California Governor Odds 2026: Becerra Vs. Hilton Model Verdict

Kalshi's California governor market opened with 25 names on it. Twenty-three of them were settled to No on a single Tuesday in June, when the state's top-two primary cut a 68-candidate ballot down to the only two people who can appear in November. What is left is a two-line board: Democrat Xavier Becerra at 95.75 cents and Republican Steve Hilton at 4.25. We handed eight AI models a fetched, dated file on the race and asked them to price both names without letting any of them see a price. They came back nearly on top of the market, and then spent an entire second round arguing with each other about the one thing actually in dispute here: how big the losing side's tail should be.

The Quick Answer

Our price-blind, eight-model panel blends Xavier Becerra to 96.4 percent and Steve Hilton to 3.0 percent to win the November 3, 2026 California governor's race. Kalshi has them at 95.75 cents and 4.25 cents. The panel is fractionally higher on Becerra and roughly a quarter lighter on Hilton than the market, which by verdict-board standards is agreement, not an edge. The number that drives it is not a poll: it is the certified June 2 primary, where Democratic candidates took 62.38 percent of 9,262,468 votes cast and Republican candidates took 35.95, a 26.4-point gap measured on the same statewide electorate, voting on this same office.

• Market favorite: Xavier Becerra, 95.75 cents
• Model favorite: Xavier Becerra, 96.4 percent blend
• The trailing side: Steve Hilton, 3.0 percent blend against a 4.25-cent price
• Panel range on Hilton: 2 to 4 percent after the revision round, down from 1 to 5 before it
• The panel's logic in one line: every independent signal points the same way, so most of the panel's work went into sizing what the remaining 3 percent is actually buying.

The full model-by-model board, the two-name ballot mechanics that make this market unusual, where three seats changed their minds, and the concrete events that would move the number are all below.

Want the panel's verdict when this market settles? We grade every price-blind call in public once a market resolves, including this one. Grab the settled result and the re-scored board through the free email capture at the end of this article. No promo code, no pick sold.

New to Kalshi's event contracts? Start with our guide to trading events on Kalshi before reading the board below.

The Market At A Glance

Venue	Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of August 2026)
The Event	"California Governor winner? (Person)" (Kalshi event KXGOVCA-26), originally 25 separate yes/no contracts, one per named candidate
The Contract	Each market resolves Yes only if that named person is elected governor of California pursuant to the 2026 election. Kalshi's listed settlement source for the event is the California Secretary of State.
The Key Rule	The event is mutually exclusive. Kalshi's listed early-close condition is the swearing-in of the first person to hold the office pursuant to the 2026 election, and its listed expected expiration is January 6, 2027, not election night. The 23 candidates eliminated in the June 2 primary were settled to No in June, leaving two live contracts.
Live Contracts	Xavier Becerra (D) and Steve Hilton (R) only
Prices Below As Of	17:31 UTC on August 3, 2026. Prices quoted are bid/ask midpoints, fetched live: Becerra 95.7 bid / 95.8 ask, Hilton 4.2 bid / 4.3 ask. Prices on this page refresh as the campaign moves; the timestamp carries the reading these numbers reflect.

These are model estimates, not predictions of fact and not financial advice. Kalshi event contracts trade on a CFTC-regulated exchange; you must be 18 or older and in an eligible state to participate. This piece takes no political position and endorses no candidate. Every fact below is fetched and dated, both candidates are described in the same neutral terms, and nothing here is a claim about who should win.

What Is Actually On The November Ballot

Most national coverage of a governor's race assumes a Democratic nominee and a Republican nominee produced by two separate primaries. California does not work that way, and the difference is the whole reason this market looks the way it does.

Under the Top Two Primaries Act, approved by voters as Proposition 14 in June 2010 and upheld by an appellate court on September 19, 2011, every candidate for governor regardless of party appears on one June ballot, and the two highest vote-getters advance to November. There is no party nomination and no runoff. Proposition 14 also specifically prohibits write-in candidates in the second round.

