Settled. Prices as of settlement: Kalshi closed this market at 9:22 p.m. Eastern on Wednesday, September 9, 2026 and finalized it at 9:52 p.m., Matos Yes and every other contract No. The price history below runs through settlement. The model estimates were generated September 1, 2026 and were never re-scored; they are graded here as they stood.
Sabina Matos won the Rhode Island Democratic primary for lieutenant governor on Wednesday, September 9, 2026, with 37% of the vote in a five-way race, and Kalshi finalized its market that night at 9:52 p.m. Eastern with the incumbent's contract at Yes. Xay Khamsyvoravong, whose contract was the favorite at 52 cents when the polls closed, finished second at 26% and conceded in a statement sent by text. Sue AnderBois took 17%, Cindy Coyne 13% and Ross McCurdy just under 6%, per Rhode Island Current. The dead heat this page is named for never broke before the vote. In the small hours of primary day the incumbent was trading in the low 40s, a 500-contract buy at 50 cents at 9:52 a.m. only brought her back to even, and the panel of nine models that saw no prices had a median of 33% on her, 15 points behind the man she beat by 11. Below: the election-night tape, minute by minute, the grade for the market and for every seat, and the November race against Republican John Loughlin.
Kalshi listed a market on the Rhode Island lieutenant governor's race on August 31, and the exchange's opening read on it was a shrug. Two candidates at 48 cents each. Three candidates at two cents each. On a prediction market, a contract that costs 48 cents pays out a dollar if it comes true, so 48 cents is the market saying roughly "call it a coin flip between these two."
Rhode Island voted on Wednesday, September 9. Nine days of trading never resolved the coin flip. On the Monday before the vote both front-runners were bid at 46 cents, and when the polls closed at 8 p.m. Wednesday the challenger's last print, at 7:45, was 52 cents; the incumbent's last two, one contract each at 7:42, were 47 and then 42. Twenty-four minutes after the close, she was at 99.
We put the race in front of nine AI models and showed them everything except the price: the complete polling table, the campaign finance filings, the endorsements, the settled result from the last time this exact contest was run, and the public record on the candidates. Then we asked each one to distribute a hundred points of probability across the five names. Not one of the nine came back with a coin flip. Not one of the nine had the winner in front, either. This edition grades both the board and the panel against the count.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
New to markets like this one? Our guide to how election prediction markets work on Kalshi covers the mechanics before you read a board.
The Race Rhode Island Did Not Expect To Be Close
Sabina Matos has been lieutenant governor since 2021, when Dan McKee elevated her after he moved up to the governor's office. She won the job in her own right in 2022, taking 47.11% of a three-way Democratic primary against Deb Ruggiero and Cynthia Mendes, on 107,628 votes cast. Incumbency in a small blue state is normally the end of the story.
Then came 2023. Matos ran for the open seat in Rhode Island's 1st congressional district, and election officials in Jamestown, Newport and East Providence referred the signatures on her nomination papers to police. WPRI's Target 12 unit reported that the disputed papers included the name of at least one dead person, and Newport residents told the same outlet that signatures attributed to them were not theirs. East Providence city council members said the same. The state Board of Elections voted 5-1 to hand every one of her nomination papers to the attorney general while keeping her on the ballot, and later found no obvious pattern of fraud by the campaign itself.
Matos said she was "deeply sorry this happened" and blamed a vendor her campaign had hired to collect signatures. That vendor's operator, Holly McClaren of Harmony Solutions, later pleaded no contest to falsifying signatures and was sentenced to three years of unsupervised probation. Matos was never charged. Three years on, WPRI still frames her current race by it, describing her in August as "clinging to a slim lead as she tries to win another term following her scandal-tarred 2023 congressional campaign."
The challenger is Xay Khamsyvoravong, the former mayor of Newport. He has the money and an unusual share of the party's establishment. Through June 30 he had raised more than $117,000 in the quarter and held $333,000 in cash; Matos raised about $43,000 and carries nearly $98,000 in outstanding loans, per Rhode Island Current's review of the filings. U.S. Rep. Gabe Amo endorsed him over the incumbent, which members of a state's congressional delegation rarely do. So did Attorney General Peter Neronha. Matos holds the AFL-CIO, the building trades, two other unions, Pete Buttigieg and Latino Victory. The Rhode Island Democratic Party itself declined to endorse anyone.
