South Carolina Senate Runoff Odds: Kalshi Beat The AP Call
Election nights have two clocks. One belongs to the networks and the Associated Press, which publish when they are sure. The other belongs to the order book, which publishes every minute and charges real money for being wrong. Tuesday's South Carolina Republican Senate special primary ran both clocks side by side, and the tape it left behind is the cleanest look yet at how an election market actually digests a race: nine days of grinding doubt, an afternoon of nervous spikes, one 527,000-contract hour that settled the first question, and a second market that refused to celebrate because it was pricing a harder one. We pulled the full hourly candle history from Kalshi for both markets, and this is the story the prices tell.
The Quick Answer
Sen. Darline Graham finished first in the South Carolina Senate special primary with about 33% of the vote at 90% counted, short of the 50% South Carolina requires, so she meets Rep. Ralph Norman in an August 25 runoff. Kalshi's first-place market traded from 73 cents to 98 in the hour after polls closed, before the AP's 8:20 p.m. ET projection, and the nominee market now prices the runoff at Graham 72 cents, Norman 28. The full hour-by-hour tape, including two 86-cent spikes the market faded before a single return was posted, is below.
Free: The Weekly PM Market Brief — the week's biggest prediction market moves, the stories behind them, and what's spiking next. One email, Sundays. Sign up in the box at the end of this article.
Nine Days Of Drift: 87 Cents To 61
The market opened on August 2 and needed about an hour to decide what the race was. The first prints on the "Will Darline Graham finish 1st?" contract ranged from a penny to 88 cents on all of 153 contracts, a thin book feeling for a level, and it closed that first hour at 87. The crowd's opening statement: the appointed senator, running for her late brother Lindsey Graham's seat with President Trump's endorsement, is a heavy favorite against a field that included Norman, Rep. Russell Fry, and former Gov. Mark Sanford.
Then the doubt arrived, a few cents at a time. The price bled to 84 by August 4 and the low 70s by the night of August 5. On the morning of August 6, one 5,245-contract hour knocked it from 73 to 67. By the weekend the drift had become a real repricing: the contract wicked to a 51-cent print on Saturday morning, August 8, its lowest trade since the opening hour's price discovery, before closing that hour back at 60, and it chopped between 56 and 64 through Sunday. On election eve it closed at 61. In nine days, without a poll closing or a vote counted, the market had cut Graham's first-place probability from 87% to 61%.
Keep that drift in mind, because the sharper signal was sitting in a different market the whole time. Kalshi also lists a nominee contract, the one that asks who actually wins the nomination rather than who leads round one, and all weekend it held Graham in the 42-to-45-cent band. The gap between the two prices was the market spelling out its actual thesis: she probably finishes first, and first probably isn't enough.
Election Day Before The Returns: A Climb And Two Spikes
Election morning opened at 61 and got busy immediately: nearly 24,000 contracts traded in the 8 a.m. hour and the price finished exactly where it started, a standoff at 61 with a dip to 60. The buyers won the hours that followed. The price climbed to 65 by 10 a.m. and 66 by 11, then a seven-cent jump in the hour before noon carried it to 73, and it touched 78 before 1 p.m. The afternoon gave back ten of the seventeen cents the climb had added, fading to 68 by 3 p.m., which is what a market looks like when everyone is guessing about turnout and nobody has information.
Except, twice, somebody acted like they did. In the 3-to-4 p.m. hour a burst of buying spiked the price to 86 cents before the book faded it back to 73, and in the 5-to-6 p.m. hour the same pattern repeated, another 86-cent print, another fade to 72. Those are the two prints I keep coming back to. Polls were still open. Whoever paid 86 was betting they knew something early, roughly three hours before the polls closed the first time and between one and two hours the second, and the market said no both times, pulling the price back to 73 by the 7:00 p.m. close of voting. Remember those spikes; the next hour would decide whether they were noise or a preview.
7 To 8 P.M.: Half The Market's Lifetime Volume In One Hour
They were a preview. The hour after polls closed, the first-place contract traded 526,746 contracts, 51.5% of the roughly 1.02 million it has traded in its entire life, and the price went 73, 70, 98. One hour, one question answered. For scale, the entire nine-day drift from 87 to 61 happened on a fraction of that volume.
Here is that night on both markets, using hour-ending Kalshi candles, all times Eastern:
| Time (Aug 11) | First-Place Market (Graham) | Nominee Market (Graham) | What Was Known |
|---|---|---|---|
| 7:00 P.m. | 73c | 49c | Polls just closed |
| 8:00 P.m. | 98c | 67c | Returns landing; no projection yet |
| 8:20 P.m. | ~98-99c | rising | AP projects Graham finishes first |
| 9:00 P.m. | 99c | 73c | Returns in; no second-place projection yet |
| 9:05 P.m. | 99c | ~73c | AP projects Norman second: runoff on |
| Overnight | 99c | 72-74c | Graham ~33%, Norman ~23.6% at 90% counted |
The 8:20 and 9:05 rows mark the AP projection times, which fall between candles; their prices are read from the adjacent candles, not separate prints.
