The 2026 US midterms in prediction markets

The 2026 US midterm elections are on Tuesday 3 November 2026. As of 4 September 2026, Polymarket priced the Democratic Party at 89.5 cents to control the House and 51.5 cents to control the Senate, on $10.1 million and $4.1 million of traded volume. The poll averages point the same way on the House. The Senate is the question nobody has settled.

Prices last checked: 4 September 2026.

Where do the markets stand today?

One chamber is priced as close to decided, the other as a coin flip. That gap is the whole story of the 2026 midterms in market terms: 89.5 cents on a Democratic House is a strong claim, 51.5 cents on a Democratic Senate is an admission that the market does not know.

Market (Polymarket)	Contract	Price, 4 September 2026	Event volume, USD
Which party will win the House in 2026?	Democratic Party	89.5 cents	10,058,325
Which party will win the House in 2026?	Republican Party	11.5 cents	10,058,325
Which party will win the Senate in 2026?	Democratic Party	51.5 cents	4,122,050
Which party will win the Senate in 2026?	Republican Party	49.5 cents	4,122,050
Balance of Power: 2026 Midterms	D Senate, D House	49.5 cents	11,006,845
Balance of Power: 2026 Midterms	R Senate, D House	35.5 cents	11,006,845
Balance of Power: 2026 Midterms	R Senate, R House	11.5 cents	11,006,845
Balance of Power: 2026 Midterms	D Senate, R House	0.95 cents	11,006,845
Blue wave in 2026? (D 218+ House and 49+ Senate)	Yes	77.5 cents	60,375
Next Senate Majority Leader?	John Thune	42.5 cents	154,985
Next Senate Majority Leader?	Chuck Schumer	34.5 cents	154,985

Prices and volumes were read from the Polymarket Gamma API on 4 September 2026 for the House, Senate and balance of power events. A price in cents is a probability: 89.5 cents means the market treats the outcome as 89.5 percent likely. Every number here moves, and the table is a snapshot, not a forecast we are standing behind. Note the one cent on "Democratic Senate with a Republican House": the market treats the Senate as strictly harder for Democrats than the House.

What does the price say about the House?

The market has been drifting toward Democrats for more than a year. The Democratic House contract traded at 70.5 cents on 1 August 2025, 79.5 cents on 1 January 2026, 85.5 cents on 1 August 2026 and 89.5 cents on 4 September 2026, according to the price history behind the Polymarket House market. That is a slow, one directional move, which usually means information arriving steadily rather than one shock.

The arithmetic is tight. The House currently stands at 218 Republicans, 214 Democrats and one independent, so Democrats need a net gain of four seats for a majority of 218 (Wikipedia, read 4 September 2026). Against that, the historical base rate is brutal for the party holding the White House: in 18 of the 20 midterm elections since 1946 the president's party lost House seats (The Conversation, 19 January 2026). Gallup found that when a president's approval sat below 50 percent, his party lost an average of 37 House seats, against 14 when approval was above 50 percent, across midterms from 1946 to 2018 (Gallup, 12 September 2018).

The complication in 2026 is redistricting. Texas, California, Florida, North Carolina, Ohio and Louisiana are all priced above 92 cents to use new congressional maps, on Polymarket's redistricting market. New maps break the link between national vote share and seat count, which is why the seat count market is so spread out: the largest single bucket for Republican House seats is "below 190" at 22.5 cents.

What does the price say about the Senate?

The Senate is priced as a coin flip, and it has been priced that way for months. The Democratic Senate contract traded at 28.5 cents on 1 August 2025, 34.5 cents on 1 January 2026, 43.5 cents on 1 August 2026 and 51.5 cents on 4 September 2026, so it has crossed 50 only recently and by a hair.

The map explains the caution. The Senate stands at 53 Republicans and 47 Democrats, counting two independents who caucus with Democrats. There are 35 seats up in 2026, including special elections in Florida and Ohio, of which 22 are Republican held, and Democrats need a net gain of four (Wikipedia, read 4 September 2026). Four flips out of a handful of genuinely competitive Republican seats is a demanding path even in a good year.

The seat count market shows how uncertain that is. On 4 September 2026 the largest bucket is Republicans holding 47 seats or fewer, at 26.5 cents, then 48 and 49 seats at 13.5 cents each, 50 seats at 12.5 cents and 51 seats at 10.5 cents. No bucket is close to a majority of the probability. A market spread that thin is saying it has no strong view, and a market with no strong view is where a call is worth the most.

Which state races are actually close?

Six Senate races carry real uncertainty in the market, and three of them are effectively tied. Prices below are the favorite's price on Polymarket on 4 September 2026, from each state's Senate election winner market.