Khamsyvoravong's own record carries one documented episode. As Newport mayor he accepted campaign contributions from developer Michael Schorsch and his affiliates; after Schorsch's company 1899 LLC won the Easton's Beach food service contract in April 2024, Newport This Week reported that "Khamsyvoravong said he donated $2,500 in Schorsch-related donations from 2023 to local nonprofits after 1899 won the beach food service contract."
Three other Democrats are on the ballot. Cindy Coyne is a retired state police lieutenant and former state senator; Sue AnderBois is a Providence city councilor with 11 state legislators behind her; Ross McCurdy ran as an independent for this office in 2022.
Here is the number that governed all of it. Across the three public polls of this primary, the largest single bloc was nobody: 42%, 43% and 48% undecided. Every candidate's polled support was smaller than the undecided share, and no newer poll was ever published. The largest bloc in this race had not chosen the last time anyone asked it. On Wednesday it chose, and it chose the incumbent: Matos went from 20% in the final poll to 37% at the finish, while the challenger went from 22% to 26%.
How It Settled
The count, the last poll, the market on the Monday before the vote, and the panel's median, side by side. A vote share is not a probability, so read the first column as what happened and the other three as what each instrument expected.
| Candidate | Vote share | Last poll (UNH, Aug. 20-24) | Market, Sept. 7 bid | Panel median | Kalshi result |
|---|---|---|---|---|---|
| Sabina Matos | 37.4% | 20% | 46¢ | 33% | Yes |
| Xay Khamsyvoravong | 26.4% | 22% | 46¢ | 48% | No |
| Sue AnderBois | 17.2% | 8% | 2¢ | 9% | No |
| Cindy Coyne | 13% | 5% | 2¢ | 7% | No |
| Ross McCurdy | under 6% | 2% | 0¢ | 2.5% | No |
The top three shares are NBC News's count at 2:22 a.m. Thursday with 96.8% of the expected vote in; Coyne's and McCurdy's are Rhode Island Current's figures as of 9 p.m. Wednesday. The one row that matters is the first. An incumbent the exchange had at 46 cents and the panel at 33 took 37% of the vote and won by 11 points, in a race that every instrument on this page had inside a coin flip or tilted the other way. The Associated Press sent its call at 8:50 p.m.; by then, as the next section shows, the market had been done for 26 minutes.
The Election-Night Tape
Wednesday was the second-busiest day this board ever had, 259 contracts behind the Saturday of the Coyne sweep, and the single biggest block of it was on the runner-up. Khamsyvoravong traded 10,022 contracts on September 9, more than half of everything he traded in the market's life, and printed his highest prices on the way in: 65 cents on 45 contracts at 12:01 a.m., 66 on 143 at 8:56 a.m., and the day's high, 67 cents, on 63 contracts at 11:55 a.m. Those were thin. The weight of his day sat at 52 cents, where takers bought three 1,000-contract lots at 6:23, 6:32 and 6:45 p.m. and another 1,619 at 7:45 p.m., 15 minutes before the polls closed. Matos opened the day at 41 cents on 47 contracts at 12:20 a.m., took a 500-contract buy at 50 cents at 9:52 a.m., and spent the evening mostly between 46 and 49. Her last two prints before the close, one contract each at 7:42, were 47 and then 42 cents, and her last fill over 50 contracts was a 57-lot sold at 43 cents at 7:21. So at 8 p.m., with the count about to start, the price history had the challenger at 52 and the incumbent somewhere between 42 and 48 on prints too small to mean much: closer than the panel ever put them, and still the wrong way round.