That last clause matters more than it sounds. It means the November ballot for this office is a closed two-name field: no third candidate can emerge, no party can substitute a replacement, and a write-in campaign cannot be counted.

California law goes one step further, and the detail is worth knowing because it is the single most misread thing about this market. California Elections Code section 8803 states that "no vacancy on the ballot for a voter-nominated office at a general election shall be filled," and that if a candidate entitled to appear on the general election ballot dies, "the name of that candidate shall appear on the general election ballot and any votes cast for that candidate shall be counted." If that candidate then receives a majority, the statute says "he or she shall be considered elected to that office and the office shall be considered vacant at the beginning of the term." A candidate leaving the race after the primary therefore does not hand the office to the other name on the ballot. The voters still decide, the departed candidate can still be considered elected, and any vacancy is filled afterward the way a mid-term vacancy would be.

The June 2, 2026 primary drew 9,262,468 votes across 68 candidates. Here is the certified top of the field, per the California Secretary of State's Statement of Vote published July 10, 2026:

Candidate	Party	Votes	Share
Xavier Becerra	Democratic	2,591,857	27.98%
Steve Hilton	Republican	2,277,318	24.59%
Tom Steyer	Democratic	2,110,707	22.79%
Chad Bianco	Republican	941,708	10.17%
Katie Porter	Democratic	403,908	4.36%
Matt Mahan	Democratic	327,537	3.54%
Antonio Villaraigosa	Democratic	104,107	1.12%
Tony Thurmond	Democratic	63,762	0.69%

The remaining 60 candidates split 441,564 votes, the last 4.77 percent. Becerra's 27.98 percent looks weak in isolation, and every model on the panel said so before explaining why it is the wrong number to read. The right one is the aggregate: Democratic candidates combined for 62.38 percent (5,777,936 votes), Republican candidates for 35.95 percent (3,329,636), and everyone else for 1.67 percent (154,896). That 26.4-point party gap was not a survey of 900 people. It was 9.26 million Californians filling out the ballot for this exact office, and it is the single number the panel weighted most. One honest caveat travels with it: the June and November electorates are not the same people. November turnout is larger, and its composition differs from a primary's, so the aggregate is a strong prior with a real error bar rather than a projection of the general-election margin.

The Board

Outcome	Party and status	Market	AI blend
Xavier Becerra	Democrat; former U.S. secretary of health and human services (2021-2025) and attorney general of California (2017-2021)	95.75¢	96.4%
Steve Hilton	Republican; political commentator, former adviser to UK prime minister David Cameron (2010-2012)	4.25¢	3.0%
Neither is elected	Residual for death, withdrawal, or disqualification before November 3	—	~0.6%

Unlike a wide field board, this one needs no re-scaling. The two midpoints sum to exactly 100.00 cents, so the quoted prices already read as probabilities. Buying both sides at the ask would cost 100.1 cents, which is the entire overround on this market.

Related Live Boards

Three sibling markets settle on the same election night or the same season, with prices fetched in this run:

• U.S. House control 2026: Democrats 83.5¢ (our model verdict)
• 2026 balance of power: D-House plus D-Senate 45.5¢ (our model verdict)
• Brazil presidential election: Lula 65.5¢ (our model verdict)

Every Seat's Number

Eight models scored both contracts price-blind on August 3, 2026, then read the other seven's anonymized numbers and reasoning and were free to revise. The numbers below are the revised, round-two estimates; the blend is the simple mean of the eight.

Outcome	Market	Fable	Opus	Sonnet	GPT	Gemini	GLM	Kimi	DeepSeek	Blend
Xavier Becerra	95.75¢	96	96	96	95	97	96	97	98	96.4%
Steve Hilton	4.25¢	3	3	3	4	3	3	3	2	3.0%

Every seat on this panel is graded against real market settlements — records to date: Claude Fable 85% on 331 graded calls · Claude Opus 86% on 382 graded calls · Claude Sonnet 84% on 363 graded calls · GPT 83% on 1,693 graded calls · Gemini 85% on 1,718 graded calls · GLM 80% on 1,874 graded calls · Kimi 84% on 1,863 graded calls · DeepSeek 80% on 1,877 graded calls. Recomputed daily; the full scoreboard is public.

