Thread by @wolfejosh

1/ Here’s an overlooked widely under appreciated thing about investing.

First, the captain-obvious no-duh punchline:

look where few others are looking,
avoid competition,
be roughly right (not precisely wrong) where superior precision analysis is unnecessary...

2/ All new things come from combos of old. Let’s take two old laws and combine them.
Sturgeons Law + Bayes Law.

I’ll show you that you can hire an analyst with 90% accuracy (!) and you will still have ONLY 50/50 odds of selecting a good investment—IF you look where others are

3/ Let’s take 100 potential investments.

Sturgeons Law playfully tells us 90% of stuff is crap

Assume
90 companies are crap
10 good

Now Bayes:
-Your analyst has 90% accuracy rate discerning between crap & good

-What are odds the good co they recommended is *actually* good?

4/ Of the 90 crap they correctly said 81 are crap but mistook 9 as good (.9*90)

Of 10 good they correctly said 9 good (.9*10) but mistook 1 as bad

They present to you 18 good co’s
But only 9 are actually good
9/18 =50%

Your 90% accurate analyst?
No better than a coin toss!

5/ SO: either increase analytical accuracy (impossible, given .300 accuracy gets you in hall of fame)

OR

Go where barriers are higher, there are fewer crappy co’s because there are fewer co’s—so its maybe half-Sturgeon and instead of 10% being good, 45% are good...

6/ Much prefer and frankly much easier to select say 1 of 5 possible companies to fund than 1 of 5,000.

It’s why I’m psychotically focused on structural competitive advantage—

Less than 5 Co’s in
-nuclear waste tech
-metal 3D printing
-robotic surgery
-satellite antennas

QED

7/ Here’s a visual I sketched of the Bayesian tree starting with Sturgeons law that 90% of stuff is crap. And assuming you have 90% accuracy rate in telling the difference between crap + good...