you wanted to see.
Luca: I just wanted to say that the 0.64 in my example is a coincidence. It’s unrelated. Yeah. Sorry.
Jake: Right. So, it’s the math is parentheses, (1-1n)n=E-1, which is 37%. Now, whatever the hell that means, I’ll let you interpret. But that’s the math behind it. There’s one other thing to bring up in here though that’s interesting, is that we’re talking a lot about risk and you’re taking stupid chances maybe that have very small probability, but a very extreme outcome.
And of course, ergodicity plays a big factor in this. There’s a big difference between 100 people playing Russian roulette at one pole and one person playing it by themselves 100 times. But there’s a positive side to all this, which is actually about habits. And if you are doing positive things in your life that have a very small upside but a consistency to it, there’s a compounding that happens here that actually makes success almost unavoidable or inevitable.
We talk a lot about kind of negative things, and maybe we’re a little biased that way sometimes, but there’s actually an uplifting story to this is that if you take the right actions and do it consistently enough, those habits then turn in and they become very powerful.
Luca: Yeah, exactly. Like, in Winning Long-Term Games, one of the things I say is that your strategy should be inevitable. And by inevitable, I mean, it should have exactly these compounding effects. This property is that even when you fail in an endeavor, you kept building skills, relationship and trust, so that you are almost guaranteed success over of a certain amount of years.
Jake: Yeah. So, you want to let that really like– You want to be on the right side of that 63%. You want to have it work for you, and then also you want to make sure that you don’t give that 1% risk, 100 chances of killing you.
Luca: Exactly. You want to make your success inevitable and not your demise. Yeah. I wanted to add one thing, Jake. When you mentioned the probability of crashing in a car, the good thing is that you can estimate your probabilities of crashing. Because probabilities of the car, the average is very different on people. Like, even if you discover that the average probability is one in x for the average population, what happens really is that this driver has three times as much as that and the other driver maybe three times less than that.
And the interesting thing, is that so, it becomes much more important, in addition to knowing the base rate to know how you’re personally doing. The tip for this for me, is to look at near misses. So, for example, in the case of driving, what you want to check is how many times are you checking the phone, or how many times do you see yourself crossing the red line, or doing something stupid? That’s what’s going to estimate your probability of crashing.
Because you can almost say that you will crash after 10 times that you check your phone or something like that. Like now, of course, maybe the number is higher, but that’s a very good estimate. That’s a relatively good way of saying it. The more frequently you catch yourself checking the phone, the more likely you will crash over the next 10 years or something.
Jake: There’s a lot of latent data points out there that you might not be seeing.
Luca: Exactly. And this applies to everything in business. How many times are you doing near misses in business? How many times are you doing near misses investing? Things that you realize it was a terrible decision, but I got lucky. Or, oh, I missed this piece of data. That’s a very good indicator of how long it will happen to before you crash if you don’t change your habits.
Tobias: There’s been a huge spike in pedestrians getting hit over the last 10 years. And it just gets bigger every year. After being virtually rubbed out 10 years ago, it’s back to where it was 10 years before that. So, that’s a very– [crosstalk]
Jake: People staring at their phones, both sides?
Tobias: I think so. Probably, yeah. One of the things you discuss in the book, Luca, is hindsight bias, which is a topic near and dear to our hearts as well. How does hindsight bias lead people to adopt non-reproducible strategies? And then, I want to talk to you about hindsight gerrymandering, which is a great term.
Luca: Yeah. Yeah. So, hindsight bias is basically the fact that– Well, it’s a bit what we were saying before with the example of the three of us going to the casino or playing the stock market.
Jake: Can I win this time, Luca, in the example. Okay, good.
Luca: Yeah. That’s the game I like– We can have three strategies that are exactly equally as worth. But if Jake makes more money, his strategy looks smarter. And if his strategy looks smarter, we are more likely to imitate him. And that’s not necessarily right. The mistake we do, is that we often think it’s true that maybe investing this way is risky, but smart investors or people with smart strategies make profit with this. And I am smart, and therefore I will also make profits. We very often tend to think this way. The problem is–
Jake: We’re an exception to the rule.
Luca: Yeah. Or, we think that we are just like the rule. It’s a real rule that smart people usually make money. The problem is that we don’t know yet if we are smart or not. And it doesn’t translate, like, it’s not because I was smart at school that I’m smart with investing. It’s not because I made money last month that I’m smart at investing. To know whether I’m smart at investing, it takes 20 years, 30 years, whatever.
The problem is that, though, that people think I am smart, and therefore I will do well. Or, I am smart, and therefore I can get away with not taking some precautions. But you don’t know yet whether you’re smart. That’s the problem.
