Michigan Governor Odds: Seven Raters, Not One Says Safe

Updated August 28, 2026 · 14 min read · by Sam Smith

Just over three weeks after Michigan settled its primaries, the exchange has already made up its mind about November. Kalshi's Democratic contract on the governor's race last traded at 88 cents, and if you want to own it right now you are paying 91.8 cents at the offer. Put up about 92 cents, collect a dollar if the contract resolves Yes, lose the stake if it resolves No.

Now go read the people who rate these races for a living. Seven of them publish a Michigan rating. Not one says Safe.

The Price And The Problem

First, the shape of the thing. Gretchen Whitmer is term-limited after two terms, so this is an open seat with no incumbent on the ballot: Democrat Jocelyn Benson, the secretary of state, against Republican John James, who represents Michigan's 10th district in Congress. The gap between the price and the ratings is the whole story of this page, and it is not the gap you would expect. There is no scandal driving it and no hidden poll. The exchange and the handicappers are looking at the same public record and putting it on two different shelves.

Start with the contract itself, because the wording matters more than the headline. The event is GOVPARTYMI-26, and it does not resolve on a person. It resolves on a party: "If a representative of the Democratic party is inaugurated as the governor of Michigan pursuant to the 2026 election, then the market resolves to Yes." Inauguration is the trigger, though the exchange can settle early once a consensus of news organizations projects the winner. New to how these contracts work? Our plain explanation of prediction markets covers the mechanics, and this page is the Michigan layer on top.

Three contracts are listed. The Democratic line for Jocelyn Benson, the Republican line for John James, and a third line for Mike Duggan, the former Detroit mayor who ended his independent campaign on May 21 after a Detroit Regional Chamber poll put him 11 points back. His contract still trades. It is the busiest line on the board by lifetime volume, 120,143 contracts, and it is quoted at two tenths of a cent. That is not a market disagreeing about Duggan. That is the paperwork of people closing out positions on a race that ended for him in the spring.

What the board does not show you is that three more names will be on the November ballot: Anthony Hudson for the Libertarians, Douglas Campbell for the Greens, and Donna Brandenburg for the U.S. Taxpayers Party. None of them has a contract. If any of them won, all three listed contracts would settle No. Nobody expects that. It is worth knowing that the listed board is not the ballot.

Jocelyn Benson And The Record Behind The Price

Benson is the strongest single argument for a price this high, and the argument rests on results rather than impressions. She has already won a statewide race in Michigan. In 2022 she was re-elected secretary of state 55.86 percent to 41.93, a 13.9-point margin, which is a wider win than either of Gretchen Whitmer's two gubernatorial victories. On August 4 she took the Democratic nomination with 83.22 percent against Genesee County Sheriff Chris Swanson, 1,251,160 votes to 252,236.

She also has the money. As of the July 19 filing with the Michigan Secretary of State, Benson had raised $10.7 million and was sitting on $6.4 million in cash. James had raised $5.6 million and had $1.4 million left. On money already raised she is ahead by about two to one; on money still in the bank it is 4.6 to one, which is the number that matters going into the expensive part of the calendar, and it is the last filing before the primary, so the real gap is unmeasured rather than closed.

Her record is not spotless, and the parts that are not spotless are on the public record rather than in rumor. In May 2025, Michigan Attorney General Dana Nessel, a Democrat, found that Benson violated the Michigan Campaign Finance Act by holding her campaign launch in the lobby of a state office building. Nessel's office said Benson appeared to have acted under a "mistaken belief" that she was complying, and that the office has no power to impose a penalty, but noted the finding could be used as evidence in any future proceeding.

The live one is data centers. Benson's husband, Ryan Friedrichs, is a vice president at The Related Companies, whose subsidiary is developing a contested AI data center in Saline Township. When the role became public in November 2025 he moved off all Michigan projects, and the company said he is not paid on commission. What makes this a general-election problem rather than a partisan talking point is that the pressure came from both directions. Republican Perry Johnson claimed her husband was "making $70 million from these data centers," a figure Bridge Michigan reported as disproven. Her own primary rival ran on the contrast, posting that his family "hasn't made a dime from data centers and never will." Benson's answer, verbatim: "None of my family is profiting from data centers, either."

