How fast the prediction markets called the 2026 midterms: 95% within an hour of the polls closing, and Kalshi ahead of Polymarket in every race

Prediction markets knew the results of the 2026 midterms long before they paid out. We timed every Senate and governor's race that had settled by Wednesday morning: when the winner's price first reached 95%, and when each platform stopped trading the market. The answer: in competitive races the market was sure within about an hour of the polls closing, but trading continued for hours more while the platforms waited for the race calls their rules require.

The data are in a CSV file.

Key findings

1. The markets were fast. In the 26 races where the winner's Kalshi price was between 5% and 95% at 6 p.m. ET on Election Day, it reached 95% a median of about one hour after the state's last polls closed.
2. Settlement was slower. Kalshi halted trading in those races a median of 3.7 hours after the polls closed; Polymarket a median of 6.75 hours.
3. Kalshi stopped trading first in every race we could match: 63 out of 63. Across all those races the median was 2.2 hours for Kalshi and 5.2 hours for Polymarket.
4. Safe races settled within the hour on Kalshi: a median of 0.7 hours after the polls closed.

Why the gap between knowing and paying

The difference comes from the rules. Kalshi's race markets can be determined early once "a consensus of media organisations" projects a winner. Polymarket's race markets resolve "once all three" of the Associated Press, Fox News and NBC "call the race for the same candidate", and then go through Polymarket's resolution process. Waiting for three named organisations, the last of which may call a race hours after the others, takes longer.

Neither approach is wrong. Kalshi's rule gets traders paid faster; Polymarket's names its sources precisely, so traders know in advance exactly which calls will count. The trade-off is time. Our guide to how election markets settle explains both.

The competitive races, timed

Hours after the state's last polls closed. "0.0" means the price reached 95% within the hour that the last polls closed; "before close" means before the state's last polls closed, since parts of some states close earlier. Source: our analysis of Kalshi and Polymarket data.

A few races stand out:

• Texas. The most traded race of the cycle was effectively decided within the hour the last Texas polls closed: Talarico's price reached 95% almost immediately, helped by the speed of Texas's early-vote reporting. Kalshi halted trading about two hours later, Polymarket about five.
• The upsets were fast too. Hinojosa in Texas and Andrews in South Carolina started the night as underdogs at 37% and 30%. Their prices reached 95% within one and two hours of the polls closing.
• Michigan. The market was sure of El-Sayed early, but Kalshi did not stop trading until 5.6 hours after the polls closed. The gap reflects how long it took for enough media organisations to call the race.
• Nevada. The slowest of the big races on both platforms: 7.2 hours on Kalshi and 10.6 on Polymarket, as Nevada counted overnight.

The safe races

Races that were never in doubt settled almost as soon as the polls closed. On Kalshi, the Delaware, Illinois and Virginia Senate markets and the California and Illinois governor's markets all closed within about six minutes of the state's last polls closing; Massachusetts within about 12. Across the 39 races where the winner was priced at 95% or more before the polls closed, the median time to Kalshi's close was 0.7 hours.

A few safe races took longer. The Nebraska Senate market, where Pete Ricketts was always the heavy favourite, did not close on Kalshi until about 6.6 hours after the polls. Montana's Senate race took five hours, and South Dakota's Senate and governor's races more than four. Those delays reflect when enough media organisations called each race, not doubt among traders: in each case the winner's price had been near 99% for hours before the market closed.

Did the time zone matter?

Source: our analysis of all settled races, safe and competitive.

The 8 p.m. group settled fastest because it contained many safe Democratic and Republican states in the Northeast and South. The earliest states, which included several competitive races in Georgia, South Carolina, North Carolina and Ohio, took longer on average. Western races, which close later and often count mail ballots for longer, took longest on Polymarket.

What slowed the slow races

• Michigan. El-Sayed's price passed 95% within an hour, but the media calls took time as Detroit and its suburbs counted. For more than four hours, the market was effectively certain while the race was officially undecided.
• Maine. Maine uses ranked-choice voting in federal elections. If no candidate wins a majority of first choices, a second round of counting follows, so media organisations are often cautious about calling close Maine races.
• Nevada. Nevada casts a large share of its vote by mail, and its biggest counties report in batches overnight.
• Arizona. Hobbs was a 93% favourite before the polls closed, but Arizona's count is slow and the governor's market stayed open for six hours on Kalshi and more than nine on Polymarket.

Compared with 2024

The speed of the 2026 markets matches what we saw two years ago. On election night 2024, Polymarket's price for Donald Trump passed 95% at about midnight ET, more than five hours before the Associated Press called the race (our replay of election night 2024). In 2026, the Senate control market on Kalshi passed 98% by about 9 p.m. ET and halted trading at 11:46 p.m. ET, once enough media organisations had projected Democratic control. In both elections, the prices worked out the answer within a few hours of the polls closing, and the formal resolution came later.

Did Polymarket's prices move more slowly?

No. The difference is in settlement, not in how fast traders worked out the result. Measured the same way, the winner's price on Polymarket reached 95% a median of about 0.75 hours after the polls closed in the competitive races, slightly faster than on Kalshi. Traders on both platforms reacted to the same county results at roughly the same time.

What this means for traders

• A price of 95% or more on election night usually means the race is over, but not settled. In this election, every competitive race that reached 95% on Kalshi went the way the price said.
• Holding to settlement costs time. On Polymarket, money sat in a decided market for several hours longer on average. For a race decided at 9 p.m. and settled at 4 a.m., that matters little; in a close race that waits for certification it can matter a lot.
• Settlement speed can explain small price differences late on election night. A contract that will pay out sooner is worth slightly more.

FAQ

How quickly did the markets know who won? In competitive races, the winner's price reached 95% a median of about one hour after the state's last polls closed.

Why did Polymarket take longer to close its markets? Its rules need the Associated Press, Fox News and NBC all to call a race before it resolves; Kalshi's need a consensus of media organisations.

Which race took longest? Of the competitive races, Nevada's governor's race: 7.2 hours on Kalshi and 10.6 on Polymarket after the polls closed.

Does a faster close mean a more accurate market? No. It reflects the settlement rules. The prices on both platforms moved at about the same speed.

Method

• Races. All Senate and governor's races that Kalshi had settled by 4 November, 65 races in all, of which 63 could be matched to a Polymarket market. Mississippi's and Alaska's Senate races, still open, are excluded.
• Poll closing. We used each state's last poll-closing time in Eastern Time, from our Election Day 2026 guide. Parts of some states close earlier, so a price can reach 95% before a state's last polls close.
• 95% time. The first hourly Kalshi candle (or Polymarket price point) at or after an hour before the last poll closing in which the winner's price was 95% or more. Hourly resolution means times can be up to an hour late.
• Close time. Kalshi's market close time, when it halted trading because the outcome was determined; Polymarket's market closed time.
• Competitive races. Races where the winner's Kalshi price at 6 p.m. ET on Election Day was between 5% and 95%.

Download the data

The CSV file lists each race with the winner, the winner's price at 6 p.m. ET, the last poll-closing time and the hours to 95%, to Kalshi's close and to Polymarket's close.

For how the night began, see our Election Day report from before the polls closed and, for comparison, our replay of election night 2024.

Data: Kalshi and Polymarket public APIs. Not investment advice.