The first ten minutes after the polls closed said nothing. At 8:10 p.m. takers bought 400 Matos contracts at 48 and 49 cents, and at 8:11:42 Khamsyvoravong printed 57 cents on 18 contracts, the last time his contract traded above 50. Then the count arrived. At 8:13:34 p.m., in one second, Matos went from 49 cents to 84: 233 contracts at 49, 50 each at 53 and 58, ten at 80 and 20 at 84. Thirty-four seconds later the largest single fill of the night hit AnderBois, 4,431 contracts sold at 2 cents, and 11 seconds after that Khamsyvoravong traded at 18 and 26 cents. At 8:20:54 a taker bought 1,000 Matos at 65 cents; five seconds after that, 1,000 Khamsyvoravong were sold at 35. At 8:22:07 p.m. she ran from 71 cents to 98 in a single second. An 800-contract buy went through at 94 at 8:23:55, her first 99-cent prints landed at 8:24:05, and at 8:24:28 Khamsyvoravong traded at one cent. The market was effectively finished. The AP's alert came at 8:50, Rhode Island Current's count had her at 37% just before 9 p.m., Kalshi closed trading at 9:22 and finalized the contracts at 9:52. Khamsyvoravong conceded later that night.
Two things about that tape are worth keeping. The first is that the incumbent's contract, the one that paid, was the thinner of the two contracts the race was about, and second-thinnest of the five, ahead of only McCurdy: 13,895 contracts over its life against 32,069 on AnderBois, 18,240 on Khamsyvoravong, 16,823 on Coyne and 11,524 on McCurdy, 15 percent of the 92,551 traded. The second is the last print on Matos the night before the vote: 22 cents, on 120 contracts at 11:09 p.m. Tuesday, in the same second that 275 more sold at 40. Anyone who read that 22 as the market's opinion was reading a stray fill on an empty book, which is the same mistake this page warned about on September 1, when the front-runners' opening 47-48 quotes sat on zero traded contracts.
The Grade
Both the market and the panel went into the vote with the wrong favorite. The difference is by how much. From September 4 through 7 the exchange never had the two more than eight cents apart on a daily close, had the incumbent in front on two of those closes, September 4 and 5, and went into the vote with the challenger a few cents ahead; the panel separated them by 15 points, the same way. On the question this page is named for, whether the dead heat was real, the market was closer to right than the models: it refused to pick, and the candidate it would not pick against won.
The standard scoring rule for probability forecasts puts a number on that. Lower is better and a perfect call scores zero. The market's September 7 bids score 0.10 against the result. The panel's median scores 0.14, and not one of the nine seats beat the market. The best seat was Claude Opus, which finished at 39% on Matos, the highest number anyone gave her, and scores 0.12; the worst single call was Gemini's first-round 63% on Khamsyvoravong, which scores 0.18 before it revised to 48. The revision round cut both ways. The four seats that had opened farthest from the result, Gemini, Claude Fable, Claude Opus and Kimi, all improved by moving toward the middle. The four that had opened closest to it and then revised, Grok, GLM, Claude Sonnet and DeepSeek, all took points off Matos to move toward the consensus, and every one of them scored worse for it. GPT, which opened at 36% on Matos and declined to revise, kept the second-best number on the board by standing still. Consensus cost the seats that had it closest.
Three things in the September 1 analysis held up, and they are the ones to carry to the next board. Opus's finding that her polling was "flat 18/20/20", not collapsing, was the counterweight the 46 cents was paying for; she had led two of the three polls, and the undecided bloc broke her way. The 2022 precedent in the section below held. And the dark horses ran far above their 3-cent quotes, which the seats that raised AnderBois to 10% and 13% were reading correctly, even though it changed nothing about who won. What did not hold was the panel's central premise, that a three-to-one cash gap and the rarest endorsement in the state would decide a low-turnout primary. It did not. Matos bought no television, carried the AFL-CIO and the building trades, and had a super PAC's radio and mail behind her in the final week, and the largest bloc in the electorate went to the name it already knew.
What Nine Models Said When We Hid The Price
Each model was given the same data card and the same instruction: form one view of how the primary resolves, then split 100 points across the five contracts. None of them saw a market price at any point. Nine ran the first round; eight returned a revised number after reading the others' anonymized reasoning, and one declined to forecast on the revision round.
