2026 Senate Control

The two regulated books no longer agree on who is favored. Nine days ago Kalshi, Polymarket and Manifold all priced Senate control at 54.5/45.5 to Republicans, and this workspace's read was that the absence of a spread was the story. That has broken.

Kalshi now prices a Republican hold at 52.5% and a Democratic takeover at 48.5%. Polymarket prices the same question at 49.5% Republican / 50.5% Democratic—it has crossed. The two venues are three points apart and on opposite sides of even money, which is the first genuine cross-exchange disagreement in this book all cycle.

Note both books also sum past 100%: Kalshi's two legs add to 101%, Polymarket's to 100%. On a strict two-outcome question that gap is the bid-ask, and at these spreads it is the same order of magnitude as the disagreement itself. Read the divergence as directional, not as an arbitrage.

Underneath, Michigan has moved seven points toward the Democrats and Texas has not moved at all.

Kalshi and Polymarket list the same binary D-versus-R Senate-control event and now disagree about which party is ahead: Kalshi 52.5/48.5 Republican, Polymarket 49.5/50.5 Democratic. Beside them, Kalshi's Range on the exact number of Democratic seats, which prices Above 52 as the single most likely bucket at 22.5% and carries $1.63M of open interest—the fatter-tailed read of the same question.

Manifold's play-money mirror was dropped from this stack this run. At $4 of twenty-four-hour volume against zero open interest it is no longer a third vote on the disagreement; its 51.1/48.9 print is a stale mark, not a price.

The Republican leg on both venues, priced directly—52.5% on Kalshi against 49.5% on Polymarket. The volume underneath the two is the mirror image of the prices: on Kalshi, where Republicans lead, the Democratic contract took $42,588 of the event's $71,089—60% of it. On Polymarket, where Democrats lead, the Republican contract took $26,775 of $27,999—96%.

Both books are taking most of their money against their own favorite. That is either two crowds converging on the same fade from opposite starting points, or one large position being worked on each venue.

The combo books price House and Senate jointly, and they are the sharpest thing on this page. Kalshi's Categorical carries $6.18M of open interest; Polymarket's carries $13.27M. Both agree closely on the shape: a Democratic sweep at 46.5% / 47.5%, a split Congress with a Republican Senate at 38.5% on both, a Republican sweep at 14.5% / 12.5%.

Then the fourth leg. R-House, D-Senate is priced at 1.2% on both venues—and on both it is the deepest contract of the day. Kalshi put $33,620 of $80,904 (42%) on it; Polymarket put $23,466 of $49,575 (47%). Two independent venues, the same rejected outcome, the same 1.2% price, and in each case more money than any of the three outcomes the market actually expects.

Read against the standalone Senate books, the combo pricing implies Democratic Senate control at 47.7% on Kalshi—within a point of its own 48.5% direct quote, so the two Kalshi books are internally consistent even as the cross-venue gap widens.

Both marquee states are past their primaries, so these are general-election books. Michigan has moved and Texas has not.

Michigan (SENATEMI-26, $169,804) now prices Abdul El-Sayed at 62.5% against Mike Rogers at 38.5%—in from 55.5/44.5 nine days ago, a seven-point move toward the Democrat with no primary or scheduled event behind it. Kalshi's second Michigan book, KXMISENATE-26 ($127,521, person-level rather than party-level), prices El-Sayed two points lower at 60.5% and still carries $98,638 of open interest on Haley Stevens at 0.5%, a candidate who lost the primary.

Texas (SENATETX-26, $154,021 against $6.07M of open interest, the deepest state book on the board) prices Ken Paxton against James Talarico at 50.5/49.5, unchanged and still the closest statewide Senate race listed anywhere.

• The convergence broke and it broke across even money. Kalshi 52.5% Republican, Polymarket 49.5%. The two venues now disagree about which party is favored to hold the Senate, after nine days of quarter-point agreement. That is the single most important change in this book this cycle.
• Both books are trading against their own favorite. 60% of Kalshi's flow bought the Democratic leg in a Republican-favored book; 96% of Polymarket's bought the Republican leg in a Democratic-favored one. If that persists, the two venues converge—from opposite directions.
• $57,086 across two exchanges on a 1.2% outcome. R-House, D-Senate is the deepest leg of the day on both combo books, priced at roughly one cent on each. Whatever explains it is not venue-specific.
• Michigan moved seven points on nothing. No primary, no scheduled event, no national move of comparable size. Either the state book is repricing a candidate or something is being unwound; the two Michigan books disagreeing by two points is the place to watch for which.
• Texas is the seat that decides it and it has not moved. 50.5/49.5 with $6.07M of open interest behind it. The chamber question is close because this one is.
• A defeated candidate still holds $98,638. Haley Stevens lost the Michigan primary and her leg is priced at 0.5% with six figures of open interest against it. Same reading trap as the closed rungs in the Fed ladders—check leg status before reading a Categorical's shape.