Thread by @KyleSamani

0/ Thoughts on predictions markets (PM)

TLDR - what matters is regularity

1/ draft kings and fan duel spend massively on CAC. so much so that they tried to merge in 2016 because they recognized that they were just bleeding each other to death (FTC blocked merger)

2/ the only way to justify large CAC is... large LTV. obviously

large LTV comes from large betting volume. obviously.

DK and FD don't care that you open an account and make a $100 bet. in fact, this is probably gross margin negative compliance costs

3/ DK and FD *need* to find people who are going to bet $10k+

well. who bets $10k+ on their first, or even 2nd, or even 5th bet?

basically no one. it takes time for people to ramp up betting size. i suspect years

4/ i think we can assume that someone who makes at least 30 bets is likely to keep on betting forever

so the question is, what set of markets are there in which a single person will make 30 bets?

5/ there is really only one answer: markets with regularity

sports is *by far* the best example of this. elections are super high profile, but extremely infrequent, and too fat-tailed in terms of attention and $

6/ the top markets on @Kalshi (which does not have US election markets) are around macro economic indicators: CPI prints, unemployment prints, will fed raise/lower rates, etc

obviously, all of these data points are released on a regular basis. so they map cleanly to betting markets

7/ i have made one bet on Kalshi: will openAI release GPT5 by EOY 2024 (i'm down mark to market)

Kalshi knows one thing about me: that I bet on this market

8/ naturally, they are trying to get me to bet on other related markets: nvidia earnings, anthropic release dates, anthropic benchmark scores, openAI BOD members, etc

i have not bet on any of those markets bc i felt 100% clueless about them

9/ let's consider what is probably the most high profile pop culture market: will travis kelce propose to taylor swift by EOY 2024

you could imagine a world in which millions of people bet on this bc it's fun

10/ but i suspect <1% of those people will convert into betting on related markets: taylor swift album sales, how long next taylor swift song stays at top of billboard, kelce's TDs in next game, etc

11/ this is the challenge with prediction markets that suffer from lack of regularity: no recurring bets, and so no LTV, and so you can't spend on CAC

12/ given how high profile @Polymarket has become both in crypto and mainstream press, we are getting a ton of pitches for new ways to approach prediction markets

most are focused on market structure mechanics (eg scalar markets, CLOBs, other derivs structures, etc)

13/ IMO those details are not the thing that matters most. what matters most is finding a set of *regular* markets that people can bet on that are currently underserved

14/ the most obvious category is esports. i am sure there are some prediction markets there, but i am not close enough to it to appreciate the dynamics of what has/hasn't worked, how to engage streamers, etc

15/ if you have some novel theories around prediction markets for some events with regularity, my DMs are open!

{fin}