And one item cuts her way. The Trump Justice Department sued her for an unredacted copy of Michigan's statewide voter list. A federal judge dismissed the case in February 2026, and the Sixth Circuit affirmed that dismissal in June. She won, and the win is the spine of the campaign she is running.

John James, And The 481 Votes

One number does most of the work in the Republican case, and it cuts both ways.

James won his primary with 466,714 votes out of 932,464, or 50.05 percent. He cleared an outright majority by 481 votes. Perry Johnson took 35.20 percent. The other 14.75 percent went to Mike Cox and Aric Nesbitt, both of whom had already withdrawn and whose names were still on the ballot collecting votes. A Trump-endorsed frontrunner who barely cleared a majority while withdrawn rivals were still collecting votes has a consolidation question to answer, and September is when the answer starts to show up.

The primary that produced him was not a friendly one. Perry Johnson sued James over his campaign logo, arguing it could mislead voters into thinking James was already the sitting governor, and rival Mike Cox attacked his family business record. Unverified personal allegations also circulated between the camps in the spring; they stay out of this piece because nobody has documented them.

The turnout number underneath it is starker. Michigan runs open partisan primaries, so every voter picks one party's ballot. On August 4, 1,503,406 people took the Democratic gubernatorial ballot and 932,464 took the Republican one. That is a 23.4-point participation gap, and it ran the wrong way for the usual explanation: the contested race was the Republican one. The Democratic race was effectively over before it started.

Set against that, Michigan Republicans keep coming close. James has run statewide twice and lost twice, but his 2020 loss to Gary Peters was 49.90 to 48.22, a margin of 1.68 points. Four years later Mike Rogers lost the 2024 Senate race to Elissa Slotkin by 48.64 to 48.30, a third of a point, and Rogers is on this November's ballot again. Donald Trump carried Michigan in 2024, 49.73 to 48.31. This is not a state that hands Republicans statewide offices, and it is also not a state where they lose by much.

That last comparison is worth sitting with, because both races settle on the same night in front of the same voters. The exchange prices the Republican running for governor at 13 cents and Rogers, running for Senate, at 39. The raters split them the same way, governor at Tilt to Likely D and the Senate race as a genuine tossup. Nobody is claiming the two races are the same race. The question this page keeps circling is one of degree: how much further toward a formality the governor's line deserves to sit than the Senate line, given that the Republican in the closer race is the one who has already lost statewide by less.

What The Panel Priced

Here is the board, with the exchange on one side and the model panel on the other.

Six AI models were given the full record above and asked to price all three outcomes in a single pass. They were not shown the market prices. Then each one read the other five's numbers and reasoning, anonymised so no model could defer to a brand, and was allowed to revise. The figures below are the revised round.

Outcome	Party	Kalshi last	Bid / ask	Panel (revised)
Jocelyn Benson	Democratic	88¢	89.0 / 91.8¢	82.7%
John James	Republican	13¢	9.2 / 13.0¢	17.2%
Mike Duggan (Withdrew May 21)	Independent	0.2¢	0.2 / 1.0¢	0.2%

The panel number on each line is the blended figure, which here means the midpoint of the six seats' numbers. That is why the three do not add up to exactly 100.

More live boards from the same panel: Michigan Senate · Democrat 61¢ · Georgia governor · Democrat 60¢ · California governor · Democrat 96¢. Prices fetched August 28, 2026.

That California line is the useful comparison. Becerra trades at 96 cents in a state where every rater says Solid or Safe D and where Democratic candidates together took 62 percent of the June top-two primary vote against 36 percent for all Republicans. Michigan trades at 88 to 89 with a rating range of Tilt D to Likely D. The exchange is separating the two, just less than the raters do.

