Here's what happened in prediction markets this week

The potential insider

Eight days ago, an account called junifund1 was down about $189,000 on Polymarket and buying LeBron James to Philadelphia at 9 cents. Almost nobody else wanted it. He bought 30 shares at a time, more than 1,200 separate orders, until he owned 87,377 of them for roughly $11,000. Friday morning the news landed, and 9 cents became 99.

The thing about cheap markets, though, is that you only ever hear about the one that hits.

The same board this week had people paying 5 cents for a new Stranger Things episode in a year when the finale aired seven months ago and no season six exists. It had Andrew Tate at 2% to walk out of a Miami cell with no bail granted. Those markets are cheap because a lot of people, instead of analyzing, are going with their gut. This is what we like to call "vibetrading". You see traders like junifund1 and think that everyone can do that. But for every 9-cent bet that turns into $121,000, there are thousands and thousands that turn into nothing and never get written up anywhere.

The outcomes nobody wants to buy

The other half of the week ran the opposite way.

Abiy Ahmed's party won 438 of the 501 seats in Ethiopia's June election, and betting on him to be prime minister pays 3%.

Brazil is being decided from inside a prison cell, and the market still calls it a coin flip, which two months of polls say it plainly is not.

Do you notice the pattern?

People overpay for the story and underpay for the boring thing that has already happened. That has been true at every racetrack since 1890, and it is true here.

Fortunately for us, we can capture these mispricings. Somebody has to take the other end of every one of those bets, and it may as well be us.

Where everybody went

Weekly notional volume across every prediction market, per Dune and @datadashboards, did $15.89 billion in the week of 29 June. Then $14.10 billion. Then $12.85 billion on the week that carried the World Cup final on the 19th. The week of 20 July is running at $6.02 billion.

The last figure covers about 5.5 days. If we run it at the same pace, we'll land near $7.66 billion, a fall of roughly 40%, which is immensely high.

Kalshi lost 61% of its volume, from $11.11 billion to $4.35 billion, and its share of the whole business went up, from 69.9% to 72.3%. Polymarket's global book fell 75%. predict.fun fell 71%.

To be honest, this doesn't seem strange at all. The World Cup brought in a crowd that came for the World Cup, and when it finished, of course they did not stick around for Japanese monetary policy.

$6-$7 billion a week is roughly where prediction markets sat back in early January, and for most of the year before the tournament turned up. So, the question is then: Does it keep sliding from here, or was that the floor? The only line on the chart with a direction of its own is Kalshi's share, and that has gone up every single week we have looked at.

Next week

Wednesday is the most important day of the week because we will know the Federal Reserve's decision on interest rates. For the first time this year, the argument is about whether they go up. $92 million has already gone through the July Fed market, and a 0.25 hike sits at 25%, up almost 21 points in 7 days.

No jobs number printed this week to explain it. Crude did all the work. WTI has touched $80, $85, and $90 in July. Oil is one of the few things that feeds straight through into the price of everything else. So, logically, it drives up inflation. And if the inflation print is ugly, which it will be, then a 0.25 hike stops being a tail risk and starts being the reality. That is exactly why the 0.25 bps hike market pumped so fast to 25%.

Another thing to pay attention to is Bitcoin. It's sitting near $64,000 and typically sells off after a Fed decision. At least that was the truth for the last 8/9 scenarios. This is what one can utilize to their advantage next week.