The thing I mentioned when I talk about hindsight gerrymandering, is this fact that we gerrymander good strategies and best strategies before we know. So, the fact that we think– I don’t know, if we can think something along the lines of– Yeah, it’s true that only– I don’t know, now, I don’t know the percentages, but it’s true maybe only 20% of investors beat the market.
Jake: I’ve got bad news for you, Luca. [crosstalk] Lower that that.
Luca: Yeah. But it’s true that those are smart investors and I am smart, so I will beat the market. Here you’re gerrymandering the strategy on a characteristic smartness which cannot be known before the fact. So, you’re using hindsight. That’s what I’m calling hindsight gerrymandering. You’re increasing your expected probability of survival based on something that doesn’t exist, and that leads to a lot of problems down the road. Yeah. And once if you’re not seeing it– [crosstalk]
Jake: What about something in the market where what works, everyone starts copying and it becomes a self-neutralizing thing? So, non-stationarity of data, how do you handle that?
Luca: Yeah. This is another big problem that maybe you were smart once you discovered a strategy or a hedge, but the question is, will this hedge remain? And if not, again, you cannot use the hindsight of the past. You cannot extrapolate the thing that stayed good for you in the past, for the future, completely. Yeah.
Tobias: Luca, you mentioned that there’s value in asking how can I achieve this in 30% of the time, and how can I achieve this in three times the time? Why is that a good way to approach a problem?
Give Yourself a Longer Time Horizon—or Only Bad Options Remain
Luca: Yeah. So, I always make the example of imagining that one day a mafia guy comes to you, and he says, “I will kill you if you don’t give me one million tomorrow.” Now, imagining that you are a normal person, you don’t have a million already in your bank account, what do you do? Maybe you sell your organs and you take the proceeds, and you go to the casino, and you put them all on the 36, and you hope that you win. The thing is, because of the very short time horizon, you only have bad options.
Now, imagine that the mafia guy tells you you have five years to give me one million. Now, your options are a bit better. You’re still not certain, but you have better options. Maybe you open a startup, you work super hard for five years, and then let’s say that you have, I don’t know, a good, decent possibility of grabbing the one million.
Now, imagine that the mafia guy tells you, “I will kill you if you don’t give me one million in 30 years. Now, it’s almost certain that of course you need to work hard, you need to make some sacrifices. But if you are decently smart and work decently hard, you can do it. So, the point is the longer your time horizon, the better options open up to you. If you want to build a successful business in 30 years, but even just in 10 years, you have some decently good options. If you want to build a very attractive physique in three years, you have very decent options and you do not need to take risk, or weird medicines or whatever.
Same thing. If you want to get a super great attractive girlfriend, spouse in five years, completely doable. But if instead you give yourself an artificially short-term horizon. I want to get an attractive physique in one month, I want to get a girlfriend tonight, I want to become a millionaire next year, then you only give yourself bad options.
And so, my suggestion is give yourself a reasonable time horizon for which you can be solidly certain, in which you can build success in a solid way that gives you relatively certain, and plus it enables you to enjoy life while you’re pursuing the success you seek. It’s a much better strategy.
Jake: What’s the name for that? Is there any math behind it or something like that we could pinpoint that or is that just a Luca special, [chuckles] the idea of optionality, opening up on farther windows?
Luca: Well, I don’t know if there is any mathematical principle. It’s just one of my rules, let’s say, yeah, one of my rules for Winning Long-Term Games is to give yourself a reasonable time horizon and everything becomes easier after that.
Jake: Call it Luca’s Razor.
Luca: Yeah, you can call it Luca’s Razor. I like that.
Jake: There we go. We got it. [chuckles]
Luca: Yeah.
Tobias: I think that there’s a lot of performance chasing, and there’s a lot of FOMO potential in this market right now, because it’s one of those times in the market where people posting huge personal accounts– [crosstalk]
Jake: A lot of victory lapse right now?
Tobias: Very, very quickly using options.
Luca: Yeah.
Tobias: I imagine that a lot of that contributes, that social pressure and that fear of missing out contributes to a lot of the short-term thinking. What do you think about that?
Luca: No, completely agree with that. The biggest limit in life, I strongly believe that is exactly the fact that you look left, you look right, you see people getting more successful than you, you feel like you’re falling behind, you’re feeling some kind of existential dread, and you think like you must success as fast as possible. It feels existential. And because of that you close yourself some good options.