The Board As It Stood On September 7
| Outcome | Polling (UNH, Aug 20-24) | Market price | Panel |
|---|---|---|---|
| Xay Khamsyvoravong | 22% | 50¢ | 48% |
| Sabina Matos | 20% | 47¢ | 33% |
| Sue AnderBois | 8% | 3¢ | 9% |
| Cindy Coyne | 5% | 3¢ | 7% |
| Ross McCurdy | 2% | <1¢ | 2.5% |
The panel figure is the median across the eight seats that returned a revised number. The market column is the midpoint of the buyers' and sellers' prices as of September 7, rounded to the nearest cent, which is why it adds to more than 100. This table is the page's record of what the board looked like two days out; the result column is in the section above.
What Moved In The Final Week
This is the price history from the market's first week, as this page read it on September 7, kept as the record of what the board did before the count. The two front-runners had not traded a contract as of the page's 2:25 p.m. Eastern stamp on September 1; Matos's first 20 traded at 4:10 that afternoon.
| Outcome | Sept 1 quote (bid/ask) | Sept 4 close | Sept 7 quote (bid/ask) | Contracts traded Sept 4 to Sept 7 |
|---|---|---|---|---|
| Khamsyvoravong | 47 / 48 | 44¢ | 46 / 53 | 5,153 |
| Matos | 47 / 48 | 47¢ | 46 / 47 | 5,208 |
| AnderBois | 1 / 2 | 5¢ | 2 / 3 | 7,504 |
| Coyne | 1 / 2 | 1¢ | 2 / 3 | 13,657 |
| McCurdy | 1 / 2 | 1¢ | 0 / 1 | 7,090 |
Closes are the last trade of each Eastern calendar day from Kalshi's daily candles. The contract counts run from midnight Eastern on Friday, September 4 through 4:35 a.m. Eastern on Monday, September 7, and sum to 38,612. Five things in that table, then two that are not in it.
- The Front-Runners Finally Traded, And The Challenger Took The Harder Road. Khamsyvoravong's first print was 47¢ on Wednesday, September 2, at 8:42 a.m. ET. He closed 48¢ that day, then slid: a 42¢ close Thursday, and a Friday afternoon low of 40¢, where takers bought 2,400 contracts in two clips at 3:48 and 3:51 p.m. ET. He recovered to a 44¢ close Friday, ranged 39¢ to 43¢ on Saturday and closed 41¢, and then Sunday was his best day on price: he traded 1,381 contracts, his second-heaviest day behind Friday's 3,149, all but 15 of them bought by the taker, including a single 1,005-contract fill at 48¢ at 3:31 p.m. ET, and closed 48¢, his highest close since Wednesday. Sunday's 51¢ high was a five-contract print at 9:26 a.m. and its 41¢ low a ten-contract one a minute later; the weight of the day sat at 48¢. The book behind that close is thin. As of that Monday update he was 46 bid, 53 ask, seven cents wide, so the 50¢ midpoint on the board above was a quote, not a print.
- Matos's Busiest Day Was Saturday, And It Cut Both Ways. She traded 3,155 contracts on September 5, her most on any day, with takers split: 1,357 bought and 1,799 sold. She printed 51¢ at 1:58 p.m. ET on 247 contracts, her high since the market opened, and closed 49¢. Sunday gave it back on 154 contracts to a 46¢ close, and she sat at 46 bid, 47 ask.
- AnderBois Had Her Day. On Thursday, September 3, 9,058 contracts changed hands on her, 8,927 of them bought by the taker, starting with a 4,000-contract buy at 2¢ at 9:23:55 a.m. ET and 2,597 more 18 seconds later. By evening she had printed 15¢ twice, on 4.71 contracts at 9:14 p.m. ET and 6.26 more at 9:34, and she closed 13¢ on a 36-contract last trade at 10:49 p.m., after a 200-lot at the same price at 8:34. It did not hold: a 5¢ close Friday and 3¢ Saturday and Sunday, though the price history was not done with her. A 1,122-contract same-second cluster at 12:22 p.m. ET Sunday spanned 3¢ to 7¢ in one second, and a single 1,000-lot was bought at 3¢ at 2:09 a.m. Monday, her first four-figure buy since Thursday, matching the size of Saturday's largest fill, a 1,000-lot sold at 5¢. She was 2 bid, 3 ask that Monday. Her open interest had not unwound: 13,976 contracts, the largest on the board, ahead of Coyne's 11,994, so the Thursday position was mostly still open. Between Thursday's 13¢ close and Friday's open, the market carried her above the panel's 9%, the only stretch it ever did.
