Prediction markets spent the last couple of years promising that they were the best representation of the everyday American's mindset. The midterm elections will put that to the test.
A test-run of the platforms’ accuracy went down in flames during last month’s Brazilian Presidential Election. In just 24 hours, operators will take on action much closer to home that could be crucial in determining their operational and regulatory future.
Regulatory questions
Last month, Luiz Inácio Lula da Silva defeated Flávio Bolsonaro, the son of former president Jair Bolsonaro, by a 2.6% margin, according to the country’s Superior Electoral Court. Prediction markets showed heavy support for Bolsonaro, including a 79% chance on the evening of Sunday, Oct. 25., the day of Round 2 of the election.
The predictive inaccuracy was an unwanted development for an industry that has lived in controversy for blurring the lines between sports betting and financial market activity. But then again, traders were targeting an out-of-country event with much less interest than the upcoming midterm elections.
“[The midterms are] the first meaningful prediction market election,” Kalshi CEO Tarek Mansour recently said.
An FAQ answer in Kalshi’s midterm elections hub also says that market prices were “well-calibrated” across over 2.2 million markets from 2021 to mid-2026. Kalshi’s Brier Scores — a measure of predictive accuracy that scores outcomes from 0-1, with 0 being perfect — fell from about 0.09 three months out to 0.02 at close.
Those claims create expectations for operators such as Kalshi and Polymarket who are fighting to continue offering event contracts nationwide. Tuesday’s events will determine if they prove to be fruitful or damning.
Gambling or legitimate information?
Prediction markets were among the first entities to predict that President Donald Trump would return to the White House after the 2024 Presidential Election. Most polls and major media outlets believed that the race was a “toss-up” and wouldn’t be decided until the late hours of election night.
That is a critically important endorsement of prediction markets’ accuracy, given what just happened in Brazil.
Still, not all are convinced.
“Prediction markets provide event contract prices, not probabilities,” said Thomas Miller, faculty director of the data science master's program at Northwestern University. “Traders are guessing. We are not.”
Prediction markets update their midterm elections data by the buying interest of consumers. In other words, an increase in demand for a specific contract will increase the price of that outcome, therefore increasing the outcome’s probability. The system largely depends on the sentiment of users instead of unimpeded evidence.
At the same time, the midterm elections could be the perfect example of an individually controlled situation in which that aids prediction markets. Unlike any NFL or NBA game, prediction traders can’t assert their influence by taking a jump shot or running into the end zone — but traders at prediction markets are the same people who could vote in the midterm elections and other public events.
According to Kalshi, Democrats have a 94% chance of controlling the House and a 74% chance of achieving a majority in the Senate, at the time of writing. That follows the general trend of political counter-balancing after years of one party’s control, in this case, the Republicans, who had Congress and the White House for the last two years.
More than $1.4 billion has been traded in midterm elections prediction markets as of Monday, just 24 hours before votes are counted. The accuracy of prediction data could help institutionalize markets and ensure their regulatory autonomy, or it could leave them as inaccurate outcasts who had already made enemies with many officials across the country.
The Trade Handle Prediction Markets Take
Politics are viewed as one of the more “legitimate” industries with which prediction markets overlap. A successful showing during the midterm elections, particularly if prediction markets are once again more accurate than polls, would put a significant damper on claims that they are not real “tools” and are simply another form of gambling, as has been suggested.