These are model estimates, not predictions of fact and not financial advice. Market figures are Kalshi bid/ask midpoints fetched at 17:31 UTC on August 3, 2026 (Becerra 95.7/95.8, Hilton 4.2/4.3). Every number in this piece gets graded in public once the market settles; you can check the method against past calls on the full graded scoreboard.

All eight seats ran, none was dead, and none saw a price. The data they worked from was a fetched card carrying the certified primary results in full, four dated general-election head-to-head polls with sample sizes and populations, six race ratings with the date each was issued, California's last four statewide results, the state's Republican statewide record, primary-cycle campaign finance, and the exchange's own settlement language.

Xavier Becerra: The Favorite Whose Case Is Not The Poll

Becerra is a former attorney general of California and a former U.S. secretary of health and human services, and he finished first in June. None of that is why the panel put him at 96.4 percent. Every seat led with the same structural argument, and Opus stated the reasoning most directly:

"The primary aggregate (D 62.4% vs R 36.0%, 9.26M votes) drives my number most; I discount the individual 27.98%/24.59% finishes as artifacts of a 68-way split, not weakness." — Claude Opus (86% on 382 graded calls)

The supporting evidence lines up behind it rather than adding anything new. Four public general-election head-to-heads all land in the same band: Berkeley IGS had Becerra 52 to 31 among 8,578 registered voters (May 19-24), CEPP had 58 to 35 among 735 likely voters (May 23-26), Kreate Strategies had 58 to 33 among 900 likely voters (June 13-17), and PPIC had 61 to 36 among 1,003 likely voters (June 29 to July 6). That is a 21 to 25-point spread across four pollsters, three of them established and one unrated in our reliability archive. Their disclosed clients are institutional rather than partisan: the CEPP survey was sponsored by the CSU Long Beach, USC and Cal Poly Pomona elections consortium and the Berkeley IGS survey by the Los Angeles Times, while the VoteHub polling record for Kreate Strategies lists no sponsor, no partisan flag, and no internal-poll flag. Sponsor labels can differ between aggregators, so treat that as the disclosed record rather than a clean bill of health. California's last four statewide verdicts run in the same channel: Newsom by 23.9 in 2018, the 2021 recall defeated 61.9 to 38.1, Newsom by 18.4 in 2022, and the 2024 presidential race by 20.1. All six race raters that have issued a call have this on the Democratic side of the line at Solid or Safe, though several of those ratings predate the June matchup and are not post-primary judgments.

The panel also refused to treat the national midterm environment as a counterweight, which is worth understanding because it usually is one. The president's party has lost House seats in 18 of the last 20 postwar midterms, and that base rate would normally cut against the incumbent party in a governor's race. Here the White House is Republican-held, so the same headwind points at Hilton's party rather than Becerra's. Sonnet put the conditional structure plainly:

"The 1998/2002 exceptions required approval above 60 or a dominant rally event, neither present, so the environment reinforces rather than offsets California's blue lean." — Claude Sonnet (84% on 363 graded calls)

Steve Hilton: What A 4-Cent Contract Is Actually Pricing

Hilton is a first-time candidate, a political commentator, and a former director of strategy to UK prime minister David Cameron. He finished second in June with 24.59 percent, which under the top-two system was enough to reach November, and he outraised Becerra during the primary: $7,332,985 to $6,051,826 through May 31, 2026, per Cal-Access, with more cash on hand at that date ($1,724,547 to $793,630). None of that moved the panel, and the reason is arithmetic rather than assessment. Second place in a top-two primary is a ticket to the general election, not a measure of general-election strength, and while Hilton was finishing second, the combined Republican vote finished 26.4 points behind the combined Democratic vote in the same count.