- Coyne's Number Is One Order. At 7:56 p.m. ET Saturday a taker bought 11,698 contracts at 2¢ in one same-second sweep of ten fills. That is 70% of the 16,674 contracts she has traded since the market opened, and it moved her quote from 1 bid, 2 ask to 2 bid, 4 ask at Saturday's close, and 2 bid, 3 ask by Monday. She traded twice more Saturday night, a 181-contract follow-on at 2¢ a minute after the sweep and an 839-contract sale at 9:49 p.m., and nothing on Sunday.
- McCurdy Went To Zero Bid. Takers sold 6,320 contracts at 1¢ between 9:48 and 9:49 p.m. ET Saturday. The bid was still zero on the Monday before the vote; the offer tightened from 2¢ at Saturday's close to 1¢ over Sunday, where it sat.
One pattern in that Friday's tape was worth flagging. Between 3:48 and 3:52 p.m. ET, takers bought both front-runners at once: 2,400 Khamsyvoravong at 40¢ and 1,271 Matos at 46¢. Paired, those cost 86¢ for a dollar if either wins, which is the same as pricing the three long shots at a combined 14¢ at a moment when their own books closed the day at 7¢ together. At 6:28 p.m. ET the shape recurred with the sides split: within 20 seconds, 749 Khamsyvoravong were bought between 40¢ and 44¢ while Matos, AnderBois, Coyne and McCurdy were all sold.
The thing that is not in the table is the lever the panel called "Late Money," which fired, and fired for the incumbent from outside her campaign. On Thursday, September 3, Khamsyvoravong accused Matos of relying on a super PAC, pointing to donations to the Democratic Lieutenant Governors Association from what he described as pharmaceutical companies, a private-prison company involved in ICE detention and Big Tech, and a $100,000 transfer from the association to DLP PAC, which paid for radio ads and mailers supporting Matos, Rhode Island Current reported, via Providence Business News. State disclosures put DLP PAC's spending in support of Matos at $92,035.81, per Newport Buzz. Matos's campaign manager, Christa Thompson, answered that Matos has not accepted corporate PAC donations directly. The same Rhode Island Current story tallied the closing television buys from FCC filings: about $122,000 for 381 spots for Khamsyvoravong, more than $96,000 for 349 spots for Coyne, more than $63,000 for 502 spots for AnderBois, and no television spending by Matos or McCurdy. No public poll was released after the University of New Hampshire survey in the board above; it stayed the newest through the vote.
What the week added up to was a market that had been thin and reactive rather than decided: the front-runners' prices were real by then but only a few thousand contracts deep, the dark horses moved on single orders, and the one fundamental that changed, outside money for the incumbent, showed up in the press before it showed up on the price history. None of it re-scored the panel. The seat numbers below are the September 1 run, and the price they argued with going into the vote was a 46-to-47-cent incumbent rather than a 48-cent one.
On Khamsyvoravong the panel and the market landed on the same number that Sunday: the seat median was 48%, and 48¢ is where his last contract traded on September 6. Six of the nine seats put him between 47% and 50%, and they got there for stated reasons that cancelled: a clear polling trend, the biggest bank account in the field and the rarest endorsement in the state pushing him up, against a plurality rule that lets a wounded incumbent survive on a committed base, two rivals pulling from his own anti-incumbent lane, and 42% of the electorate still unchosen. The two reasons pulling him down turned out to decide the race.
Everywhere else the panel and the market parted company. The models took 14 points off the incumbent's 47¢ and handed ten of them to the two candidates the board quoted at three cents. That was the whole disagreement, and the count settled it in the market's favor: the 14 points the panel took from Matos were 14 points it should not have taken.
