This podcast is probably my favorite. I consider it like a real-time class on value investing.

Anyway, here are my notes on the last episode.


The best line of the whole episode was from Jake Taylor and went completely unnoticed by Tobby and Bruce: “A drawdown for ants?!”, reference to this.

The market is in shambles but the indexes are still holding up thanks to the tech giants which are currently not expensive.

This is followed by


B. Brewster: “I don’t see what discretionary edge comes in picking a bucket of value stocks. I think a computer can do it better. So I’m looking for the outlier because I think that’s where creativity can have an edge.”

Note: this was sparked by a conversation on Google. The question is: can it continue to grow at a decent clip for years ahead, or will reversion to the mean happen.

Bill doesn’t see Google losing market share in search or Youtube losing relevance, at least not in the next 5 years.

I thought this was a fascinating debate to be had, it’s basically David Gardner vs Tobias Carlisle. I wish Tobias answered: sure you can do it with a computer, but people will end up interfering because of the broken leg problem:


A counter to that might be to then remove the human from the decision process all together, and use a fully automated (and possibly machine-learning) system to trade or invest. And we know at least one guy who made a killing with this approach, but it takes a lot of resources.

But I know someone who created a fund called Evovest, 100% based on machine learning, and they seem to be somewhat successful so far.

Anyway, surprisingly, I find myself more and more on Bill’s side of the debate here. The only exception where I like those cheap stonks is when they’re coming out of a weird situation. Examples from last year were Hallmark Financial Services, a specialty insurance company that had a big one time write-off followed by a market overreaction, or the infamous $GOED.

Other than these “special situations” (they’re not really special situations, they’re really just market overreactions), I now prefer to hunt for apex compounders: great businesses that will dominate their market for a long time, compound capital at fantastic rates, and have plenty of reinvestment opportunities.

I was hugely influenced by Chris Mayer, author of the 100 Baggers book, which I’ve just ordered and can’t wait to read. Chris has plenty of high-value content on the internet (written, audio and video).

I know Bill Brewster was also influenced by David Gardner, he was on his podcast recently and it was a great episode.


Jake says: “People either learn from their own suffering, or from other people’s suffering.”

I disagree slightly. I think people either learn from their own suffering (and by extension, the suffering of someone close to them, like a parent), or through the successes of other people because they envy that success.

I think it was Guy Spier who gave me a different perspective on envy and how it can be used constructively when you’re not designed like a Charlie Munger who seems immune to it.


Jake starts talking about Lyapunov time, and you know it’s going to be good when Tobby makes this face:


From wikipedia:

By convention, it is defined as the time for the distance between nearby trajectories of the system to increase by a factor of e.

An example of Lyapunov time is that of the Solar system: 5 million years. Beyond that, the trajectories start diverging exponentially.

The concept is linked to chaos theory. Jake explains:

Let’s say you wanted to be able to look out twice as long, with the same accuracy of a chaotic system, it’s gonna take 2x the effort of getting better measurements, it’s 10x.

So what he’s saying here is that the further away in time you’re trying to make predictions, to keep the same accuracy, the precision requirement on your inputs goes up exponentially high. Or you can look at it the other way around: with the same input precision, the variance of your predictions goes up exponentially with how far away in time they are.

In the investment realm, we often hear about DCFs 5 years or 10 years into the future. I assume not all investments have the same Lyapunov time, but if it is less than the span of the DCF, then the DCF predictions could be potentially completely useless.

So what’s interesting here is that moats and deep, entrenched competitive advantages extend the Lyapunov time of a business. It makes them easier to predict. You can more easily tell what the business will look like 10 years out. I have to find some links on this but I know that both Focused Compounding and Trey Henninger have talked about this multiple times.

And because it takes exponentially more effort to increase the precision of our measurements, it’s a much better proposition and more efficient use of our energy to wait for a moaty business to come by and swing hard at it when it does.

On the other hand, Jake mentions that anything macro-related probably has a much smaller Lyapunov window than we would think since there are so many non-linearly interacting variables.


Toby saw someone else say that oil is the natural interest rate in the economy. When energy prices go up, that’s the natural hand brake. When they go down, that’s the “hitting the accelerator”.

Is oil going up the thing that pops the bubble?



Bill: “I know fintwit is very bullish on oil.”

Jake: “Which makes you nervous, right?”

Bill: “Yes!”

Jake: “All these generalist bozos…”

Me: “Hi 👋”


How do you measure the moat of a business?

Bruce Greenwald explains his approach here or here. I haven’t watched these clips yet but I ordered his book Competition Demystified and I can’t wait to dive in.

The theory goes like this. You look at the scale that you need (say in terms of revenues) to be sustainable, i.e. be able to cover your fixed costs. And then you look at how much of the market you need to capture to get to that level. Finally, you divide that by how much market share changes hands on average in any given year.

The Acquires Multiple blog shows the calculation for the automobile industry with a few assumptions: 2% of the TAM as the minimum scale and 1% as the percentage that will change hands a year, gives you a moat of 2 years. With Coca-Cola on the other hand, because of how hard the distribution is, you need much bigger scale (say 25% of the market), and because of brand power, there’s very little change in market share year-over-year (say 0.2%), giving you a moat of 125 years (25 / 0.2).

In summary (from the blog):

Moat = [minimum viable scale] / [annual market share change in a contested environment]

Would be a fun exercise to do with Netflix (the stock got slashed 22% today).


Jake mentions the St. Petersburg Paradox in relation to what the future can hold for some extraordinary companies such as Google. It’s a fascinating topic and one I’d like to come back to.

Until next time, stay cool & stay invested!