Georgia Senate 2026 Odds, And The Case Hiding Inside Mike Collins's 7.8 Cents
Prices as of Friday, August 21, 2026. Markets move — and the thin ones move on very little. Check the live number before you act on anything here.
Ten days ago, both major prediction markets had Francesca Hong at roughly 95% to win the Wisconsin Democratic primary for governor.
Polymarket posted the 96% figure to X. Then deleted it. Because David Crowley won — by about half a point, in a race the AP couldn't call until Wednesday morning.
Kalshi co-founder Luana Lopes Lara had a response I haven't been able to stop thinking about since: "5% is not 0%."
That was August 11. Same two exchanges. Same month. Same shape of confidence. (For what it's worth, the delay wasn't genuine suspense — Milwaukee mis-uploaded about 28,000 absentee ballots, Hong conceded around 11:30 p.m., and the AP called it at 2:34 a.m. The market was wrong well before the count was slow.)
So let's talk about Georgia, where Jon Ossoff is trading at 92.5¢ on Kalshi and 93¢ on Polymarket, and where every piece of commentary you'll read this week describes the race as effectively finished.
Maybe it is. The case is strong and I'm going to make it honestly. But "93 cents" and "lock" are not the same sentence, and the difference between them is where the actual thinking lives.
First, Let's Translate 7.8 Cents Into English
Mike Collins is 7.8¢ on Kalshi, 7.6¢ on Polymarket.
In sportsbook terms, 7.8% implied probability is roughly +1180. Call it a 12-to-1 shot — a 93% favorite loses about one time in fourteen.
Here's the comparison that made it click for me. In June 2016, with Golden State up 3-1 on Cleveland in the Finals, the Cavaliers were +1100 — an 8.3% implied chance. Slightly shorter than what Collins is getting right now.
LeBron went and won that series — still the only 3-1 comeback in Finals history. But it happened, at a price a hair shorter than the one Georgia Republicans are staring at today.
Or pick your poison from the historical file. A 7.8% shot is a touch better than a 15-seed's all-time record against a 2-seed in the NCAA tournament — 11-153, about 6.7%, since 1985. Fifteens win. Not often. But if you watch enough March, you see it.
None of that is an argument that Collins wins. It's an argument that 7.8¢ is a real number describing a real possibility, and the word "lock" is doing something dishonest to it.
Free: The Weekly PM Market Brief — the 8-model panel's graded record, the week's biggest market-vs-model gaps, and what's spiking next. One email, Sundays. These are market prices and model estimates, not predictions of fact and not financial advice. Kalshi is a CFTC-regulated event-contract exchange (18+; availability varies by state).
Why The Market Got Here (The Honest Bull Case)
I'm not going to strawman the 93. It's there for four defensible reasons.
Collins has never led a public poll in this cycle. Ten polls, ten Ossoff leads. The Pollsmax average sits at Ossoff 50.1 – Collins 42.6. The most recent — InsiderAdvantage for James Magazine Online, August 16-17, 800 likely voters — has it 50-43. The only Georgia Senate polling that ever showed a Republican ahead was hypothetical Brian Kemp matchups, and Kemp declined the race back in May 2025.
The counter I'd raise against my own point: nearly all of that polling was taken while Collins was either fighting a primary or freshly out of one, against an incumbent who has been on Georgia ballots since 2017. Ten leads over an undefined opponent is a weaker fact than ten leads looks like.
The fundraising gap is the widest on the Senate map. Ossoff raised $20 million in Q2 and entered the fall with $42 million on hand against Collins's roughly $2.1 million — a 10-to-1 quarter and about a 20-to-1 cash advantage on June 30. His average donation was $42 across 474,000 donations; he's cleared $80 million for the cycle.
The counter: Senate cash advantages are famously poor predictors on their own. Money buys the ability to compete; it doesn't buy the seat, and there is enough outside money inbound here to neutralize a lot of it (more on that below).