• Texas: James Talarico (D) 50.5 cents, Ken Paxton (R) 49.5 cents. A tie inside any reasonable error bar.
• Ohio special: Sherrod Brown (D) 55.5 cents, Jon Husted (R) 44.5 cents.
• Alaska: Mary Peltola 63.5 cents, Dan Sullivan 36.5 cents.
• Michigan: Abdul El-Sayed (D) 64.5 cents, Mike Rogers (R) 35 cents.
• Iowa: Ashley Hinson (R) 63.5 cents, Josh Turek (D) 36.5 cents. The one competitive seat where the market favors the Republican.
• Maine: Troy Jackson (D) 69.5 cents, Susan Collins (R) 30.5 cents.

Two more are priced as near locks: North Carolina at 91.5 cents for Roy Cooper, Georgia at 93.5 cents for Jon Ossoff. The rating desks are more cautious. As of late August and early September 2026, Cook Political Report rated Maine, Michigan, Ohio, Texas, Alaska and Iowa as toss ups, while Inside Elections had Maine, Texas, Alaska and Iowa tilting or leaning Republican (Wikipedia, read 4 September 2026). When a market says 69.5 cents and a rating desk says toss up, one of them is wrong, and by 4 November everyone will know which.

How do the markets compare with the poll averages?

They agree on direction and disagree on confidence. The Silver Bulletin generic congressional ballot average had Democrats ahead by 6.6 points on 3 September 2026 (Silver Bulletin). Presidential approval, which historically drives midterm seat loss, sat at 38.3 percent approve against 58.3 percent disapprove in the US Polling Data average read on 4 September 2026 (US Polling Data).

A national lead of about six or seven points is consistent with the House market at 89.5 cents. It is not obviously consistent with a Senate market at 51.5 cents, because the Senate map is not a national vote.

Polymarket's House popular vote market puts most of its weight in the 6 to 12 point range: Democrats by 8 to 10 points at 27.5 cents, 6 to 8 points at 23 cents and 10 to 12 points at 21.5 cents on 4 September 2026. The market broadly believes the polls on the national number and reserves its uncertainty for the seat translation. Who wins these arguments over time is covered in market, poll or pundit.

Where have election markets been wrong before?

They have been wrong recently, publicly, and with a lot of money on the line. Joshua D. Clinton and TzuFeng Huang at Vanderbilt analyzed more than 2,500 political markets across the Iowa Electronic Markets, Kalshi, PredictIt and Polymarket in the final five weeks of the 2024 campaign, and found that only 67 percent of Polymarket markets priced the eventual outcome better than chance, against 93 percent for PredictIt, with arbitrage gaps between exchanges peaking in the last two weeks (Clinton and Huang, SocArXiv, 2025). The deepest venue scored worst.

Size is its own failure mode. In October 2024 Polymarket confirmed that four accounts which had collectively spent over $28 million on one presidential outcome were run by a single French trader, and said an investigation found no evidence of manipulation (CNBC, 24 October 2024). No manipulation is needed for one participant to move a headline number that then gets quoted as though it were a poll of the country.

The practical lesson is narrow. Trust the market most on the House, where the question is national and the volume is $10.1 million. Trust it least on state races with under $250,000 of volume, where a few thousand dollars moves the price several cents. More on that in when is the market wrong.

How do I make a call before 3 November?

Pick one race, write the disagreement in one sentence, and post it while the price still has room to move. On insiderz you make a call on the same events Polymarket lists, with no money involved at any point. The call is locked the second you post it, with the time and the Polymarket price at that moment frozen alongside it. When the race resolves it is scored against that price, so being right where the market was wrong is what counts.

Timing matters more than the take. A call on the Texas Senate race at 50.5 cents on 4 September is a real claim. The same call on 2 November, after the last week of polling has landed, is close to free. The leaderboard ranks people on Beats market, Events, Edge and Early for exactly that reason.

Here are the political events open right now.

Browse the full list on the events page, see what other people are calling in the calls feed, and use the forecaster's calendar to find the next dated event after 3 November.

Method and update log

Every price here was read directly from the Polymarket Gamma API on 4 September 2026, not from a screenshot or a third party tracker, and the API URLs are in the sources so anyone can re-read them. Prices are the yes price of the named contract as returned. Volumes are event level, in US dollars, from the same endpoint. Poll figures come from named public averages with the date each was last updated, never from a single poll presented as a trend. Race ratings are quoted as ratings, not probabilities, because Cook Political Report and Inside Elections do not publish probabilities.

This page is updated weekly until 3 November 2026, and the update date changes only when a number changes. insiderz launched in September 2026 and has no accuracy record of its own yet, so nothing here is presented as evidence about insiderz.