The Collins Contract at 36¢ Is the Best Mispricing Left in the Midterms

by Andrew Keene, The Prediction Network

The 2026 Maine Senate race has become one of the deepest pools of money in the prediction market midterms, and the price on Susan Collins has drifted to a number that the evidence doesn’t support. As of this weekend, Collins sits around 36% on average, with Kalshi at 37¢ and Polymarket at 35¢. That price says the market thinks Troy Jackson is nearly a 2-to-1 favorite. The polls, the history of this specific candidate, and the structure of this specific race say something very different: this is a coin flip, and a coin flip priced at 36¢ is exactly the bet a disciplined trader should want. 

To be clear up front: nobody knows if Collins wins. The argument isn’t that she will. The argument is that the gap between the market price and her true chance is wide, and you get paid handsomely for being right about a toss-up.

Loading chart…

A note on the numbers before we start

A 36¢ contract that pays $1 returns 64¢ of profit on 36¢ risked, which is about 1.78-to-1, or roughly +178 in sportsbook terms. That’s a bit short of a true 2-to-1, and fees shave it a little more. But the logic holds: at 36¢ you only need Collins to win slightly more than 36% of the time to profit over the long run. If her real chance is 45%, every dollar you put in is worth about $1.25 in expectation. If it’s 50%, it’s worth about $1.39. That’s the whole case in one line: you don’t need her to be a favorite, you need her to be a better underdog than the market thinks.

Reason 1: The polls say tie, the price says blowout

Start with the most basic disconnect. The market has Jackson at roughly 64%, a price you’d expect for a candidate holding a meaningful, durable lead. He doesn’t have one.

RealClearPolitics’ average from September 23 through October 20 shows Jackson ahead by just 1.4 points, 48.8 to 47.4. The Pollsmax average is slightly wider, with Jackson up 2.5 points, 48.9% to 46.4%, across 22 polls. Even Kalshi’s own adjacent market agrees the race is close: its margin market points to a Democratic win of about 1.7 points, in line with the polling averages. 

Think about what that means. Kalshi’s traders, on Kalshi’s own platform, are pricing a final margin of under two points, and simultaneously pricing the winner market as if it were a 64/36 race. A sub-two-point expected margin in a Senate race, with ordinary state-level polling error of several points in either direction, does not produce a 64% favorite. It produces something closer to 55/45. The winner market is out of line with the margin market, and the side that’s too cheap is Collins.

And the newest high-quality data point is the friendliest of the month for her. A Fox News poll released Thursday had Collins and Jackson tied at 49% among likely voters, less than two weeks before Election Day. That Fox survey was large, with 1,514 likely voters and a 2.5-point margin of error. CBS found the same thing earlier in the month: Collins is drawing enough voters, including enough Maine Democrats, to keep the race tied. Trafalgar had it tied too. co/efficient had Collins up 2, and NYT/Siena had her up 3 in mid-September. 

When the most recent big sample shows a dead heat, and the averages show a race inside two points, a 36¢ price is not cautious. It’s a market that has already decided the result.

Reason 2: The market just overreacted, and it has a history of doing that here

This contract has been a roller coaster, and that volatility is the opportunity. Kalshi traders pushed Collins from 27.3% to 36% after September 24, then kept going: by October 16, Kalshi priced Collins at 43%. Then it reversed hard. In the last week alone, Collins’ Kalshi daily close dropped from 43% to 37%, a six-point slide. 

What happened in that week to justify a six-point collapse? The polling didn’t move six points. The biggest new poll was a tie. What did happen was a stream of headlines: scrutiny of her campaign’s private jet spending alongside Republican attacks on Jackson over taxes, which even a neutral race tracker characterized as mixed and mostly limited campaign noise. Meanwhile, genuinely positive news for Collins arrived in the same window: former West Virginia Sen. Joe Manchin said he will campaign alongside her, and Collins says she secured tariff exemptions for Maine farmers 11 days before Election Day. 

A market that swings from 27 to 43 to 36 in a month, while the polling average barely budges, is a market trading on mood. Mood-driven prices overshoot. Right now the overshoot is to the downside on Collins.