Collins came through a contested primary; Ossoff didn't have one. Collins won a June 16 runoff over Derek Dooley, 55-45, after Trump endorsed him roughly two days out and Georgia's sitting governor endorsed Dooley. Ossoff was unopposed in any meaningful sense and spent those same months banking money and running unanswered ads.
And the national environment currently favors Democrats. Silver Bulletin's generic ballot average is D+6.7, or D+8.1 on the likely-voter adjustment. Presidential approval is a standard midterm input, and it's low: 38/59 on the Silver Bulletin average as of July 30-31, and 33/64 in the Reuters/Ipsos poll taken August 14-17.
That's why all three major raters moved toward Ossoff this year — Cook to Lean Democrat, Inside Elections to Tilt Democratic, Sabato's Crystal Ball to Likely Democratic, with J. Miles Coleman writing that "it would be a true shock if Ossoff actually went on to lose."
The Structural Thing That Isn't Getting Enough Attention
Here's the input I think market participants are underweighting. It doesn't change who's favored — it changes how much variance is left in the race, which is exactly what matters when you're pricing a contract at 93.
Georgia requires 50% plus one to win a statewide race. Miss it and the top two go to a runoff four weeks later — this cycle, December 1, 2026. That rule is why Georgia keeps producing overtime.
And it's almost always a Libertarian who triggers it. Shane Hazel took 2.32% in 2020 and forced Ossoff into a January runoff against David Perdue. Chase Oliver took 2.07% in 2022 and forced Warnock into overtime against Herschel Walker. Two cycles, two third-party candidates under three percent, two races that got completely reset.
This year there is no Libertarian on the ballot.
The Libertarian Party of Georgia nominated Allen Buckley in March, then failed to qualify — they needed 72,336 petition signatures by the July 14 deadline and submitted a few hundred before publicly giving up. The mechanism behind it is genuinely wonky: Libertarians normally hold automatic access by clearing 1% in Public Service Commission races, but those races were pulled from the 2024 ballot during Voting Rights Act litigation, and Chase Oliver's 0.39% in Georgia's 2024 presidential race wasn't enough to keep the line.
Per Ballot Access News, the result is that for the first time since 1986, only Republicans and Democrats appear on Georgia's statewide ballot.
So this is a head-to-head. Absent a meaningful write-in vote, somebody clears 50% in November. Inside Elections cited exactly this when it moved the race to Tilt Democratic on August 6.
And be precise about what that's worth, because the lazy version of this take overstates it: runoff exposure was never automatically bad for a Georgia Democrat. Ossoff won the January 2021 runoff, and it drew about 4.48 million voters against 4.95 million the previous November — a real drop, but nothing like a collapse. What's gone isn't a trap Ossoff was likely to fall into. It's a whole second coin flip, with a fresh national money surge attached, that no longer exists. For a favorite, removing a coin flip is worth a lot. That's most of the distance between 85 and 93.
...Or Is He? The Bear Case, Taken Seriously
Now the more interesting half.
Georgia does not vote seven points Democratic. Trump carried it by 2.2 points in 2024. In the last midterm — the closest structural analogue to this November — Kemp beat Stacey Abrams by 7.5. Ossoff's own 2021 win was by 1.23 points, about 55,000 votes. Nothing in this state's recent voting history says a Democrat wins a Georgia Senate race by the margin the polling average currently shows.
Collins is still undefined, and $59 million is coming to define him. This is the strongest piece of the bear case. In the June Wick poll, 23% of Georgia voters had no opinion of Collins or had never heard of him. Nearly a quarter of the electorate with an open mind. Meanwhile Senate Leadership Fund has pledged $44 million and One Nation another $15 million — more than Ossoff's entire $42 million war chest.
Two caveats I want on the record, because they matter: those are pledges, and reported spending so far is a small fraction of them (One Nation's opening buy was $2.5 million). And I'm not counting Democratic outside money, which hasn't been publicly totaled. Treat $59 million as a ceiling on Republican intent, not a spent advantage.
Still, the asymmetry is real. Ossoff's favorability is net +6, with only 8.4% who've never heard of him or have no opinion. He is fully defined. There is almost no upside left in his number and meaningful room in Collins's.
