South Carolina Senate Odds: Graham, Fry and Tuesday's Primary
South Carolina votes Tuesday in the special Republican primary for the Senate seat Lindsey Graham held for 23 years, and Kalshi's market on the winner has traded roughly $4.8 million, one of the largest special-election books the exchange has listed. The board prices Darline Graham Nordone, the late senator's sister, at 48 cents, with two sitting congressmen, Ralph Norman and Russell Fry, tied at 25. We handed the full 11-name partition to eight price-blind AI models with a fetched file of every published poll, the certified ballot, and the runoff rules. The panel's read: the favorite is real but overpriced, the value sits one runoff away, and half the names on the board cannot win at all.
The Quick Answer
The panel blends Darline Graham at 42.5% against a 48-cent price, a five-point trim of the market's favorite, and makes Russell Fry the board's only positive gap at 27% against 25 cents, with Ralph Norman at 23.5%. The reason is the two-stage math: nobody polls near the 50% an outright Tuesday win requires, and the one published runoff head-to-head shows Fry beating Graham by nine while Graham beats Norman by five. Whoever grabs the second runoff slot decides this market. The full board, the seat-by-seat numbers, and the six dead tickets still trading are below.
These are model estimates, not predictions of fact and not financial advice, and this page takes no side in any race; it analyzes prices. Kalshi is a CFTC-regulated event-contract exchange; you must be 18 or older and in an eligible state to participate, and availability varies by state as of August 2026. New to event markets? Start with how election markets work.
Free: The Weekly PM Market Brief — the AI panel's graded record, the week's biggest market-vs-model gaps, and what's spiking next. One email, Sundays. The signup box is at the bottom of this page, or just keep reading: the graded record it promises is printed under the board.
Prices on this page refresh through primary night and any August 25 runoff; every figure carries its fetch date. Prices as of August 7, 2026. This is the permanent home for this race: the primary board now, the general election below once the nominee exists.
The Board
| Outcome | Market Price | AI Blend | Gap (blend% − mkt¢) |
|---|---|---|---|
| Darline Graham Nordone | 48¢ | 42.5% | −5.5 |
| Russell Fry | 25¢ | 27.0% | +2.0 |
| Ralph Norman | 25¢ | 23.5% | −1.5 |
| Mark Sanford | 4¢ | 4.0% | 0.0 |
| Mark Lynch | 1¢ | 1.6% | +0.6 |
| Another ballot-certified candidate (Buckner, McBride, Parker, Shepherd) | not a listed contract | 1.2% | — |
| Nikki Haley, Nancy Mace, Trey Gowdy, Scott Bessent, Henry McMaster, Pamela Evette | 1¢ each | 0.0% | — |
These are model estimates, not predictions of fact and not financial advice. That last row is not a rounding artifact; it is a warning, explained below.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $168M traded
- 2028 U.S. Presidential Election winner? · $57M traded
- 2027 Pro Football Champion · $54M traded
- Pro Baseball Champion · $45M traded
- More tech layoffs in 2026 than in 2025? · $31M traded
- Bitcoin price at the end of 2026 · $30M traded
Every market above links to our full AI model verdict; browse them all on the OddsShopper prediction markets hub, and see how every settled call actually scored on the full graded scoreboard.
The Market At A Glance
| Venue | Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of August 2026) |
| The Event | "South Carolina Republican Senate special primary winner?" (Kalshi event KXSCRSENS-26), 11 person contracts, roughly $4.8 million traded |
| The Contract | Each name reads: "If [person] wins the 2026 South Carolina Republican Senate special primary, then the market resolves to Yes." Markets are eligible for accelerated determination once a consensus of media organizations project the winner. |
| The Two-Stage Rule | South Carolina primaries require a majority. If nobody clears 50% on August 11, the top two meet in an August 25 runoff, and the runoff winner is the primary winner. |
| Settles | On the projected winner: August 11 if someone clears a majority, August 25 if it goes to the runoff every poll implies |
| Prices As Of | August 7, 2026 (candlestick closes) |
What The Models Were Given
The card carried the race's whole verified structure, dated and sourced. Sen. Lindsey Graham died suddenly on July 11 after 23 years in office. His sister, Darline Graham Nordone, 62, was sworn in to serve out the remainder of his term, entered the special primary in her political debut, and received President Trump's endorsement on July 17. The certified ballot has exactly nine names: Duke Buckner, Russell Fry, Darline Graham Nordone, Mark Lynch, Mark McBride, Ralph Norman, Glenda Gail Parker, Mark Sanford, and Samuel Shepherd; twelve filed and three were decertified. Fry is the sitting congressman from the 7th District, Norman from the 5th; Sanford is a former governor. Three late-July polls disagree about the order: Emerson/Nexstar has Norman 22, Graham 19, Fry 12; co/efficient has Graham 30, Fry 16, Norman 15; Trafalgar has Graham 27 with Norman and Fry at 17. Every poll leaves a fifth to a quarter of voters undecided, nobody near a majority. And the sharpest instrument in the file: co/efficient's runoff head-to-heads, where Fry beats Graham by nine while Graham beats Norman by five. The eventual nominee faces Democrat Annie Andrews on the regular November 3 ballot for the seat's next term.
