Democrats at 81%: Why Prediction Markets Just Doubled the Odds of a Blue House

In February, Polymarket — the world’s largest prediction market — gave Democrats roughly a 45% chance of winning the House. A coin toss, basically. Republicans held a narrow structural advantage from redistricting, and the national environment, while trending Democratic, hadn’t broken open.

Three months later, that number is 81%.

Kalshi, another major prediction market, shows a similar story: 40% odds of a full Democratic sweep (House + Senate), up from roughly 20% earlier in the year. What happened?

Five Things That Moved the Market

1. The generic ballot stabilized at D+6 to D+8. This is the single most predictive metric for House outcomes, and it has been remarkably consistent since March. FiftyPlusOne’s aggregate sits at D+6 (May 23). Data for Progress has D+8 (May 18). RealClearPolitics shows D+7.2. PBS/NPR/Marist produced a staggering D+14 among likely voters, with independents preferring Democrats by 33 points. These aren’t outliers anymore — they’re the trend line.

Historical context matters here. A sustained D+6 generic ballot, if it holds to November, would historically project to a Democratic net gain of 15–30 House seats. Democrats need a net gain of just 6. The math is overwhelmingly in their favor unless the numbers collapse.

2. Trump’s approval hit structural lows. The American Research Group poll (May 16–20) found 31% approval — Trump’s worst number across both terms. Fox News, hardly a Democratic outlet, found 39% approve / 61% disapprove. The FiftyPlusOne aggregate sits at 36.4% approve, 60.1% disapprove. Silver Bulletin has his net approval at −19.1.

Every president who triggered a wave midterm loss saw independent approval fall below 40% before Election Day. Trump’s independent approval is at 25–34% depending on the pollster — deep into wave territory.

3. Special elections kept overperforming. Democrats have overperformed the partisan lean of every special election held since January 2025. The most notable: the Michigan state Senate District 35 special on May 5, where Democrat Chedrick Greene won a seat that kept the Michigan Senate in Democratic hands. The NJ-11 special on April 16 produced a progressive Democratic winner in a suburban district.

Special elections are the closest thing we have to real-time measurement of voter enthusiasm. And every data point has pointed in one direction.

4. The redistricting ceiling was established. Through April, the biggest counterargument to a Democratic wave was redistricting. Republicans were aggressively redrawing maps in Florida, Tennessee, Missouri, Alabama, and Virginia, while pushing for more in South Carolina, Indiana, and Louisiana. If all of these efforts succeeded, the GOP could bank 15–20 seats before a single vote was cast — potentially enough to survive a wave.

But the ceiling cracked. Indiana’s Senate rejected redistricting in December. The Virginia Supreme Court struck down the Democratic counter-gerrymander, but SCOTUS also rejected the Dems’ appeal, settling the map at the 2021 court-drawn version — not the maximally gerrymandered GOP version some had feared. And on May 27, South Carolina’s Senate killed the 7-0 Republican map. The redistricting advantage remains real (roughly +10 to +13 for Republicans), but it’s now fixed. It’s not growing. Markets can price in known quantities; it’s the unknown expansion risk that was holding Democratic odds down.

5. The Republican retirement wave broke records. As of May 2026, 36 House Republicans have announced their retirement — the most in a single cycle since records began. By comparison, 22 Democrats are retiring. Open seats are harder to defend. When the incumbents leaving include subcommittee chairs and members from competitive districts, it signals that insiders see the writing on the wall.

The Caveats

An 81% probability means Democrats lose the House roughly one time in five. That’s not trivial. Here’s what could close the gap:

The economy could improve. Wholesale gasoline futures dropped sharply in late May on Iran deal optimism. If retail gas prices fall below $4.00 before November, the economic pain narrative weakens. Trump’s worst polling is on prices and inflation (net −47 in the latest Strength In Numbers poll); any relief there directly improves Republican odds.

The generic ballot could tighten. It always does. In 2018, the final generic ballot was D+8.6, but Democrats won the popular vote by only 8.4 points. In 2010, Republicans outperformed late generic ballot numbers. There is almost always a reversion toward the party in power between May and November.

Republican cash advantage. The NRCC and RNC have amassed a significant financial edge over Democratic committees, with hundreds of millions in cash on hand. Money doesn’t win elections by itself, but in close races, a 2:1 spending advantage in the final weeks can move 1–2 points — which is the entire margin in dozens of districts.

Redistricting still matters. Even capped at +10 to +13, the Republican redistricting advantage means Democrats need to win the national popular vote by a larger margin than the raw numbers suggest. A D+6 generic ballot doesn’t translate to D+6 in every district. Geographic sorting and gerrymandering create a persistent structural bias that Democrats must overcome.

What the Smart Money Says

Prediction markets aren’t oracles. They got 2016 wrong. They were off on several 2022 races. But they aggregate information faster than polls, because traders have real money on the line and are incentivized to incorporate new data immediately.

The move from 45% to 81% over three months tells us something important: the information environment has shifted decisively. Not because of any single event, but because a pattern has hardened. The generic ballot didn’t waver. Trump’s approval didn’t recover. The special elections didn’t stop overperforming. And the redistricting expansion stopped.

At 81%, markets are saying: Democrats are heavy favorites, but a Republican hold is not impossible. It would require a significant shock — an economic turnaround, an October surprise, or a polling error — that isn’t currently visible.

Five months is an eternity in politics. But right now, the money is on blue.