My two hours spent in the future

There was no magical wisdom of crowds this time around.

When Nate Silver’s election model gave Democrats an 80% chance of winning the Senate on Oct. 31, prediction markets were at only 71%.

That’s a big win for the model.

Prediction markets did a little better on some races and a little worse on others. But, generally speaking, the surprise results were just as surprising to Kalshi and Polymarket traders as they were to pollsters. 

As late as 8 pm ET last night, for example, Kalshi gave Scott DesJarlais a 98% chance to retain his seat in the House of Representatives. He lost by about 280 votes.

As for the underlying trends, prediction markets did not seem to have any special insight into farmers flipping seats in Kansas, Nebraska, and Iowa. Or the surprisingly funny ad that’s propelled a Democrat in deep-red Mississippi to a Senate win. Or the high Latino turnout that’s sending a record 50 Hispanics to the House next year.

But something kind of magical did happen last night: Watching prediction markets made me feel like I was living about two hours in the future.

Because Kalshi and Polymarket were at least that far ahead of the TV networks, the New York Times, and my group chats.

Prediction markets got a lot of credit for being ahead of the curve in 2024, too. But that was mostly because The French Whale put his own private polls into the field and had better information than everyone else.

Otherwise, prediction markets were a mixed bag. Political scientist Joshua Clinton concluded in a definitive study of the 2024 election that their prices were neither rational nor efficient: “Prices for identical contracts diverged across exchanges, daily price changes were weakly correlated or negatively autocorrelated, and arbitrage opportunities peaked in the final two weeks before Election Day.”

This one felt different. 

Last time, prices flapped around a lot — spiking one minute and giving it all back the next; changing direction indecisively; overreacting to news. The kind of price action you’d expect in a market driven primarily by retail traders. 

Clinton’s study seemed to confirm it: “Political prediction markets appear to be driven more by idiosyncratic trading behavior, speculation, or liquidity shocks than by the systematic arrival of politically relevant information.”

Not this time, though. This time, prices were moving well before it was obvious why they were moving. They seemed to move with a purpose. They reacted decisively to race-specific news and trended confidently on news from other races. As far as I could tell, the market’s initial reaction always proved to be correct. 

In short, there was very little flapping around — which makes me think prediction market prices are now being set by professional trading firms.

This would make sense as professionals have been taking these markets more seriously as of late. Susquehanna, for example, now has 80 traders dedicated to prediction markets.

(80 traders is a lot.) 

We’ll have to wait for a political scientist’s post-mortem on the midterms to know for sure whether this has made prediction markets more rational and efficient.

But it sure felt that way.

To give you a sense of it, here’s a tick-tock of my evening spent in the future:

07:00 pm: Kalshi odds for Democrats to take the Senate spike from 75% to 88%.

I had to go looking for an explanation because there was no mention of it on ABC, CBS, CNN, or The New York Times. I finally found one on Nate Silver’s blog: A first batch of returns from an Indiana House district had been released.

But even Silver wasn’t extrapolating those results to the entire race like prediction market bettors were — correctly, as it turns out.

(Side note: The Silver Bulletin’s paywalled livestream had 115,000 viewers at one point last night. It cost me $15 to get in.) 

07:35 pm: ABC says Florida remains “very, very tight.”

Polymarket is at 93% for Republican Ashley Moody to win re-election to the Senate and 92% for Byron Donalds (no relation) to win the governorship.

ABC also calls Florida’s 13th House district a “nailbiter” — just as the Polymarket odds of Representative Anna Luna retaining the seat were skyrocketing from 50% to 83%.

The market was clearly reacting to some news. But it wasn’t news that was being reported anywhere obvious. The networks made no mention of it.

07:40 pm: James Talarico is trending, hitting new highs 20 minutes before Texas has released any results.

Prediction markets appear to be extrapolating from Latino turnout in Florida to Latino turnout in Texas (not an obvious thing to do, but again correct).

08:00 pm: Kalshi’s election dashboard suggests that if control of Congress is what you care about, you can stop paying attention.

The New York Times disagrees: Its needle says the Senate is “leaning Dem” at 74%. 

08:20 pm: In the Kansas Senate race, Democrat Adam Hamilton surges to 88%. I see no mention of this astounding development anywhere else.

08:30 pm: Update on the prediction-market odds for Democratic Senate candidates in red states: Ohio 97%, Texas 97%, South Carolina 96%, Kansas 91%, Alaska 90%.

In other words, it’s a Senate tsunami.

But CNN’s home page is devoted to news that should hardly make a ripple:

Prediction markets have had Democrat Roy Cooper above 90% since July.

08:55 pm: CNN is parsing some Texas returns that “could help lead to a win” for Talarico.

Polymarket thinks he already has. He’s at 98%.

09:00 pm: I interrupt an email debate among college friends about Ossoff’s chances in Georgia to inform them they should be debating his chances in 2028 instead.

(Weirdly, they don’t seem to appreciate me sharing my superior knowledge.)

09:06 pm: The Silver Bulletin blog notes a discrepancy between prediction markets and the needle: “NYT currently gives Sherrod Brown a 65 percent chance of winning the Ohio Senate race, while Polymarket has Brown at 98 percent.”

This was the first mention of prediction markets I’d heard, and one of only a few all night. Prediction markets went entirely unmentioned on network TV.

I suspect they wanted you to keep watching.

09:12 pm: CNN’s John King labels the Ohio Senate race “advantage Democrats,” but cautions that “we have a way to go still.”

Prediction markets have the Democrat at 98%.

09:22 pm: CBS moves South Carolina to “likely Dem” and Texas to “lean Dem.” 

Kalshi has the Democrat at 98% in both.

10:15 pm: The networks call the House for Democrats.

Prediction markets, which had been at 98% for two hours, move to 100%. 

01:00 am: The most surprising move of the night!

Prediction markets give Scott Colom an 80% chance of becoming the first Democrat to represent Mississippi in the US Senate since 1982 — up from 5% just minutes before.

The NYT needle lists the race as a “toss up” leaning in Republicans’ favor: It has Colom at just 38% to win.

07:30 am: Colom ticks up to 83% on prediction markets.

The New York Times? It’s still at 38% — because it turned the needle off at 3 am.

As of this afternoon, Mississippi remains undecided, but Colom’s odds are up to 92.6% — even though he’s just 64 votes ahead.

I don’t know how prediction markets can be so confident of a race with a 64-vote margin. There’s still 40,000 votes to count!

But that’s the beauty of prediction markets: I don’t have to know.

There may not be any wisdom-of-crowds magic at work here. But they’ve gotten magically quick at processing information.

For one night at least, that made me feel like Future Boy.

— Byron Gilliam

_{Brought to you by:}

My OnChain Net Yield Fund (MONY) is J.P. Morgan Asset Management's first tokenized money market fund, built for a new era of liquidity.

MONY, our tokenized money market fund on a public blockchain, pairs MMF relative stability and yield with onchain speed and transparency for liquidity management.