Did the Prediction Markets Call the 2026 Midterms? Election Night Scorecard

By Arthur Angelo, PredictionScout Research · Updated Oct 15, 2026, 7:40 AM ET

Key points

• What gets graded. For each race we record every exchange’s favorite and its price just before the first polls close (6:00 p.m. ET on November 3). When the market settles, we mark whether that favorite won.
• Expect some misses. A favorite at 70¢ should lose about 3 times in 10. Adding up the gaps, Kalshi’s current prices imply about 7.8 upsets across 67 races (Polymarket: 7.5 across 67). Zero upsets would actually suggest the markets were too timid.
• Most races are not close. Kalshi prices the favorite at 90¢ or more in 46 of 67 races. The real test is the 5 races under 60¢.
• Favorites that changed since Oct 1, 2026: Nevada governor.

How sure are the markets?

Sorted from closest to safest by Kalshi price. Party-level prices: where an exchange lists candidates, we add up candidates by party (same method as our midterms tracker). Click a price to open the market.

Senate races

Governor races

Control of Congress

Download the data

Race-by-race CSV

One row per race: each exchange’s favorite and price at the baseline time, the move since our first snapshot, and (after the election) the winner, the source of the result and whether each favorite won. We have archived every market price once a day since Oct 1, 2026, 10:45 PM ET (15 daily snapshots so far). Free to reuse under CC BY 4.0.

How the scoring works

• Races. All 35 Senate races, the governor races PredictionScout tracks, and the Senate control, House control and balance-of-power markets: the same set as our midterms tracker. Each race maps to one Kalshi event and one Polymarket event, checked by hand. PredictIt is included where it lists the race.
• Final pre-election price. A snapshot of every market taken shortly before the first polls close at 6:00 p.m. ET on November 3, 2026. If that snapshot fails, we use the last daily snapshot taken before then and say so on the page. Price is the midpoint of the best bid and ask when the spread is 10¢ or less, otherwise the last trade.
• Favorite. The party (or outcome, for balance of power) with the highest total price on that exchange. Candidates are grouped by party so the three exchanges can be compared.
• Result. Taken from the exchanges’ own settlement: the Kalshi contract that settled YES, or the Polymarket outcome that resolved at $1. Settlement can take days, especially for close races and recounts, so rows say “awaiting settlement” until then. If we add an AP race call before settlement, the row says so and links the call.
• Scores. “Favorite won” counts races where the favorite was the winner. The Brier score is the squared gap between the price on the winner and 100¢, averaged across races; lower is better.
• Expected upsets. If prices are accurate, a favorite priced at p loses with probability 1−p. Summing 1−p across races gives the number of upsets you would expect. It is a yardstick, not a forecast.
• Limits. About 70 races is a small sample, and many are not competitive. Races are not independent (a national swing moves many at once), so one night can look very good or very bad by chance.

Data sources: prices and volumes come from the public APIs of Kalshi and Polymarket, and PredictIt. All trademarks belong to their owners. PredictionScout is independent and not affiliated with any exchange. Nothing here is investment, betting, or legal advice.

Related: 2026 midterms tracker · How accurate are prediction markets? · All research