Every Seat's Number

Seat	Round 1: Democratic	Revised: Democratic	Revised: Republican	Revised: Duggan
Claude Fable	83.2	82.8	17.2	0.0
Claude Opus	88.0	84.0	15.6	0.4
Claude Sonnet	82.6	81.8	17.9	0.3
Gemini	83.5	84.5	15.5	0.0
Grok	73.5	82.5	17.2	0.3
Kimi	75.0	82.5	17.4	0.1

Six of the panel's eight seats ran this board. One seat declined to price an active political election, and one was unavailable.

Every seat on this panel is graded against real market settlements — records to date: Claude Fable 84% on 807 graded calls · Claude Opus 84% on 877 graded calls · Claude Sonnet 82% on 852 graded calls · Gemini 86% on 5,151 graded calls · Grok 84% on 4,427 graded calls · Kimi 82% on 2,576 graded calls. Recomputed daily; the full scoreboard is public.

Model estimates generated August 28, 2026, price-blind from live fetched data. These are model estimates, not predictions of fact and not financial advice. Kalshi is a CFTC-regulated event-contract exchange; 18+, and availability varies by state as of August 2026.

Now the part that keeps this honest. The panel reads 82.7 percent and the market says 88 to 89 cents. Do not read that as a signal. We grade every one of these calls in public, and the record says our panel loses roughly two out of three of its disagreements with the market. The gap is context, not a recommendation, and the reason to look at it is that it tells you which facts the two sides are weighting differently. Every number on this page gets graded against the certified result on the full graded scoreboard.

Set the panel number aside entirely for a moment, because the next paragraph has nothing to do with it. What a price tells you on its own is simpler and more durable than any model number. At 91.8 cents you are risking 91.8 to make 8.2, which needs the Democratic contract to be right about 92 percent of the time just to break even before fees. At 13 cents on the Republican line you risk 13 to make 87, and you need it to come in about one time in eight. Those two break-evens come from the quotes alone. They are not a comparison with anything the panel said, and what to make of them is the reader's judgment, not ours.

Where The Panel Changed Its Mind

The revision round moved two seats a long way, pulled one four points in the opposite direction, and left the other three inside a point of where they started. That is the interesting part.

Grok opened at 73.5 percent for the Democratic line and came back at 82.5. Its first-round reasoning had leaned hardest on the ceiling: "structural Dem gubernatorial cushion (Whitmer D+10/D+10) versus 2024 Trump R+1.4, so this is not a pure blue hold." Its revised note reframed the same facts as counterweights rather than as the core read: "Counterweights keeping this off Safe: open-seat variance, 9.3% unallocated, James's demonstrated ceiling, and no handicapper above Likely D." — Grok

[Editor's note: Grok's revised write-up states that it "did not revise toward the 73–75 or 88 poles." Its own stored numbers moved 9 points, from 73.5 to 82.5. We publish the quote as written and the numbers as recorded; the model's self-description does not match its own arithmetic.]

Kimi made almost the same trip, 75.0 to 82.5, and its stated uncertainty is the sharpest question on the page: "whether the August primary turnout gap (D+23 participation despite the contested race being Republican) reflects genuine enthusiasm or noise." — Kimi

[Editor's note: Kimi's revised write-up likewise says it "did not materially revise" while its numbers moved 7.5 points.]

Claude Opus went the other way, 88.0 down to 84.0, and it showed the arithmetic: "A D+6 expected margin with ~4 points of sigma yields ~84%." — Claude Opus

[Editor's note: sigma there means polling error, roughly four points of slop either side of an expected six-point win. Run that calculation and it actually returns about 93 percent, not 84. The seat landed on 84 and described a route to it that does not get there; we publish the number it filed and the reasoning it filed, and flag the gap rather than quietly correcting either one.]

And one seat simply held. Claude Fable opened at 83.2 and finished at 82.8, and said plainly why the low seats had not moved it: "a D+6 to D+8 expected margin with 3-4 points of midterm-era poll error (2018/2022 bias was near zero) does not produce a one-in-four Republican outcome." — Claude Fable

What Would Change The Panel's Mind

Four things, all of them dated and checkable rather than atmospheric. Five of the six seats named the first one unprompted.

