Prediction markets offered another way to forecast NC’s elections. But should voters trust them?

The growing popularity of platforms such as Kalshi and Polymarket has brought prediction markets into mainstream political coverage, with their odds increasingly appearing on-screen alongside traditional polling and election forecasts.

Those markets correctly anticipated Democrat Roy Cooper’s victory in North Carolina’s U.S. Senate race, but their performance across the country was more mixed. 

Two North Carolina academics see both potential value and risks in their growing influence, particularly as the platforms become a more prominent part of how Americans follow elections.

Koleman Strumpf, an economics professor at Wake Forest University who studies prediction markets, said one of their biggest advantages is their ability to respond to new information in real time.

“The prediction markets are always open,” he said. “They’re always adjusting to information.”

Traditional polls offer a snapshot of voter preferences at a particular point in time. Prediction markets fluctuate as traders buy and sell based on what they believe will happen. 

A candidate’s odds can change as new polls are released, campaigns make announcements or election results begin coming in.

That continuous flow of information can make markets useful for tracking how a race is developing, Strumpf said. But he cautioned against assuming they are consistently more accurate than traditional polling. 

“Prediction markets and the poll modelers did about the same,” he said. “They each had their strengths and weaknesses.”

In North Carolina, prediction markets gave Cooper roughly a 97% chance of defeating Republican Michael Whatley ahead of Election Day. Cooper ultimately won by nearly 20 percentage points, defeating Whatley 58.8% to 39%.

But Strumpf said neither prediction markets nor traditional polling fully captured the extent of Democratic gains across the country. He pointed to several races where Democrats outperformed expectations, including contests in which markets had favored Republican candidates.

Still, Strumpf believes prediction markets will remain an important forecasting tool heading into the 2028 elections. 

Their ability to incorporate new information quickly, he said, gives them a role alongside polling and other election forecasts. 

Sean McKeever, a philosophy professor at Davidson College, is more skeptical of claims that prediction markets provide meaningful public benefit.

For markets to provide that broader value, they would need to offer better and more reliable information than traditional polls. McKeever said he has not seen convincing evidence that they do.

“I could be proven wrong in the future, but at this point I am not aware of any solid evidence that prediction markets perform better than polls,” he said. “The track records for both are mixed.”

McKeever said another measure of prediction markets’ value should be how far in advance they can provide useful information. Knowing who is likely to win just before Election Day offers limited benefit to voters, he argued.

“It is of little value to learn, say, 24 hours before election day who will win,” he said. “This is true of both polls and betting markets.”

A question of trust

Strumpf argued that concerns about manipulation should not be limited to just prediction markets. He said traditional polling can also be manipulated and influence political decisions.

“If we think we need to regulate and put the microscope on prediction markets, maybe we need to do the same thing with polls,” he said.

He pointed to a fabricated poll in Michigan’s 2026 Democratic U.S. Senate primary. Released by Wedgewood Polls in June, the poll showed state Sen. Mallory McMorrow trailing her opponents, U.S. Rep. Haley Stevens and Abdul El-Sayed.

McMorrow suspended her campaign less than a week later, saying the poll contributed to a loss of campaign support and momentum. 

Wedgewood later admitted that it faked the results as an experiment to demonstrate how easily false polling data could enter political coverage and election forecasts.

Strumpf said the stunt showed how misleading information can affect campaigns and public perceptions, even when it comes from a source presented as traditional polling.

Prediction markets diverge from traditional polls in NC races as Election Day nears

He also pointed to safeguards implemented by Kalshi, including identity verification and monitoring of accounts belonging to people who may have access to nonpublic political information. 

Kalshi is regulated by the U.S. Commodity Futures Trading Commission.

Strumpf said prediction market platforms have a financial incentive to prevent insider trading and manipulation because users are less likely to participate if they believe the markets are unfair.

McKeever said it’s true that people working closely with a campaign will often have access to insider information about a candidate that the public doesn’t, but the significance of that information is not always clear. 

He said campaign insiders might not be able to accurately predict which developments will matter most to voters.  

“Some scandals quickly sink a candidate while others quickly fade away,” he said. “Insiders might not be very reliable at telling one from another.