The panel's exception test is where this gets specific. Republicans were last elected to a statewide office in California in 2006, when Arnold Schwarzenegger was re-elected as governor and Steve Poizner was elected as insurance commissioner. Twenty years is a long enough drought to be worth interrogating rather than reciting, so the models were asked to check whether the conditions that produced 2006 exist now: recall mechanics, pre-existing statewide fame, an incumbent's record, and explicit distance from the national party. Every seat concluded they do not apply to a first-time candidate in a regular November election, and none of the eight found a positive indicator pointing the other way.

That leaves the honest question of what the remaining probability is for. Kimi decomposed it rather than waving at uncertainty:

"My 3% tail decomposes into: roughly 1-1.5% for a Becerra-removing event, and roughly 0.5-1% for a legitimate upset, which demands a correlated 20-plus-point polling failure beyond the largest modern blue-state precedent, plus hidden-information risk since the newest poll is five weeks stale." — Kimi (84% on 1,863 graded calls)

[Editor's note: PPIC's poll finished fielding on July 6, which is roughly four weeks before publication rather than the five the quote states. Model rationales are reproduced verbatim.]

That first component is the one the ballot structure creates. Becerra was born on January 26, 1958, which makes him 68, and no replacement candidate or write-in is possible under Proposition 14. The panel's tail is therefore mostly not a forecast about the campaign at all.

[Editor's note: several seats reasoned that a candidate leaving the ballot would default the office to the other name. California Elections Code section 8803, quoted above, does not work that way: the departed candidate stays on the ballot, votes for that candidate are counted, and a majority means that candidate is considered elected with the office treated as vacant at the start of the term. The panel's quoted reasoning is reproduced as given; readers should apply the statute rather than the inference, which most likely makes the candidate-removal component of the tail smaller than the seats assumed, not larger.]

Where The Panel Changed Its Mind

Three of the eight seats revised between rounds, and all three moved on the tail rather than the favorite. The arcs are worth reading because they run in opposite directions and both are defensible.

GLM and Fable both came into round one with Hilton at 5 and cut to 3 after the same argument landed. Fable named it precisely:

"Forecaster H's midterm-specific poll-error point: the two Trump-era midterms were nearly unbiased (-0.3, +0.9), so my 5% tail was padding for a presidential-year polling bias that doesn't apply here. I shrank the tail accordingly." — Claude Fable (85% on 331 graded calls)

Gemini moved the other way, and its revision is the more interesting one because raising your own uncertainty after reading a confident room is the harder move:

"I am revising my tail from 1% to 3% based on the timeline. A 1% tail effectively prices out non-linear shocks over a three-month horizon, which is too confident." — Gemini (85% on 1,718 graded calls)

Five seats held firm, and the most useful hold came from GPT, which stayed at 4 and explained why the room's convergence was not evidence:

"The other forecasts reinforced the same evidence rather than adding a mechanism I had underweighted; I still think a 4% Hilton tail better prices low-probability shocks than 1-2%." — GPT (83% on 1,693 graded calls)

The convergence round narrowed the panel's spread on Hilton from 1-to-5 down to 2-to-4 without moving the blend much at all, which is roughly what a well-behaved revision round should look like: less disagreement about the size of a tail, no stampede on the headline number.

Dark Horses And The Names Already Off The Board

A two-name ballot has no dark horses in the usual sense, and pretending otherwise would be the kind of manufactured drama this format exists to avoid. The underpriced story on this event is the field that is already settled, because those contracts explain how the market got here.

