Prediction markets on Kalshi and Polymarket correctly predicted the winner in more than 87% of election races 24 hours before the midterm vote, a CNBC analysis found. Traders shifted toward Democrats after Indiana results began arriving before 7 p.m. on Nov. 3.
Large sums moved on specific races election night. At least $1 million traded on a Kalshi market for Texas Governor Greg Abbott's reelection after 8 p.m., though Abbott lost to Democrat Gina Hinojosa by over 700,000 votes. A little after 11 p.m., more than $1.4 million traded on Kalshi regarding Republican House control.
Kalshi exceeded $3 billion in total midterm contract volume. Polymarket's two exchanges reported nearly $700 million in midterm volume. Trading volume more than tripled from late October into midterm week.
Koleman Strumpf, an economics professor at Wake Forest University, said prediction markets "did pretty well" but "didn't outperform the best" traditional forecast polls.
Polymarket CEO Shayne Coplan and Kalshi CEO Tarek Mansour credited prediction markets for accuracy. Nate Silver, a political forecaster and Polymarket advisor, disputed this, writing that "prediction market probabilities strongly converged toward" his forecasting models and arguing that "prediction markets are low-key stealing valor from the models."
The report notes that odds and gambling platforms do not use methodologies used by traditional political polling and therefore are not substitutes for political polls.