I personally saw it myself. I remember when I was 16 years old, 17 years old, and almost everyone around me had a girlfriend, and I did not and I was feeling it existentially, and so I wanted to find a girlfriend tonight. I only had bad options like going to the club. [Jake laughs] But if I wasn’t attractive myself, it wasn’t helping. Problem is that I literally did this for three or five years. That was my mindset. I didn’t have a girlfriend until I was 22, I think something like that. What changed? The change that I started having as little longer time horizon, so instead of going out and meeting people, I was taking some time to go to the swimming pool, get a bit more back in shape, and things like that, and then the problem solved itself.
So, it’s really about this noticing that the fact that you feel like you’re falling behind is number one, the people who are getting ahead, maybe they will crash if they’re getting ahead in ways that’s not reproducible. And number two, even if it’s true that you’re falling behind, trying to get back as fast as possible will not help you. Yeah.
And for the people still don’t believe it or they believe it but they want an argument to show it to people, because maybe they need to justify to their boss or to the clients and so on, I give an example in the book which is the example of Alice and Bob, I think the skiers. I don’t remember exactly, because I made different variations and I don’t remember which final example went into the final version in the current version of the book.
But the basic idea was that Alice and Bob, they ski. One take is more conservative. So, he has a lower chance of winning each race, but also a lower chance of breaking his leg. It’s Bob. And instead, Alice takes more risks, so she has a higher chance of winning a trace and a higher chances of breaking her leg. And the question is who has the best strategy. And the answer which is very– I think that the percentages was Alice 20% chances of winning, 5% chances of breaking her leg, and Bob 10% chances of winning and 1% chances of breaking his leg.
I don’t remember the exact numbers, but the answer was the best strategy is neither Alice, it’s neither Bob, it depends on when we measure. I was showing that if we measure it in five races or less, Alice has the highest numbers of expected wins. And if we measure it after with more than five races, Bob has the more experience.
However, there are two psychological problems. The first one is that even if Bob has the best long-term strategy, the risk is that after three or four races, he notices that Alice is ahead and he switches to Alice’s strategy. That’s the first risk. And then, the second problem is that even though Bob is expected to win more races over 10 races, if there are a hundred Alice’s and 100 Bob’s and after 10 races, we only look at the top 10 skiers, they will all be Alice’s.
The problem is that there are a few Alice’s at the top, then all the Bob’s, and then all the Alice’s at the bottom. But if we only look at the top, we think that Alice strategies is better. However, if you are a single person, Bob’s strategy will be much better for you. We have trouble humans at understanding this, at understanding that if we see someone at the top, it doesn’t mean that his strategy is better. That’s very– [crosstalk]
Jake: I think you just described 90% of the problem in the investment world right there.
Luca: Exactly. Exactly.
Tobias: It’s very similar to Taleb has that idea about dentists being– The median dentist is more successful than the median sports star– Not sports star, like someone who tries to turn pro in sports or music or acting or something like that. The successes that you see are the ones that are the unusual outliers. And then, the median dentist or the median Bob is better than the median Alice, I think. Would the median be the right measure there? Probably.
Luca: Yeah. Yeah. But that’s an interesting example, because when I talk about those things, some people, they tell me, “Yeah, but I still want to be one of the top 10. Well, I still want to become a singer and to be a super successful singer.” Okay, let’s imagine that that’s your choice and you are conscious of the tradeoffs, and you are conscious what’s the next best thing you want to do.
It’s not to copy what the people at the top did, because there is an incredible amount of luck. And most likely, even if they did something good, you probably don’t know what amongst the 10 or 20 things that they did, which is the one that actually brought them success. You will probably not be able to know that, and you will copy the wrong thing.
Instead, the much better question to ask is how did the other people that tried fail? So, what caused other talented singers to not become superstars? That’s the key question that you want to ask yourself. What caused other talented investors smart as me to not become Warren Buffett or whatever? That’s the key question that you want to ask yourself.
Tobias: I like that. That’s [crosstalk] via negative.
Jake: Yeah, does it change anything, Luca, if the payoff to the outcomes are incredibly hugely power law driven?
Luca: Well, the more they’re power driven, the more is the impact of chance, of course. And then, the good question is then how do you expose yourself to chance? So, I make an example for myself. I am an author, I write books. And the thing that you realize as an author, is that writing a great book is the prerequisite, but it’s definitely not sufficient to get the kind of success that I don’t know Stephen King got.
So, the question is what do you need to do as an author to get that kind of success? And then, you start thinking and I’m not sure that I still have the answer, but you start noticing that it has things that have relatively little to writing a great book and much more for example, with getting your book in the right places and things like this. So, the thing is, the more the outcome is fatal, the more there is a difference from the kind of things you need to do to become good and the things that you need to do to become great, let’s say. And then, of course, what this mean is extremely domain specific. Yeah.
