Kalshi wants prediction markets to be seen like polling. Senators aren’t ready to make that bet.
WASHINGTON – With midterm elections fast approaching, the prediction market platform Kalshi thinks it has something special to offer candidates and voters looking for insights into high-stakes races, perhaps even to compete with polling data.
Read more In Waltham, big landlords want to replace an old office complex with new homes. The mayor says no.
The platform just launched the “Midterms Hub,” a new home base for trading on and predicting upcoming 2026 congressional elections, majorities, and governors’ races. The company hopes that visitors will see it as a data site as much as a trading hub.
“Kalshi was always built to be an information source for people,” Kalshi spokesperson Jacki McGavick told the Globe. “Polling just drops, you know, weekly or every few weeks, depending on the race … [Kalshi] updates instantaneously when someone makes a trade on our platform.”
But when asked about prediction markets this week, senators on Capitol Hill were hesitant to endorse Kalshi’s new initiative or commit to using any data it provided. In fact, the Globe could not find a single senator who embraced the idea of ditching their pollster. Many said they weren’t that aware of prediction markets in the first place.
Senator Andy Kim, a New Jersey Democrat, said he would not base campaign decisions on the data.
“There’s a scientific method to polling,” he said. “Without knowing how many people are involved in this … there’s no transparency, probably, in releasing who they are, where they’re from, things of that nature, so I just don’t see how I can trust that as anything scientific. I would not trust that as a candidate.”
Since it first received permission to list contracts on political outcomes in fall 2024, Kalshi has steadily expanded its influence in politics at a time when prediction markets are growing in popularity nationwide. The platform opened its office in D.C. in January and spent $990,000 on lobbying during the first half of 2026.
The total volume of trades on 2026 elections has already surpassed the 2024 election cycle, according to McGavick. But she said internal data found that 75 percent of site visitors are simply checking odds, while 25 percent are trading.
As of Friday afternoon, the live forecasts gave Republicans a 55 percent chance to control the Senate next year and Democrats an 84 percent chance to control the House.
While polls ask voters which candidate they hope wins, markets ask them to choose the candidate they think will win, with their own money on the line. Kalshi users trade on event contracts by buying either the “Yes” or “No” position. The two options always add up to $1, so odds are calculated based on price.
In the election for Senate in Maine, for example, the “Yes” for Democrat Troy Jackson was listed at 62 cents on Friday afternoon, predicting he has a 62 percent chance of winning against incumbent Republican Susan Collins.
The prices are constantly changing, and people can trade in and out of their positions up to election day. The markets can shed light on smaller races that polling doesn’t reliably cover and provide a nonpartisan outlook, McGavick argued.
ButSenator Thom Tillis, Republican of North Carolina, mused about the capacity of politiciansto influence wins and losses on the platform – including himself.
“I find that whole thing fascinating. We are also talking about markets that are setting lines on things that any one person can influence the outcome,” he said, referring to a previous prediction market regarding former Federal Reserve chair Jerome Powell. Tillis had held up the confirmation of Powell’s replacementthis spring until the Trump administration ended an investigation into Powell that was widely seen as a pressure campaign to get him to resign.
Read more Haeran Ryu takes a one-shot lead after 36 holes of the British Women’s Open
“Everybody’s hearing the narrative, ‘I’m betting on Powell, betting against Powell,’ but they had no earthly idea that I could have completely just changed my position and destroyed the line,” Tillis said. He also said that he’s “not a gambler.”
Other senators said the data could be valuable, and they don’t mind if constituents are betting on politics.
“I think people look at them as an alternative to polling already,” said Senator Mark Kelly, an Arizona Democrat. “Understanding where the public is on issues and candidates would be a positive thing.”
“I don’t know how to distinguish in law what you should be allowed to bet on and what you shouldn’t be allowed to bet on,” said Ohio Republican Senator Jon Husted. “Whether one should is an individual choice.”
Still, many expressed the need for more detailed information and more stringent regulation to prevent insider trading on the platforms.
In April, the Senate approved a bipartisan resolution to ban its members and their staff from using prediction markets. Another proposed bipartisan bill would ban insider trading and impose strict rules on the platforms. The House has not passed a chamber-wide ban, but Kalshi doesn’t allow members of Congress to make accounts, and campaign staff cannot trade on election markets.
Even so, stories of insider trading have made recent headlines. Earlier this month, Kalshi announced that a White House teleprompter operator won $100,000 placing bets on what President Trump would say during speeches.
“We need regulation in this area and a set of rules that Kalshi has to play by,” said MassachusettsSenator Elizabeth Warren, the topDemocrat onthe Senate Banking committee. “I’m very worried about how insiders keep doing great on these so-called prediction markets while everyone else gets hosed.”
Despite the reasons for caution, economists argue that data from prediction markets can be useful as a complement to polling and traditional methods of election forecasting. Proponents of the platforms reference the “wisdom of the crowds,” or the principle that the aggregated predictions of a large group can be far more accurate than individual opinions.
Mikhail Chernov, a finance professor at the University of California, Los Angeles, created a model with combined data from polls and market prediction prices on Polymarket, a non-US-based competitor of Kalshi. When applied to the 2024 presidential election, he found the prediction market data to be a much stronger and more informative forecast.
Rajiv Sethi, an economist at Columbia University, compared 2024 election data from Polymarket with the three majorstatistical models: The Economist, Silver Bulletin, and FiveThirtyEight. He found Polymarket to be the best forecast for the presidential contest, but the other models were better at predicting the popular vote and down-ballot congressional races.
“The debate about prediction markets is extremely polarized,” he said. “There are people who think that they’re absolutely wonderful, best things since sliced bread, and then there are other people who think that they are really actually evil … and I’m trying to really argue that neither of these extreme positions is warranted, and that one needs to take a more nuanced and more careful look at these markets.”
Post Comment