Midterm accuracy claims put prediction markets’ favorites under scrutiny

Polymarket and Kalshi are touting their accuracy in the 2026 midterms, according to World Today Journal. Polymarket says its favored candidate won 419 of 426 resolved races, a success rate of 98.4 percent.

What the tally measures

Polymarket’s comparison uses the candidate its market favored the day before Election Day. The figure therefore measures how often that pre-election favorite won among races that had been resolved, rather than every midterm contest.

World Today Journal also reports that Kalshi is claiming high accuracy. The supplied summary does not include a corresponding race count or percentage for Kalshi, leaving no numerical basis for comparing the two platforms.

Misses remain part of the record

The platforms did not get every outcome right. According to World Today Journal, favored candidates lost in some races, including contests in South Carolina and Texas. The supplied summary does not identify those candidates or specify which platform favored them.

Polymarket’s tally leaves seven resolved races in which its pre-election favorite did not win. That limits the claim to a strong favorite-win record, not a perfect forecast of the election.

What it means for prediction markets

• Scope matters: The 98.4 percent figure applies to 426 resolved races and a snapshot taken the day before Election Day.
• Winner selection is not probability calibration: Counting winning favorites does not, by itself, establish whether the probabilities assigned to candidates matched their likelihood of winning.
• Platform comparisons remain limited: Without an equivalent numerical tally for Kalshi in the supplied material, the accuracy claims cannot be compared on the same basis.