The row that matters is the first one. At 7:00 p.m., as voting ended and before any projection, the crowd had Graham at 73 cents to lead and 49 cents, a coin flip, to be the nominee. By 8:00 p.m. the first-place question was priced at 98, and the AP's projection that Graham would finish first did not land until 8:20. The order book beat the wire by at least 20 minutes, and it did it the same way it always does, by paying up while the networks were still double-checking. The AP then projected Norman into second at 9:05 p.m., and with Graham short of the 50% plus one South Carolina requires, that confirmed the runoff, set for August 25. The nominee market had spent the previous two hours pricing exactly that outcome.
The Market That Refused To Celebrate
A 99-cent print looks like a party, but the nominee market never joined it, and that restraint is the most instructive part of the tape. While the first-place contract was sprinting to 98, the nominee contract touched 72 and closed at 67 in the hour after polls closed, then peaked at 77 between 8 and 9 p.m. on 246,000 contracts, its biggest hour since the market opened in mid-July, and settled in the low 70s. It sits at 72 cents this morning, with Norman's Yes at 28.
The reason is the same arithmetic the weekend prices were pointing at: Graham's 33% leads the field by 9.4 points but is nowhere near the 50% South Carolina requires, so the nomination now runs through a one-on-one with Norman on August 25, and a runoff is a genuinely different race. A Kalshi price in cents is just the crowd's probability with a dollar sign on it, so Norman's 28-cent Yes is the market handing him a 28% chance, about +257 in sportsbook terms, to flip a race he just lost by about nine points. Run Tuesday's arithmetic forward and 72/28 stops looking generous to either side. If Tuesday's coalitions simply revoted, Norman would need about 61% of everyone who voted for neither of them; Graham needs about 40%. The tape has already put prices on pieces of that question. Fry's and Sanford's nominee contracts, roughly 2.8 million lifetime contracts between them, finalized at a penny, which means the biggest bloc in this runoff, their combined 35% of Tuesday's vote, is now priced nowhere except inside the Graham-Norman spread. And the overnight fade from the 77 high back to the low 70s is the crowd applying its own small discount to two weeks of consolidation work. The verdict, for now, is 72/28. Our pre-election breakdown of this race walked through the field, and the steelman case for Norman we published before the primary is now, at 28 cents, the live counterargument. Our AI panel's verdict on the November race frames what the winner inherits.
One more honest detail: the first-place contract sits at 99, not 100, even though the race call is in and coverage has moved on. The contract settles on certified results, and certification does not happen on election night. That last penny is the price of paperwork, a small standing reminder that a market is never done until the thing it measures is.
A Worked Example: What The Poll-Close Coin Flip Paid
To make the night's repricing concrete, walk one position through it. At 7:00 p.m., a Yes contract on Graham winning the nomination cost 49 cents and pays $1 if she wins the runoff, so the crowd was pricing her nomination just under even money at the exact moment voting ended. By 9:00 p.m. that same contract was marked at 73 cents. On 100 contracts, a $49 position was carrying a paper gain of $24 two hours later, before Kalshi's trading fees and the bid-ask spread, real costs on both ends that get steeper in a fast market (our guide to how Kalshi trading works covers both). The first-place contract ran the same play with more force: 73 cents at poll close and 98 an hour later, 25 cents of repricing in sixty minutes. None of this is a trade to copy; the runoff is a new race and the easy repricing already happened. It is simply what it looks like, in dollars, when a market answers a question an hour before the official projection does.
What The Tape Teaches
Zoom back out and the night's real lesson is about contract wording, not speed. Kalshi asked two different questions about the same race, and traders who treated them as one bet paid for it: first place resolved to 98 cents within an hour of poll close, while the nomination, the thing a casual buyer probably thought they were trading, never cleared 77. The tape rewarded whoever kept those two probabilities separate all week, faded both 86-cent panics, and read 33% for what it was, a win that settled nothing. The market was not psychic. It was right twice before it had any right to be, and it argued itself back down both times. It was simply the only forecaster in the room publishing a correctable number every minute, and by 8:00 p.m. that habit had it 20 minutes in front of the wire.
The runoff market runs at 72/28 through August 25, and the tell worth watching is whether Graham's price moves on consolidation news, Fry and Sanford picking sides, endorsements landing, or on nothing at all, which would be its own answer about how settled this race already is. Our daily Kalshi picks page grades a fresh board every morning, our verdict-page calls are graded against real market settlements on a public scoreboard, and the 2026 midterms board prices what the winner walks into in November. The edge on Tuesday was never believing Graham would win; it was separating first place from the nomination before the returns forced everyone else to. Our analysts run that same discipline, the number over the narrative, every day on the free expert picks board.
Kalshi event contracts are CFTC-regulated and available to adults 18+ in eligible states as of August 2026; availability varies by state. All prices in this article come from Kalshi's hourly candle history for the two markets named, pulled August 12, 2026. Nothing here is advice about what anyone should trade or how anyone should vote.
