Brazil Election: Lula, Bolsonaro and Prediction Market Odds

Today's a special day. Why? Brazil election-

Yeah

... president. Bolsonaro,

Lula, right? Day one, you guys had him at 80%.

That's right.

And if you go to it, uh, widen it a little bit more.

Yeah, so day one, go back to, like, uh, two days ago.

Go back two days ago, you had him at 87%.

That's about two weeks ago, a week ago. But that's fine, 87%.

That's about two weeks ago, a week ago. But that's fine, 87%.

Or like 75, yeah, 75.

So and all of a sudden, this flips.

That's right.

Lula wins. 15 out of 7, 17 districts

that were doing the counting of the, of, of, the votes

said Bolsonaro's gonna win.

Last minute, he pleads with the Supreme Court to open

up the voting in areas that are pro-Lula, and all of a sudden he comes out of

nowhere, wins. I don't know how many he gets,

plus minus 10.6 million.

That's right.

This has never happened in the history, right?

So for you, when you watch it, 'cause you have a data that we

don't have, what do you speculate is going on there?

So this is what's important to understand. There's two things.

One, a market doesn't create

new data. A market aggregates all the data, right?

Like Apple's stock

doesn't run the product for Apple,

right? Now Apple's

stock is at $100 a share

today, let's call it.

If Apple does a series of actions, like for example, if they...

Some bad regulatory action comes

out that they were doing something bad, uh, in

some of their factories, what's gonna happen to the stock?

It's gonna go down.

Mm-hmm.

Why? Because new information has come to light

that is now gonna be priced in by

the market. The pri- market is gonna absorb that information-

Right

... in real time. If Apple releases a bunch of great

products over the next three

weeks, what's gonna happen to the stock? It's gonna go up. So what...

And this is always

important. Markets, what they do is they absorb new

information

in real time. They're an aggregate of, and in the elections, is of the polls, the

news, uh, uh, candidate quality.

If a candidate has a scandal,

what's gonna happen

to their odds? They're gonna tank. So markets, that's what they do.

They do a very good job at summarizing all the information that's out there.

And you know that the incentive to summarize it is clean, because there's no

partisan lens or bias. Like why is Fox saying something

or CNN saying something?

There's a partisan lens in the air.

Whereas the only lens for prediction, right, is that people

are trying to get to,

to ground truth. Because if they're closer to the truth they're

gonna make money,

and if they're further from the truth they're gonna lose money.

Now

sometimes there's new piece of information that's very material, right?

Like a, if a cand- candidate is doing very well,

but then this scandal happens,

they're gonna tank very quickly. And that's what happened.

There's a big new shift that happened yesterday on the election,

which was-

Could you guys pinpoint at what it was?

Or c- could-

I think a, a big shift happened when the, uh,

I think the extensions, I think,

started happening. But I, I hadn't looked

at it, uh, quite yet.

But the more interesting way to evaluate the prediction markets, I think, if, if

you go back to the calibration plot, I think it's...

We have it in open cal- yeah, right now. If you go down a little bit, is this.

So if you, if you zoom in a little bit on some of

these charts, that's what we call

a calibration plot. So you cannot

evaluate a prediction market

accuracy on one

market. There's a study on, like, two million markets that's public,

anyone can

replicate it, where it says what is the probability

that we're saying something is

gonna happen with, and how often does it happen?

So if we say something has 20% chance, that doesn't

mean it's not going to

happen.

It means it's gonna happen one in five times.

It's gonna happen two in 10 times. Like it's gonna happen 20% of the time,

right?

Got it.

And that's so important. It's like sometimes the 10 percenter,

if you run the race 10 times, what's gonna happen?

It's gonna happen one out of 10 races.

Mm-hmm.

It's not... And so that's what shows you in the spot,

which is actually very

accurate. It's like when we

say something has a 20% chance,

it actually does happen

20% of the time.

So on the back end, what did it show to you guys?

Did, did, did you, did Kalshi know that there is a possibility that

Lula wins?

There was a fif- 25% chance,

one in four possibility that Lula wins.

And, and that it was, you know,

looking at all the prevailing factors.

And that changed, obviously, dramatically when the new material

informations came

to light. But like again, if Lula, if you play that election four

times and Lula wins zero out of those four times, then we would be wrong.

He should win one out of those four times, and that happens.

And so what, what I say with the midterms right now

is people ask me, "Oh, how do

you evaluate the midterms?" Like look, look at 100 races

that Kalshi's predicting,

and assume that the, you know, they're, it's saying that the Republican is winning

50% of the time or, like, 40% of the time.

How many of those 100 races should go Republican? 40.

It shouldn't be zero. It shouldn't be, it's not like every one of those races

should go, uh, uh, uh, Democrat.

Yeah, and that one makes more sense because it's

a volume of races, right?

You're not dealing on one.

That's right. But you have to look at elections over

time.

It's a probabilistic model, right?

A probab- like right now we're seeing the Senate

is gonna go 64% Democrat, 60

whatever. That, it's like a three-sided coin where two

sides are Democrat,

one side is Republican. It's a three-sided die.

There's three sides, two Dem, one Republican.

If you roll that die you could

still land Republican, one in three chances.

It's not a poll. And people,

people really confuse markets with polls

because polls

are what are you, what are you, what are you gonna vote for?

It's checking who's gonna vote for what.

Markets are what is the probability, what

do you think will happen-

Mm-hmm

... that something will happen?

So of the guy that does 100 million, uh, does he do a lot of

trades? Is, is it-

Yes, yeah

... personal, like so it's not one.

It's like diversifying, or is that kind of a-

For sure, for sure.

Got it.

It's very diver- you're trading. You're, you're, you're trading in and out, and

you're, you're... And, and that,

those are the guys that make the markets

accurate.

Those guys would never be allowed on a sportsbook. Why?

Because if they were making that amount of profit, who

is it going from?

The casino.

Yeah.

Yeah.

There's no, it's a very different model.

So you, you brought up the cas-

The casino is a way to move money from people to the casino.

Handcrafted across five Italian factories,

with builder technology underfoot, it may be the most

comfortable shoe you will ever own.

Introducing the all new FLB shoe. Finished

in new chocolate karezza leather.

Made to move, built to endure, designed to

impress.

If you enjoyed this video, you wanna watch more videos like this,

click here.

And if you wanna watch the entire podcast, click here.