Three Texas Senate Polls in Three Days Put Talarico Ahead, and the Market Prices the Lead but Not the Size

A Democrat is priced to win a United States Senate seat from Texas, and the three polls released between September 21 and September 23, 2026 explain why the price is in the low sixties rather than the eighties. As of 4:35 p.m. Central on September 23, 2026, across two venues, Kalshi priced the Democratic contract, which both exchanges label with James Talarico's name, at 59% and Polymarket at 63%, against 43% and 39% for the Republican side. The three surveys put Talarico ahead of Ken Paxton by six, one and three points. A lead that shows up in every poll and is small in each is what a price near 61% looks like: the market believes the direction and discounts the size.

What the number says

The two books sit four points apart, 59% on Kalshi and 63% on Polymarket as of September 23, 2026, which is a normal spread for a contract with this much money in it and not a disagreement about the race. Prediction Genius, which pairs the venues, showed $6.6M traded on the Democratic outcome and $8.9M on the Republican one, so this is the deepest board on the Texas site by a wide margin, and the price is not a thin book being pushed by a few orders. A near-61% price says a Democratic win is the likelier outcome and that a Republican win would not be a surprise, which is roughly what the polls say when read together.

Read one at a time, the polls say different things. Marist, published September 23, 2026 from interviews of 1,139 registered voters conducted September 17 to 20 with a margin of error of 4.4 points, had Talarico at 50% and Paxton at 44%. Texas Southern University's Barbara Jordan Public Policy Research and Survey Center, released September 22, 2026 from 1,800 likely voters surveyed September 15 to 19 with a margin of 2.31 points, had Talarico at 47%, Paxton at 46%, the Libertarian Ted Brown at 3% and 4% undecided. The NBC5 Dallas and Telemundo poll published September 21, 2026, from 625 likely voters surveyed September 8 to 10 with a margin of 4 points, had Talarico at 46% and Paxton at 43%. What the three share is the direction; what they do not share is the size, and the size tracks the sample, which means registered voters give Talarico six and likely voters give him one or three.

Marist asked a second question that cuts the other way. Regardless of whom they supported, 52% of its respondents said they expected Paxton to win, against a horse race that had Talarico ahead. A price near 61% for the Democrat on September 23, 2026 sits between those two answers: closer to the preference than to the expectation, but not all the way to either, which is the position of a market that treats a six-point registered-voter lead as real and a one-point likely-voter lead as the number that will matter in November.

Why: where the one-point race lives

The Texas Southern survey is the one that explains the shape of the price, because it shows where the Republican side's missing points are, which is the part of the poll the topline hides. Ninety-nine percent of Democratic likely voters told the center they would vote for Talarico and none for Paxton; 84% of Republican likely voters said they would vote for Paxton and 7% for Talarico. The center put 15% of Texas Republicans in a group it labels moderate Republicans, and among them Paxton drew 59% and Talarico 27%, against 98% and 0% among the voters it labels MAGA Republicans. Marist's independents broke 55% for Talarico. A one-point race in a likely-voter sample is therefore not a state split down the middle; it is a Democratic coalition fully consolidated against a Republican one that has not finished consolidating, and the price is a bet on how much of that 16% comes home by November 3, 2026.

That framing also explains why the price is not higher. If the Republican voters who are not yet with Paxton return at anything like their usual rate, the one-point race becomes a Paxton lead, and every poll in the set is within reach of that outcome. Marist's six points is outside its own margin of error by a little more than a point; the other two leads are inside theirs. A 61% price is what a market pays for a lead it can see in every survey and cannot rule out in any of them.

The case against this read

The strongest case against this read is that the market has priced a registered-voter lead into a likely-voter election. The two likely-voter polls in the set, Texas Southern's and NBC5's, have the race at one and three points, both inside their margins, and the professional raters are split on which way the seat leans: Ballotpedia's summary on September 23, 2026 listed the Cook Political Report, Decision Desk and Sabato's Crystal Ball at toss-up and Inside Elections at lean Republican. A reader who holds that view would say the honest price is nearer 50% than 61%, and that the market has let the largest lead in the set, which is also the one from the loosest screen, do too much of the work.

There is a second objection, about what could shift the number between now and the election, and it is that very little is scheduled to. Talarico accepted three proposed televised debates, the first set for September 22, 2026 in the Rio Grande Valley and the others for October 8 in Dallas and October 13 in Austin, according to the Austin American-Statesman's August 12, 2026 account of the invitations. WFAA reported on the evening of September 22, 2026 that the Paxton campaign had not responded to its messages about a debate since August 24 and that a debate now appeared unlikely. A race without a debate is a race without the one scheduled event that reaches undecided voters at once, which leaves the price to be pushed around by polls and by the money the campaigns spend, and it is a fair objection that a market with nothing on the calendar to move it can sit at 61% out of inertia as easily as conviction.

What would change it

Catalysts: the voter registration deadline of October 5, 2026, the two proposed October debates if the Paxton campaign agrees to either, the start of early voting on October 19, 2026, and every likely-voter survey published between now and then. Invalidation: this read, that 61% is the price of a small lead that appears in every poll, is wrong if the next likely-voter surveys put Paxton ahead and the Democratic contract holds above 55% on both books anyway, because at that point the market would be pricing something the polls are not showing. It is also wrong in the other direction if the contract trades above 70% before October 19, 2026 without a likely-voter poll putting Talarico ahead outside its margin, because that would be a market running ahead of its evidence.

The number to watch is not the topline but the Republican share of Republicans. The Texas Southern center found 84% of its Republican likely voters with Paxton; a later survey from the same center that puts that figure above 90% would erase the one-point lead on its own, and a figure that stays in the low eighties into October would justify a price higher than the one the exchanges show.

What is next

The next fixed date is October 5, 2026, the last day to register to vote in the November 3 election under the Texas Secretary of State's calendar, and early voting runs from October 19 to October 30. The contract on both venues resolves on the party of the winner, once the Associated Press, Fox News and NBC have called the race for the same candidate or, failing that, on the certified result.