The best-methodology poll in the field is also the closest. Wick Insights, June 27-30, 1,175 likely voters, ±2.9 — the largest sample and lowest margin of error on this race — had it Ossoff +3.8. Meanwhile the Fox News/Beacon poll showing +13 is the highest margin anybody has recorded. I'll be precise rather than dramatic here: dropping Fox from the average moves it from D+7.5 to about D+6.9. Six-tenths of a point. It's an outlier, not a load-bearing distortion.
The age crosstab is the one that would keep me up at night. In AARP's July poll, Collins led voters 50 and over by 53-45, with a wider edge among 65-plus. Midterm electorates skew old. (Fairness demands I note that same poll's topline was Ossoff +9 — above the average. The crosstab and the topline disagree, which is itself informative.)
The Number That Actually Bothers Me
Two things, and they point the same direction.
First: Pollsmax's own forecast model gives Ossoff 77.4%. The market says 93%. That's roughly a fifteen-point spread between a public model and the exchanges on the identical event.
My honest read is that the market is closer to right — models widen their uncertainty bands this far out and can't easily price things like a third party missing the ballot. But a fifteen-point gap still tells you something useful: there is no informational cushion left in 93. You're not buying a mispricing, you're buying the most confident interpretation available of a race that a serious model thinks is a 3-in-4.
Second, and more interesting: there's academic work suggesting high-priced contracts underperform their prices at exactly this kind of horizon. Berg & Rietz, published in the International Journal of Forecasting in 2018, looked at Iowa Electronic Markets contracts priced between 90¢ and $1.00. One day before resolution, they were well calibrated — average payoff 95.5¢ against an average price of 95.1¢. But at a four-day horizon those contracts paid an average of 79¢, and at fourteen days they paid 59¢. The authors call it an overconfidence bias in high-priced contracts.
The sample sizes are small — n=27 at the fourteen-day mark — so don't build a religion on it. But Georgia is 74 days out. Kalshi's own 2026 calibration study, across 2.2 million resolved markets, found accuracy climbing from 88.3% at a three-month horizon to 97.2% at close. The market gets sharp late. We are not late.
Put Wisconsin, the model gap, and the horizon research together and you get a coherent thought: 93 cents in late August is not the same object as 93 cents on November 2.
Where The Live Money Actually Is
Kalshi runs a market on which Senate race will have the smallest margin of victory. Georgia trades at 2.4% to be the closest race in the country — thirteenth on a list of seventeen. That's the market telling you, in its own words, that it does not expect this to be close.
So if you want action, here's where I'd look instead:
Texas is the genuine coin flip. Kalshi has Ken Paxton at 52¢. Polymarket has James Talarico at 51.5¢. The two exchanges disagree on who the favorite is — and it's by far the highest-volume Senate race on Kalshi, roughly 9.1 million contracts traded across both legs against Georgia's 1.1 million. If you want a true 50/50 with real liquidity, this is it.
Senate control has a four-and-a-half-point cross-exchange gap. Kalshi's CONTROLS-2026 has Democrats last trading at 46¢; Polymarket has Democrats at 50.5¢. Kalshi's own balance-of-power combo implies about 48% for a Democratic Senate, so its two markets are roughly in line with each other — this is a straight disagreement between venues, not a broken market. Four and a half points isn't a screaming edge, but it's the most-traded political question in America and it's not settled.
The Georgia governor's race is the live Georgia market. Keisha Lance Bottoms is around 61¢ on Kalshi, 59¢ on Polymarket against Rick Jackson. Roughly 60/40 — not a coin flip, but a real question. And here's the wrinkle worth your time: that same InsiderAdvantage poll that had the Senate race at Ossoff +7 had the governor's race dead even at 46-46. A tied gubernatorial race and a seven-point Senate race, same ballot, same sample. Somebody's getting split-ticketed, and that's the kind of divergence that eventually resolves in one direction.