Reason 3: No incumbent in America has a stronger record of embarrassing her polls

This is the core of the bull case, and the market is discounting it to nearly zero.

In 2020, Collins was supposed to lose. Not narrowly, convincingly. She won by nearly 10 points even though every state poll on RealClearPolitics had her behind Sara Gideon. Some of those misses were enormous: Quinnipiac had her down 12 in September, and Emerson had Gideon up six at the end of October. Colby’s final poll is the cleanest illustration: it showed Gideon ahead by 3.2 points a week before the election, and Collins won by 9. That’s a 12-point miss from a reputable in-state pollster. 

Why did it happen? The people who ran the polls told us. Colby’s Daniel Shea said his polls failed partly because there were more split-ticket voters than expected, people who voted Biden and Collins, and the polls also underestimated her big margins among blue-collar workers in rural Maine. Shea also floated the theory that voters who aren’t hard partisans, and so are more likely to split tickets, may also be less likely to answer polls. 

Read that again, because it’s the whole trade. The exact voter who keeps Collins alive, the soft-partisan ticket-splitter in rural Maine, is the voter pollsters have the hardest time reaching. That’s not a random error; it’s a structural one that points in Collins’ direction. Pollsters say they’ve adjusted, but University of Maine political scientists describe polling as being in a bit of a crisis, with more ways for it to go wrong today than before.

Even the Bangor Daily News, reporting the Fox tie, made the point explicitly: Collins hasn’t led a public poll since late September, but she also trailed in every single one in 2020 before handily beating Gideon. If she runs even a third of her 2020 polling overperformance, she wins this race. The market is pricing her as if that history doesn’t exist. Kalshi even lists a separate market on whether Collins will outperform the polls again this year, which tells you traders know it’s a live possibility. They just aren’t paying for it in the main contract. 

Reason 4: Her coalition is already visible in the crosstabs

You don’t have to rely on a hidden-voter theory alone. The current polls show the crossover support that wins Collins elections, in plain sight.

In the Fox poll, she leads independents by 13 points, and one in five likely voters who disapprove of Trump are backing her. That second number is extraordinary in a polarized era. An AARP poll last month found Collins with more swing support than any other Republican across seven battleground states. CBS found the same dynamic from a different angle: she’s bolstered by voters who say they’re casting their ballot on state matters rather than national ones, and she holds an edge on who would bring more federal money back to Maine. The Washington Post poll, which was one of Jackson’s better results, still found that likely voters favor Collins’s ability to steer federal dollars to the state. 

Here’s the thing about a candidate who wins Trump disapprovers and independents: she has a ceiling far above her party’s. A generic Republican in a Democratic-leaning state facing a hostile national environment is dead. Collins isn’t generic, and the polls are already showing the split-ticket behavior that the market treats as a relic of the past.

Reason 5: Mainers themselves think she’s going to win

One of the most underrated indicators in election forecasting is the “who do you think will win” question, which captures voters’ read of their own communities, not just their own vote. In the Marist poll, Jackson led the head-to-head 51–47 among registered voters, yet 57% of Maine voters said they think Collins will actually win, compared to 41% for Jackson.

That’s a 16-point gap in expectations running directly opposite to the market price. Mainers have watched Collins win five statewide races. They know their neighbors. And they’re telling you, by a wide margin, that they expect her to pull it out. The market is pricing the opposite.

Reason 6: Jackson is a replacement nominee with a short runway

The Democratic nominee isn’t the candidate Democrats spent a year building. Graham Platner withdrew on July 10 amid a rape allegation, which he denies, and the state party had to find a new nominee through a compressed convention process by a July 27 statutory deadline. Jackson has been the nominee for barely three months.

That matters. Even sympathetic coverage noted the drop-off: Platner was described as a charismatic, dynamic speaker, and Jackson, by his own admission, is not. Jackson also carries an ideological profile that gives Collins a target. In the Fox poll, 46% of voters say Jackson is too liberal. Republicans are pressing attacks on Jackson over taxes into the final stretch. A less-known challenger who got the job by emergency succession is exactly the type of candidate whose numbers soften late, and late softening is precisely what happened to Gideon in 2020, when a Democratic source saw danger signs as the race tightened late. 