Every Seat's Number
The blend is the equal-weight mean of the eight models' round-two numbers, generated after each model read the others' anonymized round-one reasoning and could revise. All eight seats returned in both rounds.
| Outcome | Market | Fable | Opus | Sonnet | GPT | Gemini | Kimi | GLM | DeepSeek | Blend |
|---|---|---|---|---|---|---|---|---|---|---|
| Darline Graham Nordone | 48¢ | 47.0 | 51.0 | 38.0 | 40.0 | 45.0 | 40.0 | 44.0 | 35.4 | 42.5% |
| Russell Fry | 25¢ | 25.0 | 23.0 | 27.0 | 28.0 | 27.0 | 26.0 | 27.0 | 33.3 | 27.0% |
| Ralph Norman | 25¢ | 22.0 | 21.0 | 23.0 | 24.0 | 23.0 | 25.0 | 22.0 | 28.3 | 23.5% |
| Mark Sanford | 4¢ | 3.0 | 3.0 | 8.0 | 5.0 | 3.0 | 4.0 | 4.0 | 2.0 | 4.0% |
| Mark Lynch | 1¢ | 2.0 | 1.0 | 2.0 | 2.0 | 1.0 | 3.0 | 1.0 | 1.0 | 1.6% |
| Another certified candidate | not listed | 1.0 | 1.0 | 2.0 | 1.0 | 1.0 | 2.0 | 2.0 | 0.0 | 1.2% |
| The six names not on the ballot | 1¢ each | 0 | 0.0 | 0 | 0.0 | 0 | 0 | 0 | 0 | 0.0% |
These are model estimates, not predictions of fact and not financial advice. A 0 means the seat priced the outcome at zero. The panel: Claude Fable, Claude Opus, Claude Sonnet, ChatGPT, Gemini Pro, Kimi, GLM, and DeepSeek, each pricing the full partition in one pass on August 7, 2026, without seeing any prices.
Every seat on this panel is graded against real market settlements — records to date: Claude Fable 84% on 334 graded calls · Claude Opus 85% on 385 graded calls · Claude Sonnet 83% on 366 graded calls · GPT 83% on 2,013 graded calls · Gemini 85% on 2,062 graded calls · Kimi 83% on 2,099 graded calls · GLM 81% on 2,138 graded calls · DeepSeek 79% on 2,145 graded calls. Recomputed daily; the full scoreboard is public.
More live boards from the same panel: New Hampshire Senate 2026 · Democratic side 88¢ · Maine Senate 2026 · Collins 37¢ · Michigan Senate 2026 · El-Sayed 58¢ · 2026 Senate map hub
Six Names That Cannot Win: Read the Ballot Before the Board
The board's strangest feature is that Nikki Haley, Nancy Mace, Trey Gowdy, Scott Bessent, Henry McMaster, and Pamela Evette all still carry one-cent contracts, and none of them filed for this race. The certified ballot was set at nine names in late July; a person not on it cannot win the primary. Kimi stated the panel's unanimous position: they are "structurally near zero, not longshots — they cannot win a primary they are not on the ballot for." Those six contracts exist because the exchange listed the board before filing closed, and they are the clearest example in any current market of why the first read on an election board is the ballot, not the prices.
Darline Graham: A Real Favorite, Priced Like a Safer One
Nobody on the panel disputed that Graham is the favorite; all eight seats ranked her first. She leads two of three polls, carries the endorsement that historically dominates South Carolina Republican primaries, and holds the seat's surname in a low-information special election. The five-cent trim is entirely about the second stage. Claude Opus, her highest seat at 51, put the two-stage shape best: the endorsement and name make her "the near-lock for a runoff slot," but 50% on August 11 in a nine-name field is unlikely, and a runoff is a different election. Claude Fable's summary carried the panel's core caution: the endorsement arrived "only July 17 for a political debutante in a nine-way field against sitting congressmen with geographic bases, and the sharpest data point on the board says the endorsed candidate loses at least one plausible runoff pairing."
