New Jersey governor's race: Prediction markets edge out opinion polls, again
Add Axios as your preferred source to
see more of our stories on Google.
Traditional polls said the New Jersey governor's race was a toss-up. Prediction markets were steadfast for months that it wasn't even close.
- They were right — again.
Why it matters: Just as bettors foresaw Donald Trump's victory over Kamala Harris last year, the New Jersey governor's race is another example of crowd wisdom (with money on the line) having an edge over opinion polling.
How it works: Participants in prediction markets buy shares that represent an outcome at some price less than a dollar.
- More-likely outcomes will be priced higher than less-likely outcomes.
- When the market resolves, bettors on the correct side will get $1 for every share they bought. Those on the wrong side will get nothing.
Context: Democrats, looking at the poll numbers, were anxious about the Jersey race.
- The most recent poll before the election suggested a nailbiter, giving Democrat Mikie Sherrill only a 1% edge over her opponent, Republican Jack Ciattarelli.
- But bettors saw it very differently. On Kalshi, one of the top domestic prediction markets, they have been buying a Sherrill victory at 80 cents on the dollar or more since April.
- In the end, Sherrill won by 13 points. That gave her a much wider margin than her predecessor Phil Murphy had when he beat Ciattarelli in 2021.
The intrigue: The margin of victory may have been helped by a group that is particularly likely to use prediction markets: cryptocurrency investors.
- Stand With Crypto rates Sherrill as strongly pro-crypto and held a large voter rally amid the primaries.
Between the lines: Polls and prediction markets answer slightly different questions.
- Polls report on who people say they plan to vote for, not the odds of a candidate winning.
- Prediction markets report on what outcome bettors expect.
- So, a voter who would prefer that Ciattarelli wins might not be willing to actually bet a single dollar that he will.
The two methods can give very different results, and yet both can be technically correct.
- Imagine an election with a 100-person electorate, all of whom absolutely always vote and answer questions honestly. Suppose that when polled, 51 of them said they were going to vote for A and 49 would vote for B.
- The poll would show a 51%-49% race in A's favor, but based on the known behavior of those voters, the prediction market would come out with nearly 100% odds for A.
- Both right, but both suggesting different levels of conviction to the uninformed observer.
Reality check: Prediction markets are also subject to influences that polls are not.
- A pollster in a given race will only survey voters who are eligible to cast ballots in that contest; a prediction market is open to the world.
- That's how Kalshi ended up with Zohran Mamdani favored to win the NYC mayor's race overall — even with Andrew Cuomo having much more support from bettors actually living in New York City.
What we're watching: How and when political pros start to take these markets more seriously.