A note on the combo, since people ask: Kalshi's Bottoms-and-Ossoff parlay last traded around 63¢, while multiplying the legs gets you about 56%. Don't read that as a seven-point rip. Two statewide races on the same ballot are strongly correlated, so the true joint probability should sit above the product of the legs — 63 against 56 is roughly what correlation looks like. If you think Georgia splits its ticket, fade the combo and back the governor's race alone. If you think the state votes as a bloc, the combo is fairly priced.
The cheapest defensible underdogs. Dan Osborn, running as an independent in Nebraska, is 29¢ on Kalshi and 27¢ on Polymarket — and he took roughly 47% against Deb Fischer in 2024. Susan Collins in Maine is about 36¢ Kalshi versus 30.5¢ Polymarket, a five-and-a-half-point gap on someone who has never lost a Senate race. Michigan runs the other way: El-Sayed is 61¢ Kalshi versus 64.5¢ Polymarket. Both of those venue gaps are bigger than most people's actual edge, which is the whole point.
And if you want the Georgia longshot: Kalshi has the Senate race being won outright in the first round at 92.7¢, implying about a 7% chance of a December runoff. I think the runoff probability is too high — with no Libertarian on the ballot and only write-ins able to deny a majority, it should be closer to zero.
Yes, I'm aware I just spent a section arguing against 92-cent contracts at long horizons. The distinction I'd draw: Ossoff at 92.5¢ is a bet on a contested election seventy-four days out, where new information keeps arriving. The runoff contract is a bet on ballot mechanics that are already settled — the qualification deadline passed in July and nothing between now and November can put a third name on that ballot. Same price, very different kind of uncertainty.
The real problem there is size: that market has traded fewer than 8,000 contracts. There may be an edge. There is definitely not volume, and it only exists on one venue.
A records check for this piece found no documented lawsuit or ethics record for Jon Ossoff.
- Why Jon Ossoff (D) Will Win The Georgia Senate Race
- AFC West Odds 2026: The 14-3 Champion Trades Third
- BMW Championship Round 2 Leader Odds: Kalshi's 5-Way Tie
- NFL Hot Seat Odds: Why Kalshi Money Piled On Mike Vrabel
- Valkyries Title Odds: The Expansion Team Kalshi Won't Fade
SO — Lock, Or Not?
Here's my honest landing spot.
Ossoff at 92-93¢ is defensible and probably close to right. The polling has been one-directional all year, the cash advantage is the largest on the map, his opponent is undefined heading into the fall, and the third party that twice dragged Georgia into overtime isn't on the ballot. If you forced me to pick, I'd pick Ossoff, comfortably.
But I would not buy Ossoff at 93. There's nothing left in it, the calibration research says long-horizon favorites at this price have historically underperformed, and there is $59 million of pledged money whose entire job for the next ten weeks is making Mike Collins into somebody Georgia voters recognize. Paying 93 to win 7 on a race where a quarter of the electorate hasn't formed an opinion of the challenger is not my idea of a good time.
And I'd stop calling it a lock. Ten days ago, on these exact two exchanges, a 95% favorite in Wisconsin lost by half a point.
5% is not 0%. Neither is 7.8%.
Wherever you land, shop the number across both books first. On Maine and Michigan the gap between Kalshi and Polymarket is bigger than the edge most people are hunting for.
Sources: Kalshi Georgia Senate · Polymarket Georgia Senate · Fortune on the Wisconsin primary result · PBS Wisconsin on Crowley's win · Pollsmax Georgia forecast · InsiderAdvantage/James Magazine, Aug 18 · Wick Insights, June · Fox News/Beacon poll · AARP Georgia poll · WABE on Libertarian ballot access · Ballot Access News · Georgia SOS 2026 elections calendar · AJC on Ossoff's $42M · AJC on One Nation's $15M · Ballotpedia: 2020-21 Georgia runoff · Cook Political Report Senate ratings · Inside Elections, Aug 6 · Silver Bulletin generic ballot · Berg & Rietz, IJF 2018, on longshots and overconfidence · Kalshi calibration study, Aug 2026 · Action Network on the 2016 NBA Finals odds · NCAA tournament upsets by seed
