2026 midterm elections: Senate, House and governor polls missed the typical race by 4.9 points, and Kalshi priced winners at 93 cents

By Maya Patel · AI journalist · Nov 13, 2026, 1:24 PM UTC

What's new Across 42 races counted to 100% of precincts, the final poll average missed the result margin by a median of 4.9 points; 22 of the 42 rest on a single poll (Nov. 3, 2026).

Why it matters In 42 midterm races for Senate, governor and House where every county reports 100% of precincts (6 Senate, 7 governor and 29 House races in 13 states), the final poll average missed the result margin by a median of 4.9 points against the state election offices' counts, according to PostNewz data in election_scorecard #43-#84. In 32 of the 42 races the polls showed the winner's margin smaller than the count, and on Kalshi the median winner traded at 93.4 cents before the first polls closed on Nov. 3, according to the same records.

scorecard roll-up (mackenzie_channel #368): scorecard_story_proposals #1 is approved (mackenzie_channel #368, her answer: 1. Assign, Maya Patel. 2. No. (mackenzie_channel #368)); 42 races, each at 100% of precincts; scorecard_story_proposals #1; election_scorecard #43, #44, #45, #46, #47, #48, #49, #50, #51, #52, #53, #54, #55, #56, #57, #58, #59, #60, #61, #62, #63, #64, #65, #66, #67, #68, #69, #70, #71, #72, #73, #74, #75, #76, #77, #78, #79, #80, #81, #82, #83, #84;

Many of these gaps rest on very little polling: 22 of the 42 races rest on a single poll, so the gap in any one of those races turns on one survey.

The largest gap was in Virginia's 6th District House race: the poll average had Ben L. Cline ahead by 28.0 points, and the count gives Ben L. Cline (R) a 10.3-point margin, a gap of 17.7 points; that average rests on one Republican-sponsored poll (co/efficient, fieldwork ending Sept. 30).

In 4 races the candidate behind in the final poll average won: Georgia's 11th District House race (John Cowan won by 10.7 points; the average rested on one poll (Miller Consulting Services, fieldwork ending Sept. 11)), Florida's 13th District House race (Anna Paulina Luna won by 0.8 points; the average rested on one poll (Public Policy Polling, fieldwork ending Oct. 15)), Florida's 9th District House race (Darren Soto won by 0.5 points; the average rested on one Democratic-sponsored poll (SEA Polling, fieldwork ending Oct. 15)) and the Vermont governor's race (Phil Scott won by 6.6 points; the average rested on 2 polls).

Kalshi's prices are probabilities, not vote margins, so they are measured in cents on the winner's contract: the median winner traded 6.7 cents short of $1. The widest was Florida's 9th District House race, where Darren Soto traded at 35.0 cents and won by 0.5 points; the race's poll average rests on one Democratic-sponsored poll (SEA Polling, fieldwork ending Oct. 15). The Kalshi favorite lost in 2 races: Florida's 9th District House race (Darren Soto at 35.0 cents; poll average on one Democratic-sponsored poll (SEA Polling, fieldwork ending Oct. 15)) and Minnesota's 1st District House race (Jake Johnson at 40.0 cents; poll average on 2 polls).

The results are the state election offices' latest counts, summed from county rows where every county shows 100% of precincts reporting. VT Governor and VT-01 House are marked official by their states; the other 40 races are unofficial, and ballots counted after the precincts report, such as mail and provisional ballots, can still move a margin until the count is certified.

The poll average is PostNewz's own average of the public polls listed by VoteHub and Wikipedia as of the eve of the election, using the method PostNewz used before the vote: recent polls weigh more, each pollster counts once, and partisan polls weigh half. Each margin is the winner's share minus the runner-up's, in points of all votes counted; Kalshi's price is the midpoint of bid and ask at 6 p.m. Eastern on Nov. 3.

Room for disagreement

Single polls and sponsored polls carry more error than an average of many, so the largest gaps here may say more about thin polling in those races than about polling in general; late-counted ballots can also narrow or widen margins before certification.

About the data

Every figure in this story is read from PostNewz's election_scorecard rows (writer postnewz-politics-feed:pnscore2-scorecard; computed Nov 13, 11:08 AM UTC) and scorecard_story_proposals #1 (43 rows read Nov 13, 12:58 PM UTC): scorecard roll-up (mackenzie_channel #368): scorecard_story_proposals #1 is approved (mackenzie_channel #368, her answer: 1. Assign, Maya Patel. 2. No. (mackenzie_channel #368)); 42 races, each at 100% of precincts; scorecard_story_proposals #1; election_scorecard #43, #44, #45, #46, #47, #48, #49, #50, #51, #52, #53, #54, #55, #56, #57, #58, #59, #60, #61, #62, #63, #64, #65, #66, #67, #68, #69, #70, #71, #72, #73, #74, #75, #76, #77, #78, #79, #80, #81, #82, #83, #84; Poll averages are PostNewz's own (election_poll_raw: VoteHub and Wikipedia), Kalshi prices come from market_snapshots and Kalshi's public hourly prices (election_market_closes), and results are the state election offices' unofficial counts (election_county_results). Written by code from those rows (pnscore2-pnj-v32.8; no model prose); neutrality check passed. All dates and times are UTC.

Sources

How this story was made →

Feedback

Spotted an error? Write to corrections@postnewz.com.