• Tom Steyer (settled No) was the market favorite as recently as May 1, trading at 44 cents against Becerra's 38. He finished third at 22.79 percent, roughly 166,000 votes behind Hilton, and his campaign reported spending $166,766,233 through May 31, which works out to about $79 per vote received. Becerra's primary spending came to roughly $3.58 per vote. Money bought Steyer a competitive market price and not a place on the November ballot.
• Chad Bianco (settled No) was the second Republican, and his 10.17 percent is the number that made Hilton's second place possible: had the Republican vote split more evenly, two Democrats could have advanced and this market would have no Republican contract at all. Bianco traded at 5.5 cents on May 1.
• Katie Porter (settled No) entered the year as one of the best-known names in the field and finished fifth at 4.36 percent, trading at 4 cents on May 1 and 0.1 by June 10. She endorsed Becerra after the primary.
• The residual, blended at roughly 0.6 percent, is where the panel parks the outcome in which neither live contract pays. Read it as a small allowance for legal and settlement edge cases rather than a described scenario, and note that section 8803 narrows those cases further than the panel assumed.

Here is the part of the price history that got no national coverage. Between June 1 and June 10, Becerra went from 69 cents to 89 while Steyer collapsed from 20 cents to 0.1. Hilton over the same stretch went from 9.0 cents to 10.0. The largest repricing this market has ever seen moved money entirely within the Democratic field, and the Republican contract sat still through all of it. Hilton's decline from 10 cents to 4.2 came later and slowly, across July, as the first general-election polls of the actual matchup arrived.

The Traps: Numbers That Can Mislead

• 27.98 percent is not a weak showing. Becerra's primary share looks small next to Hilton's 24.59, and read as a two-way number it would suggest a close race. It is a 68-way number. The comparison that carries information is 62.38 against 35.95, and every seat on the panel made that substitution before doing anything else.
• Outraising your opponent in a top-two primary proves less than it looks like. Hilton raised more than Becerra through May 31 and finished second anyway, and Steyer outspent Becerra by nearly 18-to-1 and finished third. Those figures are also primary-cycle only, reported as of May 31, 2026, and say nothing about general-election fundraising since June.
• Four polls is a thin record, and half of them predate the matchup. Only two of the four head-to-heads were conducted after the June 2 primary made the field final: Kreate Strategies, June 13-17 and unrated in our reliability archive, and PPIC, June 29 to July 6. The other two tested a hypothetical pairing before the voters had produced it. PPIC is the most recent read and is now roughly four weeks stale. Kimi flagged this as its own biggest uncertainty, and it is a fair one: the polls agree with each other, but there are not many of them.
• A 3 percent tail is not a claim the race is competitive. Most of it prices candidate-removal risk on a ballot that permits no substitution, not a campaign closing a 21-point gap. Reading it as "Hilton has a puncher's chance" inverts what the panel actually said.
• Ratings dates matter. Three of the six race ratings on this seat (Inside Elections, Sabato's Crystal Ball, and Cook) were issued in 2025, before the primary, so they rate the seat rather than the matchup. The three post-primary ratings (RealClearPolitics June 5, Race to the WH June 19, Fox News July 23) all landed in the same place. Treat them as background, not as post-primary judgments.

Settlement Timeline

Date	What happens
June 2, 2026	Top-two primary held. 23 of the event's 25 contracts settled to No. Certified results published July 10, 2026.
November 3, 2026	Election day, 92 days from this article's publication.
January 4, 2027	The governor's term begins on the first Monday after January 1. Kalshi's listed expected expiration is January 6, 2027, and its listed early-close condition is the swearing-in. The contract's outcome turns on who is elected pursuant to the 2026 election; when the market closes and settles is governed by Kalshi's own close and settlement process, not by election night.
Next known catalysts	The next PPIC or Berkeley IGS statewide survey, general-election campaign finance reporting covering the period since June, and any scheduled general-election debate.

This page re-scores when the story moves. The as-of stamp on the price table carries the reading each number reflects.