Reason 7: Why a deep, heavily traded market can still be wrong

The obvious objection: if this is one of the most traded midterm contracts, isn’t the price efficient? Not necessarily, and the data on who’s trading actually supports the Collins side.

The volume is real. Kalshi’s Maine Senate market alone has more than $14.4 million in volume. The race sits in the top tier of midterm markets: a watchdog analysis found the Senate races in Texas, Maine, Michigan and South Carolina, plus Kentucky’s 4th District, each draw several times more trading than the next-highest race, with Texas the clear leader, since Kalshi names Texas Senate its most traded competitive midterm race. (I could not find a source confirming Maine is specifically #2 across all platforms; it is firmly in the top handful, behind Texas.) 

But high volume isn’t the same as smart volume. On Polymarket, the top 1% of wallets by volume account for 70% of all congressional betting this cycle, and ten wallets control 17% of all congressional volume on that platform. A market dominated by a small number of large accounts can be pushed by conviction, by hedging, or by partisan enthusiasm. In a midterm with a strong Democratic national environment, the natural partisan lean of an engaged online trading crowd is toward the Democrat in every close race, and the national narrative is a powerful gravitational pull. The Maine race is the one place where that narrative has repeatedly failed.

There’s also a logic gap in the market structure itself. Remember the margin market: a Democratic win of about 1.7 points. A trader who believes that number should not believe 64/36. One of those prices is wrong, and Collins at 36¢ is the cheap side of the inconsistency. 

Reason 8: The national environment is already in the price, and then some

The bear case on Collins is the national mood, and it’s real. 62% of Mainers disapprove of Trump’s job performance, and 53% are concerned Collins votes with him too often. Maine Democrats hold a nine-point lead on the generic congressional ballot.

But that’s the point: all of this is in the polls already, and the polls are still tied. Collins is running roughly seven to nine points ahead of her party on the generic ballot. That isn’t a candidate being dragged under by the environment; it’s a candidate who has already absorbed the hit and is still standing at 49%. The market is double-counting the national environment, once in the polls and again in the vibes.

The math, plainly

Let’s be concrete about what the bet is.

• At 36¢, the market implies about a 36% chance.
• The polling averages imply a race inside about two points, which, with normal state-level polling error, is roughly a 40–45% proposition for the trailing candidate before you account for anything Collins-specific.
• Layer on a candidate with a documented history of beating her polls, a 13-point lead with independents, a fifth of Trump disapprovers in her column, and an electorate that expects her to win by 16 points, and a fair estimate in the mid-40s to 50% is defensible.

At a 45% true probability, a 36¢ contract has about 25% positive expected value. At 50%, it’s about 39%. You will not find many liquid, high-volume contracts with that kind of edge eleven days before resolution. The market has handed you a toss-up at underdog prices.

What could make this wrong

A case worth making is worth stress-testing, so here’s the other side, and it’s serious.

The polls could be right this time, or wrong in Jackson’s favor. Pollsters have changed their methods since 2020, tweaking designs, finding new ways to reach people, and rethinking their likely-voter models, so the 2020 miss may not repeat. Collins hasn’t led a public poll since late September, and Jackson leads in most of them, if narrowly. The national environment is genuinely bad for Republicans, and with ticket-splitting in decline, Collins is widely seen as the most vulnerable Republican incumbent of the cycle. Midterm electorates can break late and decisively against the president’s party. There are also live controversies in the background, including a ProPublica report that FBI agents got approval for a broader public-corruption investigation after examining allegations involving Collins and other members of Congress, which could resurface. And the market may simply know something the polls don’t: prediction markets aggregate a lot of informed money, and Kalshi says its prices have historically been well calibrated, with accuracy improving as markets near resolution. 

Finally, the practical warning: as CNN noted, most people who trade on prediction markets lose money. A contract that resolves to zero 55–65% of the time will lose you money most of the time, even if it’s a good bet in the long run. Size it as such. 

I’m not a financial advisor, and this is the bull case for one side of a trade, not a recommendation to make it. The case is this: at 36¢, you’re being offered roughly 7-to-4 on a race the polls call a tie, against a candidate who has made a career out of proving Maine’s polls wrong.