Multiple post-Labor Day live-caller polls showing James within two points. The current picture is a four-aggregator average of Benson 49.3 to James 41.4, updated August 23. Only four of the 13 public head-to-head readings were taken after both nominees were set, and those four came from just three polling operations, because one of them reported the same fieldwork at both a registered-voter and a likely-voter screen. September is the first month of clean data. This is the single trigger the panel named most often, and it pushes the Democratic number down.

James consolidating the withdrawn candidates' voters. Mike Cox and Aric Nesbitt took 14.75 percent of the Republican primary while off the trail. Endorsements, joint appearances, or a September Republican poll showing James at or above 45 would all point the same way, and all of them push the Republican number up.

The 9.3 percent the aggregators do not allocate to either candidate. The four-aggregator average leaves nearly a tenth of the electorate parked between the two names. The midterm environment points that block toward Benson, since she is the out-party candidate in a cycle where the president is about 21 points underwater. If September polling shows it breaking toward James instead, the whole board moves.

A national shift in presidential approval. Republicans hold the White House this cycle, so the midterm base rate runs against them and for Benson. Presidential approval sits at 37.7 percent approve against 59.0 disapprove across ten aggregators as of August 27, with a generic ballot of D+6.2. The president's party has lost House seats in 18 of the last 20 midterms, and the two exceptions both required a dominant national event, an impeachment backlash in 1998 and a post-9/11 rally in 2002. Neither condition exists today. If one appeared, that base rate would weaken, and the number that rises is James's.

Settlement, And When This Page Gets Re-Scored

Election Day	November 3, 2026
Contract Settles	On inauguration, expected January 1, 2027
Early Settlement	Allowed once a consensus of news organizations projects a winner
Next Catalyst	First post-Labor Day independent polling, early September 2026
Second Catalyst	Post-primary campaign finance filings, which will show general-election money for the first time
Third Catalyst	Any scheduled Benson-James debate

Prices on this page are as of August 28, 2026, fetched from the exchange, and the panel's numbers were generated the same day, price-blind. This page gets re-scored when the story moves, and the panel's call gets graded against the certified result whichever way it lands.

Hottest Prediction Markets Right Now

• 2028 Democratic presidential nominee · $196M traded
• 2027 Pro Football Champion · $73M traded
• Pro Baseball Champion · $62M traded
• 2028 Republican presidential nominee · $60M traded
• TOUR Championship Winner · $34M traded
• Bitcoin price at the end of 2026 · $33M 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.

More on this:

• Rhode Island Governor Primary: McKee Says 'I Am the State
• Alaska Governor Odds: Dropping Out Won't Clear The Field
• Wyoming Governor Primary Odds: Kalshi Fades Trump's Pick
• Minnesota Governor Primary Results: Demuth Beats Lindell By 10.9
• Florida Governor Primary Results: The Majority Kalshi Missed

The Bottom Line

The exchange is charging near-lock money for an outcome that seven professional handicappers all decline to call a lock. That is not a contradiction so much as a difference of shelf: everyone agrees Benson is favored, and they disagree about how far past "favored" this race has travelled.

What the record actually supports is the narrow version. Benson has won statewide by more than either Whitmer margin, she has a 4.6-to-1 cash lead, and she is running in a midterm where the president's party is about 21 points underwater. James has never won statewide, and he needed 481 votes to clear a majority in his own primary. But Michigan voted for Trump in 2024, two Republicans have lost Michigan Senate races by 1.68 points and by a third of a point since 2020, nearly a tenth of the electorate has not picked a side, and only three polling operations have measured the actual matchup since the nominees were set.

The number that decides this is not on the board today. It arrives in September, in the first independent polling of the actual matchup, and everything on this page is provisional until it does.

Hero illustration: OddsShopper, in the house collage style. Jocelyn Benson photo by JFM Photo 73, licensed CC BY 4.0; photos cropped, toned, and composited.