More live boards from the same panel: Rhode Island governor primary, settled the same night: Helena Foulkes unseated Gov. Dan McKee, per Ocean State Media · California governor 2026, where Xavier Becerra sat at 96¢ when that price was fetched on September 7, 2026.
Every Seat's Number
| Seat | Khamsyvoravong | Matos | AnderBois | Coyne | McCurdy |
|---|---|---|---|---|---|
| GLM | 50% | 32% | 9% | 6% | 3% |
| Grok | 49% | 33% | 9% | 7% | 2% |
| Gemini | 48% | 34% | 10% | 6% | 2% |
| DeepSeek | 48% | 34% | 9% | 7% | 2% |
| Kimi | 48% | 33% | 9% | 7% | 3% |
| Claude Fable | 47% | 33% | 10% | 7% | 3% |
| Claude Opus | 44% | 39% | 8.5% | 6.5% | 2% |
| Claude Sonnet | 40% | 32% | 13% | 10% | 5% |
| GPT (First Round Only) | 39% | 36% | 11% | 10% | 4% |
| Panel (Seat Median) | 48% | 33% | 9% | 7% | 2.5% |
The GPT seat's first-round figures are shown for completeness and are not in the median. Read the Matos column against her 37.4% of the vote and the Khamsyvoravong column against his 26.4%, remembering that a share of the vote and a probability of winning are different quantities: the highest number any seat gave the winner was 39, and the lowest any seat gave the runner-up was 39.
Every seat on this panel is graded against real market settlements, this one included — records to date: Claude Fable 90% on 2,636 graded calls · Claude Opus 89% on 2,775 graded calls · Claude Sonnet 88% on 2,752 graded calls · GPT 86% on 9,450 graded calls · Gemini 87% on 8,358 graded calls · Grok 85% on 6,847 graded calls · GLM 85% on 3,796 graded calls · Kimi 87% on 4,139 graded calls · DeepSeek 84% on 3,911 graded calls. Recomputed daily; the full scoreboard is public.
- Every number here gets graded in public once the market settles, on the full graded scoreboard.
One more thing about that price, because it is the thread that runs from the first edition of this page to the settlement. When this page went up on September 1, as of that afternoon's 2:25 p.m. stamp, every one of the board's 5,775 traded contracts was on Coyne, AnderBois or McCurdy; the two front-runners had traded exactly zero, and their 47-bid, 48-ask was a market maker's quote sitting on an empty book. By the Monday before the vote the leaders had traded about 6,500 contracts each, and by settlement the board had traded 92,551, still with 65% of it on the three long shots. The crowd that set the dead heat was never more than a few thousand contracts deep, and the contract that paid was thinner than every contract but McCurdy's.
Where The Panel Changed Its Mind
The revision round moved the board hard. Gemini opened at 63% on Khamsyvoravong and came back at 48%. Claude Fable came down from 55% to 47%. The models that started most confident in the challenger all retreated toward the middle.
Claude Opus came down too, from 54% to 44%. What separated it was where it put the incumbent: it lifted Matos from 34% to 39%, the highest number any seat gave her, because it stopped taking one of the card's headline facts at face value.
That fact is the Emerson crosstab, and the panel split evenly on it. Helena Foulkes leads the governor primary 45% to 30%, and among Foulkes voters Khamsyvoravong leads Matos 27% to 14%. Four of the eight seats that revised read it as a clean structural advantage for the challenger: the bigger coalition prefers him. The other four multiplied it out.
"Run the card's own crosstab: Foulkes 45 × 27/14 gives Khamsyvoravong 12.2 and Matos 6.3; McKee 30 × 34/10 gives Matos 10.2 and Khamsyvoravong 3.0 — Matos 16.5, Khamsyvoravong 15.2 among governor-decided voters. Composition is net-neutral."
— Claude Opus
Claude Sonnet did the same arithmetic independently, then still trimmed Matos from 34% to 32%:
"Matos's McKee consolidation (24-pt margin) is proportionally stronger than Khamsyvoravong's Foulkes margin (13 pts), so this crosstab is closer to a wash than the clean tilt toward Khamsyvoravong that a naive reading suggests."
— Claude Sonnet
The multiplication checked out, and so did its direction. Among the voters who had already picked a candidate for governor, Matos was weaker inside the bigger coalition but gripped the smaller one far harder, and the two effects very nearly cancelled. Foulkes won the governor primary on Wednesday, and the lieutenant governor's race did not follow her coalition. Opus, the seat that did the arithmetic, finished with the best score on the board, as the grade above records; Sonnet, which did the same arithmetic and then still trimmed Matos from 34% to 32%, finished third. That is a case where working a poll finding through was worth more than agreeing about it.
The Argument For Matos The Panel Had To Answer
The strongest thing going for the incumbent is that all of this has been tried before, and it failed.
In June, the Rhode Island Association of Democratic City and Town Chairs endorsed Khamsyvoravong over Matos. That same organization endorsed Deb Ruggiero over Matos in 2022, and endorsed Nellie Gorbea for governor the same year. Matos won anyway with 47.11%. Gorbea finished third. The identical body, against the identical incumbent, in the identical office, went nought for one within living memory of every voter in this primary.
The models were made to test that. Most concluded the difference this time is money and a single well-funded challenger rather than the endorsement itself. Claude Opus laid out the check most explicitly, anchoring near an 85-90% renomination rate for sitting statewide officeholders and then asking which historical exceptions actually apply:
"Fresh unpriced scandal: PRESENT — the 2023 signature-vendor episode postdates her 2022 primary, she was uncharged, the vendor pled no contest, and WPRI still frames the race by it. Severe fundraising deficit: PRESENT and the strongest single fact. Elite/party defection: PARTIALLY present. Polling collapse: ABSENT — flat 18/20/20, led two of three polls."
— Claude Opus
That last line held. Matos had not collapsed. She led two of the three published polls. She was flat, not falling, and the count bore it out. The Association of Democratic City and Town Chairs is now two for two in endorsing against her in this office and nought for two in beating her.
Dark Horses The Panel Won't Dismiss
The market quoted Coyne and AnderBois at 2 bid, 3 ask apiece going in. The panel said seven and nine. For one evening the market went past it: AnderBois closed at 13¢ on September 3, with two thin prints at 15¢, before the book took it all back, as the price history above shows. In a five-way race decided by a plurality with no runoff, that gap deserves a sentence.
AnderBois is the one nearly every seat raised. Claude Sonnet put her at 13% and Claude Fable at 10%, on the theory that a Providence councilor with 11 state legislators behind her has an organized turnout base in exactly the kind of low-turnout September election where organization is worth more than name recognition. Roughly 17% of Rhode Island voters showed up for the 2022 state primaries. A candidate does not need a statewide following to win a plurality of a small, motivated electorate.
Coyne held the police, fire and government employee unions, a bloc that votes. GLM flagged the scenario worth watching: if AnderBois consolidated the anti-incumbent vote that Khamsyvoravong was counting on, she would not just help herself, she would fragment his coalition and help Matos. That is close to what the count shows. AnderBois took 17% and Coyne 13%, 30 points between them, and the challenger finished at 26%. The count fits GLM's scenario; nothing published yet shows whose voters the two of them took. Neither dark horse was ever near winning, which is the question the exchange's 3 cents was answering; the panel's 9% and 7% were reading something real about their organizations, which is a different question.
What Would Change The Panel's Mind
Four things the models named on September 1, each with the direction it pushed, and how each one answered.
- Any Poll Fielded After August 24. The newest public survey of this race is the University of New Hampshire poll, in the field August 20-24 and released August 27. Matos above 26% moves the board back toward her; Khamsyvoravong above 28% pushes him past 55%. Answered, September 9. None was ever released; the UNH survey stayed the last word; from that poll to the count Matos gained 17 points and Khamsyvoravong four.
- Early-Vote Composition. In-person early voting ran from August 20 through 4 p.m. on September 8. Returns concentrated in Providence-area or Foulkes-leaning municipalities favor Khamsyvoravong; returns weighted toward labor-heavy or McKee-aligned precincts favor Matos. Answered, September 9. No public breakdown of the early vote by municipality appeared before the count, and the count made the question moot.
- Late Money. Third-quarter filings or independent spending that closes the gap raises Matos by several points, in the panel's read. Khamsyvoravong's $333,000 in cash on hand, set against an incumbent who raised about $43,000 in the quarter and carries nearly $98,000 in loans, was the most-cited finance fact across all nine seats. Note that those are different measures: the filings give his cash on hand and her quarterly total, and Rhode Island Current's writeup does not state hers side by side. Answered, September 9. It fired, September 3 to 4, in the direction the panel named, and the panel was never re-scored for it. Whether the DLP PAC's $92,000 of radio and mail was the margin is not something the count can say; the margin was 11 points.
- Whether 2023 Is Re-Litigated On Air This Week. Several seats flagged that the signature episode was, by then, old news rather than an active attack. If it returns to television in the closing days, it moves against Matos; if it stays quiet, its effect is already in her flat-but-stable 20%. Answered, September 9. FCC filings showed Matos bought no television at all, so any closing air war was her opponents' to run, and what their spots said is not in this page's record.
When This Settled
| Polls Closed | 8:00 p.m. Eastern, Wednesday, September 9, 2026 |
| First 99¢ Print On Matos | 8:24 p.m. |
| Associated Press Call | 8:50 p.m. |
| Khamsyvoravong Concedes | Wednesday night, after the 9 p.m. count, in a statement sent by text |
| Kalshi Closes Trading | 9:22 p.m. |
| Contracts Finalized | 9:52 p.m.: Matos Yes; Khamsyvoravong, AnderBois, Coyne and McCurdy No |
| General Election | Tuesday, November 3, against Republican John J. Loughlin II, who was unopposed in his primary |
The contract's rule was one sentence: it resolved Yes if the named candidate "wins the nomination for the Democratic Party to contest the 2026 Rhode Island Lieutenant Governorship," and it was eligible for what Kalshi calls accelerated determination "after a consensus of media organizations project the winner." That consensus formed inside an hour of the polls closing. The AP's projection, which The Boston Globe carried with Matos's statement that this was "a victory of the voices of working people, who are the backbone of the Democratic party, over the backroom insiders, the wealthy donors, and the corporate interests who want to keep themselves at the top," came 26 minutes after the price history had already settled the question at 99 cents. The listed November 3 expiration was only ever an outer bound.
One quirk worth pinning down, because several reference sites will keep printing the wrong date: this primary was moved. It was set for Tuesday, September 8, until Gov. Dan McKee signed a bill on April 20 pushing it back a day. Labor Day falls on Monday, September 7 in 2026, and more than 430 of Rhode Island's polling places sit in schools, churches and town halls that are closed on the holiday, leaving no time to set them up, per the Providence Journal. The state had last voted on a Wednesday in 2018.
This page is settled. The prices on it are final and the panel's numbers are graded.
The Bottom Line
The market opened this board at a coin flip between an incumbent and a challenger, and nine days of trading never moved it off one. Nine models given the same facts and none of the prices kept the challenger at the market's number and marked the incumbent 15 points below it. Then Rhode Island voted, and the flat-polling incumbent, outraised by her main challenger, collected the undecided bloc and won by 11. The exchange had the wrong favorite by a few cents; the panel by 15 points. The dead heat this page is named for was the more honest number of the two, and the seat that came closest to the truth was the one that did the arithmetic and moved the way the arithmetic pointed.
Every number above is now graded in public, and the same nine seats will price the next board with this one on their record.
Hero illustration: OddsShopper, in the house collage style. Sabina Matos photo by Kenneth C. Zirkel, licensed CC BY-SA 4.0; photos cropped, toned, and composited.
More on this: 2026 Governor Races: Every Kalshi Board We Track · Rhode Island Governor Primary: McKee Says 'I Am the State · Why Jack Reed Will Win The Rhode Island Democratic Senate Primary · Why Raymond McKay (R) Will Win the Rhode Island Senate Primary · Iowa Governor Odds 2026: The Man Who Beat Trump's Pick