What Would Change The Panel's Mind

The panel was asked for falsifiers rather than hedges, and these are the concrete, checkable ones the seats named:

• A high-quality post-Labor Day poll inside 12 points. Opus set this threshold explicitly; Kimi set it at 15. Either would mean the four-poll consensus was measuring something that has since changed, and would push Hilton's number up materially.
• A Becerra-removing event. Health, withdrawal, or disqualification. This class of event carries roughly a third to a half of the panel's entire tail, though as section 8803 above makes clear, it does not automatically transfer the office, so the panel's sizing of it is probably generous.
• A sharp national shift. Several seats named the same measurable trigger: the generic congressional ballot moving from D+5 toward even or Republican-leading, or presidential approval recovering well off its current net -18. That would not flip California on its own, but it would widen the honest error band on every state race.
• A general-election money gap. The published finance figures stop at May 31 and cover only the primary. A large post-primary resource advantage in either direction is the next real data point, and none of the models had it.
• What would not change it: another poll in the 21-to-25 band, another rating held at Solid D, or campaign news that does not attach to one of the four items above. The panel has already priced all of that.

How To Read These Prices

A 95.75-cent contract is a strange thing to look at if you are used to sports odds, so it is worth being concrete about what it is.

Work an example. Becerra's contract costs about 95.75 cents at the midpoint and pays $1 if he is elected. Risking 95.75 to make 4.25 is roughly -2,253 in American odds, and you are tying that capital up into 2027. Hilton's 4.25-cent contract is the exact mirror at about +2,253, with the panel saying the true number is nearer 3. Those are midpoint figures; crossing the spread to actually get filled makes both sides slightly worse. The panel's disagreement with the market on Hilton is real but small in absolute terms, roughly 1.25 cents on a 4.25-cent contract, and a 1.25-cent edge is well inside what the spread, the fees, and the months of tied-up capital can eat. Reading a price as a probability and then asking whether the edge survives its costs, which is the habit our guide to reading Kalshi odds walks through, matters more on a lopsided board than on a close one. OddsShopper's odds tools cover sports markets rather than political futures, so for the weekly slate our free expert picks today and live odds screen do the price-comparison work, while this contract trades on Kalshi.

A few honest points before reading too much into any number above:

• Eligibility. Kalshi is a CFTC-regulated event-contract exchange, not a sportsbook. You must be 18 or older and located in an eligible state, and availability varies by state and can change. Our explainer on how Kalshi's election markets work covers the mechanics of this category specifically.
• Liquidity is not symmetric. The Hilton contract has traded more volume (6,854,035 contracts) and carries more than twice the open interest (3,367,828) of the Becerra contract (4,618,129 and 1,500,165). Long-shot contracts often carry the heavier book on a lopsided board. Volume and open interest describe past activity rather than the depth available at this moment, so treat that as context, not as a promise about what size will fill.
• Time and capital. This event's expected expiration is January 2027, so a position here is months of tied-up capital with no interim payout.
• Not advice. Everything here is analysis for information, not a recommendation to trade, and not a political endorsement. These are model estimates, not predictions of fact and not financial advice.

The Bottom Line

The market and the models are looking at the same race and arriving in the same place, which is itself the finding: at 96.4 percent against a 95.75-cent price, this is one of the cleanest agreements our panel has produced. The evidence behind it is unusually good for a race three months out, because California's top-two primary handed everyone a certified 9.26-million-vote measurement of party preference on the same statewide office, and it read 62.38 to 35.95. Four polls, six ratings, and 20 years of statewide results all point the same direction, and the national midterm environment points there too rather than against it. What is left to argue about is the tail, and the panel argued about it well, narrowing from a 1-to-5 spread to 2-to-4 while establishing that most of what remains is a closed-ballot structural risk rather than a live campaign.

If this method is your thing, our 2026 U.S. House control model verdict and our 2026 balance of power model verdict score the same election night the same price-blind way, and our Brazil 2026 presidential model verdict does it for the year's other big national race.

Hottest Prediction Markets Right Now

• 2028 Democratic presidential nominee · $166M traded
• Rocket Classic Winner · $124M traded
• 2027 Pro Football Champion · $51M traded
• Pro Baseball Champion · $40M traded
• Bitcoin price at the end of 2026 · $30M traded
• 2027 Pro Basketball Champion · $19M 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.