Kalshi traders correctly favored Democrats to take Congress and win several of the election’s biggest Senate races.
They were less prepared for how far the Democratic gains would go.
In DeFi Rate’s Election Day preview, Kalshi gave Democrats a 78% chance of winning the Senate and a 94% chance of taking the House before polls opened Tuesday.
Both outcomes overwhelmingly occurred. The preview also favored Democratic winners in:
-
Iowa
-
Kansas
-
Georgia
-
Maine
-
Michigan
-
North Carolina
-
Ohio
-
Texas
But the markets missed important Republican losses. Annie Andrews defeated Sen. Darline Graham in South Carolina after traders continued favoring the Republican. In Texas, Gina Hinojosa unseated Gov. Greg Abbott, while Vikki Goodwin defeated Lt. Gov. Dan Patrick.
The early verdict is therefore more useful than a simple declaration that prediction markets won or failed, as traders identified the election’s direction, adjusted in several competitive states before voting began, and still underestimated the breadth of the Democratic result.
Election Day snapshot backed eight called Senate winners
The cleanest comparison starts with a fixed timestamp, not prices that continued moving as results arrived.
DeFi Rate’s Tuesday guide recorded nine Senate battlegrounds before polls opened. Eight have since produced Democratic winners, matching the candidate favored in that morning snapshot.
Alaska remained unresolved in the results and market recap reviewed for this story.
That makes nine called winners among the guide’s 10 listed races, with Alaska excluded from the completed comparison.
The range of probabilities matters. Turek’s 53% was almost a coin flip, while Cooper and Ossoff were overwhelming favorites. Counting all three as equally successful “calls” is convenient, but it discards how much uncertainty the market assigned to each.
Iowa and Kansas show why timing matters
A week before the election, Kalshi favored Republicans in Iowa and Kansas. By Tuesday morning, both had moved into the Democratic column, and both Democratic candidates won.
A review using week-out prices can therefore reasonably identify them as Republican-favorite upsets. A review using DeFi Rate’s Election Day snapshot records them as Democratic favorites that won.
Neither comparison is inherently wrong. They answer different questions.
The change also gives traders credit for updating. Markets did not have to maintain their original position to be useful; recognizing that a race had changed was part of the forecasting process.
But an assessment needs to specify whether it measures a week before voting, Election Day morning, immediately before polls closed, or after returns began. Otherwise, a platform can appear more accurate simply because the evaluation moves closer to the result.
South Carolina was the clearer miss
South Carolina exposed the limits of traders’ confidence in Republican strongholds.
Graham was at 72% in a DeFi Rate snapshot a week before Election Day, despite a New York Times/Siena poll showing her tied with Andrews at 48%. Reuters reported that Graham remained favored at roughly 70% before her defeat.
Andrews won with approximately 51.1% against Graham’s 46.7%, with nearly all votes counted in the reviewed results.
The market had recognized some danger: Graham’s probability declined substantially during October. But traders still assigned her a clear advantage when the most prominent late survey showed an effectively even race.
That does not prove the poll was a perfect forecast. It does show that the market’s additional confidence in Graham did not pay off.
Texas traders saw the Senate flip, not the full statewide result
Texas delivered a particularly useful comparison within one state.
Kalshi favored Talarico at 74% on Election Day morning. He defeated Republican Ken Paxton, validating the market’s willingness to back a Democrat in a state where the party had repeatedly fallen short.
But Hinojosa also defeated Abbott after the Republican governor had been favored at odds similar to Graham’s. Goodwin’s victory over Patrick added another major Democratic statewide win traders did not forecast.
Together, the results illustrate the broader limitation with traders acknowledging an opening for an individual Democrat but not fully anticipating how widely the same political environment would affect the Republican ticket.
Kalshi’s 98% claim needs qualification
Kalshi said Wednesday that favorites won 479 of 487 resolved races, approximately 98%, and reported more than $2.5 billion in midterm trading volume. Its recap covered races resolved by 6 a.m. Wednesday, excluding still-unresolved contests.
Those are company-reported figures, not an independently audited DeFi Rate accuracy score.
The company also emphasized how quickly markets reached 99% on congressional control after results began arriving. That is potentially useful information aggregation, but it is a different test from forecasting an outcome before anyone starts counting ballots.
And many election races are not competitive. A high overall winner-selection rate can coexist with meaningful misses in the battlegrounds readers care about most. The Hill’s reporting on the preliminary claims highlighted both that issue and the exclusion of unresolved races.
Picking winners is not the entire test
The strongest conclusion is narrower than either side’s victory lap.
Kalshi’s Election Day snapshot correctly favored the winners in all nine called Senate contests in DeFi Rate’s guide, and traders correctly anticipated Democratic control of both chambers.
Markets also moved toward Democrats in Iowa and Kansas before voting ended.
They nevertheless underestimated several Democratic opportunities, particularly South Carolina and the broader Texas statewide result. Kalshi itself acknowledged that traders anticipated a blue wave but did not fully price the larger one emerging from the count.
The next test is calibration: whether candidates assigned a particular probability win at roughly that frequency across a properly defined set of races. That requires fixed timestamps, complete results, and transparent methodology, not just counting how many favorites won.
For now, the evidence supports a measured verdict: Kalshi captured the election’s direction and many of its biggest flips. It did not capture every place the wave would reach.