Russell Fry: The Narrow Gate and the Wide Path
Fry polls third in two of three surveys, and the panel still makes him the board's only value at 27% against 25 cents. The mechanism is the inversion several seats named independently: Norman is likelier to reach the runoff, but Fry is likelier to win one. The lone published runoff head-to-head has Fry ahead of Graham by nine points, the only direct second-round evidence anyone has, and it points at the endorsed favorite losing a consolidation ballot. Gemini Pro compressed the whole trade into one sentence: Fry's "real fight is edging Norman for the second runoff slot." Get through that gate, and the data says the path widens.
Ralph Norman: The Mirror Image
Norman leads the Emerson/Nexstar poll at 22, the survey Kimi's seat called the field's most reliable, holds the 5th District base, and prices identically to Fry at 25 cents. The panel puts him 3.5 points lower because his shape is the mirror of Fry's: the best odds of making the runoff, and the published head-to-head shows him losing it to Graham by five. Same price, opposite geometry, and the panel thinks the market is paying for first-round strength that stops mattering on August 12.
Dark Horses The Panel Won't Dismiss
- Mark Sanford (4.0% blend, 4¢). The former governor polls fourth everywhere; Claude Sonnet carried the panel's high number at 8 on residual name recognition, while noting his political baggage caps the runoff upside.
- Mark Lynch (1.6%, 1¢). The Greenville businessman made the five-candidate debate stage; a self-funder's late surge in a low-turnout special is the scenario, with no polling evidence behind it.
- The unlisted four (1.2%). Buckner, McBride, Parker, and Shepherd are on the ballot but have no contract; a win by any of them would resolve every listed contract No.
Where The Panel Changed Its Mind
The revision round tightened the Fry trade rather than the favorite. Kimi flipped its second and third places, moving Fry above Norman after six peers "converged on the inversion point — Norman's higher probability of making the runoff is offset or more by Fry's superior runoff matchup." Gemini Pro moved the other direction on the same evidence, cutting Fry and raising Norman after peers flagged that it had "over-indexed on Fry's strong runoff head-to-head numbers" without discounting for his gate risk. Claude Sonnet raised Graham from 33 to 38, conceding it had underweighted how safely she reaches the runoff, and GLM cut her from 47 to 44, persuaded that in a two-person consolidation ballot "the head-to-head data — not first-round plurality — is the more predictive signal." DeepSeek held its 35 and said the endorsement argument "does not move me" because its conditions were already priced.
What Would Change The Panel's Mind
- Tuesday's first-round margins. The single biggest repricer: the identity and gap of the top two rewrites every conditional above.
- Any post-August 1 poll. Everything in the card predates August 1; the final week is unmeasured, and the panel said so rather than pretending precision.
- An outright majority Tuesday. Roughly a one-in-ten scenario on the panel's math; it would settle the market two weeks early.
- A second runoff head-to-head from any pollster. The Fry-beats-Graham number is the board's load-bearing datum; one corroborating or contradicting poll moves the whole trade.
Settlement Timeline
| Date | What Happens |
|---|---|
| August 11, 2026 | Special primary day; market settles now only if someone clears 50% |
| August 25, 2026 | Runoff between the top two if no majority; the runoff winner is the primary winner |
| November 3, 2026 | The nominee faces Democrat Annie Andrews in the general election (a separate market question this page will carry) |
| January 2027 | November's winner is sworn in for the seat's next term |
This page re-scores after Tuesday's results and again after any runoff; prices and blends above are as of August 7, 2026.
How To Read These Prices
Buy YES on Fry at 25 cents and you are paying 3-to-1 odds on a candidate the panel makes 27%, a two-point edge that lives or dies on one number: whether he beats Norman to the second runoff slot. The Graham trade is the reverse: her 48 cents prices a 52% miss chance, while the panel says 57.5, and the difference is the head-to-head everyone else is ignoring. Two-stage markets punish reading only the first stage, the same way our Michigan primary-day board rewarded reading the electorate over the name. Every number in this piece gets graded in public once the market settles, most likely within eighteen days; the model verdict scoreboard carries the running record. These are model estimates, not predictions of fact and not financial advice.
The Bottom Line
Four days out, the panel agrees with the market's order and disagrees with its confidence: Graham is the favorite at 42.5%, not 48; Fry's 25 cents is the only price on the board the panel would rather own than sell; Norman's identical 25 cents buys the weaker half of the runoff geometry; and six famous names are trading above their true worth of zero. Tuesday night answers the only question that matters, who gets the second slot, and this page re-scores the moment it does. To be explicit one final time: these are model estimates, not predictions of fact and not financial advice, and this page endorses no candidate. 18+, eligible states only.
Verdicts generated August 7, 2026, price-blind, revised once after an anonymized cross-read. Prices as of August 7, 2026 (Kalshi candlesticks). Panel: Claude Fable, Claude Opus, Claude Sonnet, ChatGPT, Gemini Pro, Kimi, GLM, DeepSeek.
