...
By Shane Croucher
Senior Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
On Tuesday, Americans will learn who controls Congress. They will also learn whether to keep believing the two Ps that signal what may happen: the polls and the prediction markets.
This time, the two agree: a blue wave is coming. And they are surfing it toward the fate that awaits them on November 3.
The RealClearPolling (RCP) average of polls for the 2026 generic Congressional vote puts Democrats at an 8.2-point lead, 51.4 to the Republicans' 43.2. Nate Silver's polling average puts it at 51.2 to 42.5 in Democrats' favor.
Read More on Analysis
Traders on Kalshi give Democrats a 69 percent chance of taking the Senate and 92 percent of a House majority. Polymarket is similar, with Democrats at 72 percent probability for the Senate and 94 percent the House.
This is seen, too, in some of the most closely-watched races. Look at the polling in red Texas, for example, which has Democratic candidate James Talarico leading the GOP's Ken Paxton, a state that may turn blue in the U.S. Senate for the first time since 1988.
Prediction markets make Talarico the clear favorite, saying it is his to lose.
All this consensus is the danger. The two Ps will rise or fall together, because, as several political scientists told Newsweek, the second P is largely a reading of the first. In recent times, both have faltered. These midterms are a stress test of their credibility.
"I tend to look at the variation in the polls. I kind of like variation. If there's too much clustering, I get a little nervous," Christopher Wlezien of the University of Texas at Austin told Newsweek.
Related Story
Jon Ossoff Is Suspiciously Shy About 2028. Should We Believe Him?
Andrew Gelman, a professor of statistics and political science at Columbia University, says the public polls overstated Democrats by two or three points on average in 2016, 2020, and 2024 but were close to the mark in the 2018 and 2022 midterms.
Robert Erikson, a Columbia political scientist, says state polls have been impressively accurate with one exception: with Donald Trump on the ballot, Republican Senate candidates beat their polls.
"When Trump hasn’t been on the ballot (2018, 2022), this bias disappears," Erikson said.
Gelman’s tentative explanation is turnout: the occasional voters whom surveys struggle to reach leaned Republican in recent presidential races, and fewer of them vote in midterms.
Pollsters have spent a decade hunting for those voters blamed for earlier misses that have come to shape perceptions of survey accuracy. By 2024, almost 70 percent of national polls weighted their samples to party or past vote.
Wlezien guesses they are doing all right, but says only Tuesday will show empirically whether Republicans who answer surveys resemble those who vote.
Emory University’s Alan Abramowitz warns that this year’s misses need not run in 2024’s direction. Last November, with Trump off the ballot, the RCP average understated the Democrat by about 10 points in New Jersey’s governor’s race and five in Virginia’s.
Polling errors are also correlated, Gelman notes. Nonresponse that favors a party in one state tends to favor it in others, so most of the close races could miss in the same direction.
Then there is the raw material. On October 20, a self-described pollster called Wedgewood Polls admitted fabricating surveys after the aggregator FiftyPlusOne caught it out. Wedgewood had made its way into polling trackers and reporting, including Newsweek’s.
The markets made their name in 2024, favoring Trump on election morning while the polls showed a toss-up. They have since gained a reputation for indicating that events will happen before they come to pass.
But this year has been less kind. Kalshi gave single-digit odds to Angie Nixon in Florida’s Democratic Senate primary and to David Crowley in Wisconsin’s Democratic primary for governor. Both won.
In Brazil, both platforms had Flávio Bolsonaro at roughly 80 percent on the morning of the October 25 runoff. Luiz Inácio Lula da Silva won by nearly 4.4 million votes.
Abramowitz said, "I don’t pay any attention to prediction markets," and thinks they largely track the polls and pundit commentary.
"I don’t think of markets as competing with polling," said Gelman, who sees prices as bettors’ reading of the polls plus clues such as early voting.
The research backs them up. Studying presidential betting markets as far back as 1880, Erikson and Wlezien found that once scientific polling arrived, election-eve prices added nothing to the polls’ predictions, and where prices departed from the polls, "they added error rather than further accuracy."
In 1948, the markets made Thomas Dewey a nearly 9-to-1 favorite, in step with the polls. Their earlier study found adjusted poll forecasts beating the Iowa Electronic Markets’ winner-take-all prices 87 percent of the time, partly because traders overvalued long shots.
Wlezien, a fan of prediction markets in theory, notes that this work predates Kalshi and Polymarket, which might do better. But misses like Brazil give him pause.
When colleagues flagged that Kalshi had Ohio’s Democratic candidate for governor, Amy Acton, near 80 percent, his reaction was: "That’s where Bolsonaro was." She leads the polls by a few points.
...
He is also puzzled that Kalshi and Polymarket can price the same outcome differently for long periods of time. In a currency market, he notes, traders would pounce on such obvious arbitrage opportunities.
With more than $1.4 billion traded on the midterms, federal regulators have vowed to crack down on insider trading, and Kalshi has suspended six candidates for betting on their own races.
Polymarket says it has investigated "dozens" of election bets, at least two reportedly referred to the Commodity Futures Trading Commission. The Senate has banned its members and staff from trading on them.
Related Story
The Trump Era Is Dying Across the World
Next week will test the whole shebang. The polls will pass if the national House vote lands within about three points of the eight-point Democratic lead they show, whichever way states like Texas and Kansas break.
They fail if Republicans outrun them by the two or three points that, on one Fox News projection, would keep the Senate Republican, or if Democrats beat them by New Jersey margins. An error that flatters you is still an error.
For the markets, watch the headline results, but also where they have emphatically backed an outcome, like in Ohio. A second 80 percent favorite falling within nine days of Brazil would look less like bad luck than a habit for misreading scale of the probability.
Erikson’s advice to bettors is the old one: back the favorites. On Tuesday, the favorite is a big Democratic wave. If it arrives, both Ps saw it coming. If it doesn’t, they will have missed together.
In the latter scenario, many voters—and perhaps even the media—may stop listening to either. And the next time the numbers say a race is over, fewer Americans will believe them.
Related Story
Michigan Needs a Two-State Solution
Newsweek’s reporters and editors used Martyn, our AI assistant, to produce this story. Learn more about Martyn here.
Register for our November 10 discussion, 'Newsweek Webinar: How the Politics of AI Could Shape Midterms,' which will examine how AI policy played in key races, data centers, lobbying and the political lines that emerged.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines