Which party will control the U.S. House of Representatives after the November 3, 2026 midterm elections?

The tape is not close. Polymarket sits near 82% on ~$8.1M, Kalshi 84-86%, Metaculus 80-85%. Deep money says Democrats. But the same money knows the map, so the win it prices is a squeaker, not a rout — I take the modal 48-58% flip and leave the wave to the dreamers.

The crowd anchors on a narrow win because a narrow win feels safe, yet the structure whispers otherwise. Every one of the four modern flips arrived as a wave, unpopular presidents surrender ~37 seats on average, and the retirement ledger runs ~36 Republican open seats against ~22. Fold those together and the wave is no fringe — it holds 20-30% of the field, the fat tail the tape underweights.

Count them across the long roll of midterms: 20 of 23 since 1934 the president's party has bled, and every escape belonged to a beloved man sitting near two-thirds approval. This president sits far below, at the wrong end of the ledger, so the pattern turns against him as it has turned against nearly all before. Yet the map holds an old and stubborn line, and so I read not a flood but a firm crossing — the modal flip, 48-58%.

The mechanism is plain. Democrats need three net seats. The cushion is two or three. Eighteen toss-ups sit on the table and the environment leans their way. But break-even runs at D+4-5, so the same edge that wins caps the surplus. The route delivers a majority, not a landslide — 48-58%.

Numbers bow, in the end, to will. A thin majority is not weakness but concentration — the defenders draw close around the handful of seats that decide everything, and a party that knows exactly which doors to hold can hold them against a wider tide. Redistricting variance breaks their way at the margin, and the map's old arithmetic stands guard. I back the hold at its honest 14-19% — the underdog with the firmest grip on the few gates that matter.

Everyone has already spent Election Night — Democrats win, film at eleven. That is precisely when the map arranges its little joke. A D+4 edge and a D+4-5 break-even line is a recipe for a dead heat, and history obliges: roughly two of the last dozen midterms left control genuinely unknown at a two-or-three-seat margin. Priced at 6-10%, the near-draw is the live thing nobody wants to hold.

Read the shape of the thing, not the tape. Four times the modern republic has turned on a hated midterm president, and four times out of four the turning came as a flood — 1994, 2006, 2010, 2018, no exceptions, a rhythm too clean to be accident. A man at minus-seventeen, the ledger promising ~37 seats, thirty-six of his own party's doors standing open: this is a story that ends in a wave, and the ending is already written. The odds say 20-30%; the omens say it is fated.