Prediction Markets Picked Texas Democratic Primary Winner But Missed on GOP Race

James Talarico’s victory over Jasmine Crockett in the race for the Democratic nomination for Senate in Texas came as a surprise to many poll watchers. Prediction-market traders never had much doubt.

An average of polls collected by Decision Desk HQ showed Crockett leading for nearly the entire campaign until Talarico captured a narrow lead in its closing days. But within a week of Crockett announcing her plans to run in December, traders on markets like Kalshi and Polymarket placed her odds of winning below 40% — and kept them there through much of the race.

The Texas Senate primary has been the most closely watched early test of how voters might tilt in this year’s midterm elections. Talarico, a religious populist who drew national attention while championing higher taxes for the wealthy, beat Crockett, a Dallas-area congresswoman and progressive firebrand, by more than six points.

Prediction-market traders were less successful in picking a winner in the state’s Republican Senate primary, which advanced to a runoff between incumbent John Cornyn and state Attorney General Ken Paxton. Most polls and prediction markets expected the race to advance to a runoff with Paxton leading Cornyn in the first round, but with 94% of votes counted, Cornyn topped Paxton by more than 25,000 votes.

Both Cornyn and Paxton are well-known figures in Texas politics, making the race a tough call. While Cornyn is a longtime senator backed by some of the state’s wealthiest residents, Paxton sought to appeal to hardcore followers of President Donald Trump. The presence of Representative Wesley Hunt, who finished a distant third, also made the GOP race more complex.

Traders had favored Paxton in the runoff by ten percentage points on Wednesday morning, but activity was volatile in the afternoon after Trump said on Truth Social that he plans to endorse a candidate soon.

Texas primaries can be difficult to forecast because voters can choose to vote in either party’s race. Surveys released in February ranged from a six-point Talarico win to a 12-point Crockett romp. On Election Day morning, markets on Kalshi and Polymarket gave Talarico an 87% chance of victory.

Prediction markets allow users to trade contracts on the outcomes of real-world events. Users have placed billions of dollars in bets on everything from crude oil prices to the Super Bowl. Yet some observers have questioned the extent to which the markets reflect real developments, as opposed to simple changes in sentiment among traders themselves.

“It’s not clear how much actual information people have when they’re trading on this stuff,” said Joshua Clinton, a Vanderbilt University professor who has researched prediction market efficacy in elections. “A lot of the time, there’s not really all that much information, so it’s just aggregating vibes.”

On Tuesday, as voters cast their ballots, traders poured money into Texas bets. The Democratic primary became the 10th most actively traded contract on Polymarket, according to data compiled by Bloomberg. By Wednesday, the Democratic primary markets on Kalshi and Polymarket had fielded $20.8 million in bets.

Supporters of prediction markets argue that they are more reliable at predicting outcomes when trading activity is high. “The more trading’s involved, the more accurate they tend to be,” said Robin Hanson, an associate professor of economics at George Mason University who is considered a founding father of prediction markets.

In recent days, criticism of prediction markets has grown as traders placed bets on the Iran war. Businesses like Kalshi and Polymarket are licensed and supervised by the US Commodity Futures Trading Commission.

News outlets including Dow Jones, CNN and CNBC have signed partnerships with prediction markets to supplement coverage with their live data. On Monday, the Associated Press said it would share election-results data with Kalshi.