Tobias: How about network effects and compounding? What’s the relationship between network effects, compounding, and long-term games?
Luca: So, I think that there is one aspect that you want– So, modern network effects, I will talk about relationships, because the problem with network effects is that if we are talking it in an entrepreneurial way, then it’s easy to then become a winner takes all– We get into a winner takes all, and winner takes all is not compatible with Winning Long-Term Games in the sense that–
Again, there is the discussion of let’s assume that you decide that you want to enter a winning long-term game race, but you need to know that because it’s a race with a limited number of winners, that means that you cannot have a certainty of winning. Because even if you have the best strategy, all it takes is another person running the same best strategy and then it becomes a flip of the coin. That’s why I don’t like the term when network effects much in this context, because it brings to these considerations.
But let’s talk about relationships, which is similar to that or about trust, about which is also similar to that. I think that that’s a key element in Winning Long-Term Games. The fact that you need to build relationships, you need to build trust, those are all compounding assets that become more valuable when the time passes and become more valuable as you keep doing it. So, each relationships can bring more relationships and acquire more value, because with your relationships you can give more to your existing relationship and so on. So, that’s the angle I will get.
Tobias: You’ve published several books, and you’re a consultant. How do you use these ideas in your own life and career?
Jake: Aside from getting girls when you were younger? [chuckles]
Luca: Yeah. Well, you can see it in my career in the sense that if you subscribe to my newsletter, you will see that I definitely don’t consume the trust of my readers. And instead, I write the emails to build trust. There is extremely low amount of selling. Of course, when I publish a book, I mention it to my newsletter, but that’s not the topic. Most of the email I send, I send them to build trust, so that people are more likely to read my next email to buy my next book, not the current one, and so on. That’s the same thing.
Another thing you will notice is, for example– Well, another thing is that I published 11 books, because I know that you can write what you believe is the best book, you will have no idea how much a book can sell. There is extremely high variance, low control of that. If you want to become a successful author, you need to write more than one book. For example, if you want to have a high degree of certainty, that’s another way in which I use that.
And then, another thing is, I am very often on Twitter and I am definitely resisting the temptation to write viral tweets. Because if you want to write viral tweets, that will bring you to consuming the trust of your people. Because the quickest way to write a viral tweet is to exaggerate, or misrepresent, or playing on the outrage, or telling a funny joke. The problem is that it gets you a lot of likes, but it’s not the right kind of likes. It’s the kind of likes of I liked the joke or I trusted the lie, but it’s not the kind of like that says I trust the author more. That’s the kind of what you want when you write on Twitter is to write things that makes you trust the author more.
You will realize it yourself when you are on Twitter very often when you put a like, it’s not a like that means I trust the author more, I will give the author my money. It’s a like of, I like the tweet, the tweet made me laugh. But maybe you even trust less the author.
Jake: Luca, can you give one habit that you have that you think is the least likely that people listening would have and have the biggest impact for them?
Luca: So, I think that the biggest habit is to have a ask yourself, did I last week do at least one thing that is unnecessary in the short-term but necessary for to grow beyond a certain level in the long-term? I think that’s one of the greatest habits. And then, the second thing is to keep asking yourself. I always say the risk, the risk, the risk. So, always ask yourself how did other people with the same goal as mine fail? And that will give you a lot of insight into improving your life.
Tobias: Ladies, liquor, and leverage, [Jake laughs] according to Charlie Munger and Warren Buffett.
Hey Luca, we’ve come up on time. It’s always a pleasure. The hour goes really quickly. If folks want to follow along with what you’re doing, buy one of your books, get in touch with you, what’s the best way of doing that?
Luca: Yeah. So, you can find all my books on Amazon on most online web stores or some print web stores as well. And otherwise, my website is luca-dellanna.com. L-U-C-A dash D-E-L-L-A-N-N-A dotcom. And I’m also very active on Twitter and on YouTube.
Tobias: And on Twitter, you’re @dellannaluca. You’ve got to start with a D if you’re looking for Luca.
Luca: Yeah, exactly.
Tobias: Author of books, Ergodicity and Winning Long-Term Games, the ones that I’ve read. Thanks so much, Luca. We’ll hope to have you back again in the future.
Luca: Thank you so much. I would love that very much.
Tobias: And thanks, Team. JT, any final words?
Jake: Just always a pleasure to have Luca on. One of my favorites.
Tobias: Check out Journalytic, folks.
[Transcript provided by SpeechDocs Podcast Transcription]For all the latest news and podcasts, join our free newsletter here.
Don’t forget to check out our FREE Large Cap 1000 – Stock Screener, here at The Acquirer’s Multiple:
