Updated August 25, 2026 · 15 min read · by Sam Smith
Yesterday, Rep. Al Green filed another article of impeachment against Donald Trump, accusing the President of turning federal immigration agencies into "his vicious, quota-driven, unaccountable paramilitary police forces" and of "devolving the United States democratic republic into an authoritarian regime, where unjustified killings and cover-ups by ICE and CBP are normalized." It was at least his sixth such filing, and the House has voted to table every one that reached the floor.
In April, the ranking Democrat on the House Judiciary Committee had asked the President's own doctor to run a full neuropsychological assessment and release the results publicly. Four days after that, Rep. Jamie Raskin and 50 House Democrats introduced a bill to create a 17-member Commission on Presidential Capacity, staffed with physicians, psychiatrists and former statespeople, to judge whether the President can do the job. "Public trust in Donald Trump's ability to meet the duties of his office has dropped to unprecedented lows," Raskin said, warning that the country sits on a "dangerous precipice."
The White House answered at the end of May. Its physician's memo, reported by TIME, recorded "strong cardiac, pulmonary, neurological, and overall physical function," a perfect 30 out of 30 on the Montreal Cognitive Assessment, and a conclusion that the President is "fully fit to carry out all duties of the Commander-in-Chief and Head of State."
All of it has been running alongside a war. American forces have been fighting Iran since February, Congress never authorized it, and both chambers have now formally told the President to stop: the Senate 50 to 48 on June 23, the House 214 to 208 on July 23. Neither vote binds him.
Those are the conditions traders are being asked to price against. And yet the market on whether Trump will leave office early has spent the whole year going the other way. Start with what has actually happened on the floor of Congress, because that is where the answer lives.
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.
The Record Is Loud. The Roll Calls Are Empty.
Here is the part the volume of news does not prepare you for.
The last time the House of Representatives voted on impeaching this president was December 11, 2025. Al Green forced H.Res. 939 onto the floor. The House did not vote on the articles. It voted to make them go away, and the Clerk's own tally file records what that looked like: 237 to table, 140 against, 47 voting "present." Republicans went 214 to nothing to set it aside. Every one of the 140 nays was a Democrat, and so was every one of the 47 "presents," and so were the 23 members who crossed over to help table it.
Since then, nothing. We pulled every recorded House roll-call vote of 2026 from the Clerk's own vote files, all 283 of them, January 6 through July 23, and searched every vote description for the word "impeach." Zero hits. The House has not recorded a vote of any kind since July 23. That sweep covers floor votes only, so it says nothing about how many resolutions have been introduced or referred, and yesterday's filing is a reminder that introductions keep coming. What it does tell you is that in a year of physicians' letters, war-powers rebukes and majority disapproval, the mechanism itself has not been used once.
The war votes point the same way once you look past the headline. Four Republicans crossed in the Senate on June 23: Bill Cassidy, Susan Collins, Rand Paul and Lisa Murkowski. Four crossed in the House on July 23: Tom Barrett, Brian Fitzpatrick, Warren Davidson and Thomas Massie. Both resolutions were non-binding, which makes them the cheapest possible vote a Republican could cast against this President.
And the crossover did not survive the week.
"These Senators have just made my job more difficult." — Donald Trump, on the four Republicans who voted to rein in his Iran war powers
Trump called those four senators "losers." The next night, the Senate reversed itself: Paul switched to "present," Cassidy switched to no, and a second war-powers resolution was blocked 47 to 50, with one senator voting present. Four defectors on a symbolic vote, two of them gone inside 24 hours, in a chamber where conviction needs 67.
Public opinion is not the constraint here and never has been. A Washington Post-ABC News-Ipsos poll in late April found 59% of Americans saying Trump does not have the mental sharpness to lead and 55% saying he is not in good enough physical health for the job. None of those people gets a vote in a Senate impeachment trial or a seat in the Cabinet room. Every path out of the presidency is operated by officeholders, and the officeholders have moved four seats, twice, on votes that cost them nothing.
The Four Contracts, And What Each One Actually Pays
Kalshi is a CFTC-regulated exchange where each outcome trades as a yes-or-no contract priced between one cent and 99 cents, settling at a dollar if it happens and zero if it does not. The price is the crowd's live probability: a 23-cent yes contract means the market is treating it as roughly a 23% chance. Kalshi is 18+ and availability varies by state.
New to any of this? Kalshi sports betting in four steps walks through the mechanics before you read a price.
So it is worth knowing what people betting real money think happens next, because their answer is stranger than either side of the argument. And to read it you need to know that there is not one Trump-departure market. There are four, and they pay on different things.
- Impeached. Pays if a simple majority of the House votes articles of impeachment. That is all it takes, and it does not remove anybody. Settles Jan 20, 2029.
- Impeached And Removed. Pays only if the House impeaches and the Senate then convicts by a two-thirds vote, 67 of 100. Settles Jan 20, 2029.
- Resigns. Pays if Trump resigns the office. Settles at the end of the term.
- Leaves Office Early. Pays if he leaves for any reason other than death, or if he publicly says he will leave within the next year. This one has three deadlines: 2027, 2028 and Jan 20, 2029.
That last one is what this piece is about, and its fine print carries two clauses most people never read.
The first clause widens it. The contract's own text says an announcement "that the President will leave the office within the next year is also encompassed by the Payout Criterion." He does not have to be gone by the deadline. He has to have said it by the deadline. Acknowledging that his term simply expires does not count.
The second clause narrows it, and it narrows it hard. If Trump leaves office "solely because they have died," the contract does not pay yes. It settles at the last traded price before the death, or at a committee's determination of fair value. Nine presidencies have ended early, and eight of them ended with the president's death. For an 80-year-old sitting president, that carve-out removes the reason almost every early exit in American history actually happened. Anyone pricing this ladder off mortality tables is pricing the wrong contract.
Four Contracts, One President, And The Numbers Do Not Add Up
Now the prices, all read from Kalshi's book at 1:30 PM ET on August 25, 2026.
The market thinks Trump probably gets impeached. The contract on the House voting articles before his term ends is bid 64 cents and offered at 65. The contract on the House impeaching and the Senate convicting is bid 13 cents and offered at 14. That 51-cent spread between the two is the two-thirds Senate, priced. Our piece on why that wall holds at 67 votes walks the arithmetic in full.
Set impeachment aside, then, because the market already has. What the departure ladder is really pricing is Trump choosing to go.
| Deadline | What the market pays | The panel's number |
|---|---|---|
| Before Jan 1, 2027 | 5¢ (bid 4.9 / ask 5.1) | 1.0% |
| Before Jan 1, 2028 | 17.5¢ (bid 17 / ask 18) | 4.0% |
| Before Jan 20, 2029 | 23¢ (bid 22 / ask 24) | 6.7% |
Prices fetched August 25, 2026. Panel numbers are the seat median of eight models, generated without showing any model a price. Model estimates, not predictions of fact and not financial advice.
More live boards from the same panel: Trump impeachment and the Senate wall, where conviction is bid 13¢ against an impeachment contract at 64¢; will Trump resign, bid 19¢ on the narrower resignation-only question; and the impeachment board itself, which prices a House vote at better than even money. Prices fetched August 25, 2026.
And the departure ladder has been fading all year. Kalshi's daily closing prices, pulled for this piece, put the full-term contract at 49 cents on February 13 and 21 cents on Monday. The 2028 rung peaked at 38 cents on February 14 and closed Monday at 16. The near rung topped out at 20 cents on February 10 and closed at 5. Every rung has more than halved while the record above it got louder.
Read those numbers together and something breaks. Resignation is bid 19 cents. Impeachment-and-removal is bid 13. Neither one can happen twice, they mostly cannot both happen, and either one on its own would settle the departure contract yes. Stack them loosely and you get something near 32 cents' worth of ways for Trump to be gone before January 20, 2029. The contract that pays on both of those, plus a mere announcement, plus any other non-death exit, is bid 22.
Three separate order books do not have to agree, and these three plainly do not. The honest caveat is that all of them are thin day to day: the 2029 departure rung traded 49 contracts in the last 24 hours, resignation 566 and removal 314, against lifetime volumes in the hundreds of thousands. Nobody is arbitraging these against each other. But the direction of the inconsistency is worth sitting with, because it says the widest contract on the board is also the cheapest one.
The market's shape tells its own story. Of the 23 cents on the full-term contract, five sit in the next four months, twelve and a half sit inside calendar 2027, and five and a half cover all of 2028 and the first three weeks of 2029. Traders are concentrating more than half the risk in a single year, and it is the year after the midterms. That is coherent: the House flips on November 3 or it does not, and Kalshi's own board has Democrats bid 85 cents to take it. The impeachment ladder agrees. It sits at 4 cents for a House vote before January 1, 2027, jumps to 20 cents for one before March 1, 2027, which is the window that opens the moment the new Congress is seated on January 3, and reaches 58 cents by January 2028.
Every seat's number. Eight models were shown the contract text, the Clerk's vote records, the war-powers votes, the fitness record and the White House's rebuttal, the polling frontier and the full presidential precedent ledger. None of them saw a price.
| Seat | Before 2027 | Before 2028 | Before Jan 20, 2029 | The line each seat filed with its number |
|---|---|---|---|---|
| Claude Fable | 1.6% | 6.0% | 9.5% | "Resignation/announcement is the only live channel; age-80 fitness fight and an unauthorized war lift it well above the 2% ledger base, but 67 votes stays unreachable." |
| GPT | 1.2% | 5.8% | 9.5% | "I hold: peers sharpened the downside case, but no new mechanism beats the announcement/resignation path plus hostile-House pressure I already priced." |
| Claude Opus | 1.0% | 4.5% | 7.5% | "Resignation or a self-timed announcement is the only live path; 67 Senate votes are not adjacent to four crossovers." |
| Gemini | 0.8% | 3.4% | 6.8% | "Impeachment faces a 67-vote Senate wall, and Section 4 divests power without vacating office; thus, the only viable non-death exit remains a coerced or health-forced resignation." |
| Claude Sonnet | 0.7% | 3.2% | 6.5% | "No peer surfaced a new mechanism or contract reading; the board's spread is all in how much weight to give noise (letters, polls, rebukes) around an unchanged 67-vote wall." |
| GLM | 0.7% | 3.2% | 6.5% | "No peer surfaced a mechanism or contract reading I missed; the 67-vote wall and death carveout remain binding, and the announcement path still lacks any trigger evidence." |
| DeepSeek | 1.2% | 4.5% | 6.0% | "Holding: no new evidence surfaced that changes the mechanism gap—67 Senate votes remains the wall, and the announcement path is already priced." |
| Kimi | 1.0% | 3.0% | 5.0% | "Conviction needs 67 senators against an observed crossover of four; nearly all live probability sits in the health-driven resignation/announcement path, which compounds over 878 days at age 80." |
| Panel Blend | 1.0% | 4.0% | 6.7% | seat median of all eight, method always disclosed |
Every seat on this panel is graded against real market settlements — records to date: Claude Fable 84% on 709 graded calls · Claude Opus 84% on 779 graded calls · Claude Sonnet 82% on 757 graded calls · GPT 84% on 6,053 graded calls · Gemini 86% on 4,760 graded calls · Kimi 83% on 2,494 graded calls · GLM 81% on 2,504 graded calls · DeepSeek 80% on 2,542 graded calls. Recomputed daily; the full scoreboard is public.
Model estimates generated August 25, 2026, price-blind, from a fetched data card, after a revision round in which every model read the others' anonymized reasoning. These are model estimates, not predictions of fact and not financial advice. Models are frequently wrong, and the market price reflects real traders' money. Every number above is stored and graded when this market settles, and our running record on every event we have priced sits on the full graded scoreboard.
All eight seats landed below the market on all three rungs. The panel's disagreement with traders is not about whether Trump is under pressure. It is about whether pressure has a mechanism attached to it.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $193M traded
- 2027 Pro Football Champion · $72M traded
- 2028 U.S. Presidential Election winner? · $62M traded
- Pro Baseball Champion · $60M traded
- 2028 Republican presidential nominee · $60M 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.
Where The Panel Changed Its Mind
After the first round, every model read the others' reasoning with the names stripped off and got one chance to revise. Both rounds are stored. Three seats moved, and what they wrote on the way is more interesting than the numbers.
Gemini cut its full-term figure from 7.8% to 6.8% after correcting a reading of the Constitution. In round one it had filed this:
"Senate math forms an absolute wall against removal, leaving health-driven resignation or its announcement as the primary mechanism, which scales slowly over time." — Gemini, round one
In round two it added the piece it had been missing, which is that the 25th Amendment's Section 4 does not actually empty the office:
"Impeachment faces a 67-vote Senate wall, and Section 4 divests power without vacating office; thus, the only viable non-death exit remains a coerced or health-forced resignation." — Gemini, round two
Claude Fable, the panel's highest seat, came down from 11% to 9.5% while explicitly holding onto the reasons it had gone high in the first place:
"Resignation/announcement is the only live channel; age-80 fitness fight and an unauthorized war lift it well above the 2% ledger base, but 67 votes stays unreachable." — Claude Fable, round two
Kimi went the other way, the only seat to raise its full-term number, from 4.5% to 5%, on an argument about time rather than about politics:
"Conviction needs 67 senators against an observed crossover of four; nearly all live probability sits in the health-driven resignation/announcement path, which compounds over 878 days at age 80." — Kimi, round two
GPT held at 9.5% and said so in as many words, which is the more useful kind of stubbornness:
"I hold: peers sharpened the downside case, but no new mechanism beats the announcement/resignation path plus hostile-House pressure I already priced." — GPT, round two
What Would Change The Panel's Mind
The panel's number is low. It is not fixed. Four things would move it, and the seats mostly agree on which one matters most.
A named Republican senator says publicly they would vote to convict. Three seats independently called this the single highest-impact development. The whole low number rests on a crossover ceiling of four, on votes that carried no legal force, two of which reversed the next night. One senator saying the word "convict" on the record reprices the wall itself. Pushes the number up, hard.
A visible medical episode, or a physician's memo that concedes impairment. Not death, which this contract does not pay on, but an event that makes a managed handover the dignified option. Every seat named some version of this as the live path. Up.
A Democratic House seats in January 2027 and opens fitness hearings. Kalshi has Democrats bid 85 cents to take the chamber on November 3. Subpoena-armed hearings do not produce 67 senators, but they lengthen the window in which a self-timed exit announcement becomes thinkable. Up, modestly, and only on the 2028 and 2029 rungs.
Nothing happens through the midterms. If November 3 passes and the December quarter closes with no crossover, no hearing calendar and no health event, the near rung expires worthless and the remaining hazard compresses into a shorter window against a term that is already ending. Down.
When This Settles, And What To Watch
| Date | What it is |
|---|---|
| November 3, 2026 | Midterm elections. Kalshi has Democrats bid 85 cents for the House, Republicans bid 53 for the Senate |
| January 1, 2027 | First rung settles. It expires worthless unless he is gone or has announced by then |
| January 3, 2027 | The new Congress is seated. The chamber that could impeach starts here |
| January 1, 2028 | Second rung settles |
| January 20, 2029 | Final rung settles at the end of the term |
One rung on this ladder has already settled. The contract on Trump leaving before August 1, 2026 resolved NO on the day it expired, after trading at three tenths of a cent. This page re-scores whenever the story moves, and every number on it carries the August 25, 2026 stamp above.
- What Will Trump Say This Month? 12 Words Left, One Is Epstein
- Does Trump Leave Office Early? Kalshi's Departure Odds
- What Will Trump Say This Week? Kalshi Odds
- What Will Trump Say? Kalshi Mention Market Odds
- Will Trump Be Impeached? What The Market Says Right Now
The Bottom Line
Traders have built a coherent, slightly uncomfortable picture: a president who probably gets impeached, almost certainly does not get convicted, and leaves early only if he decides to. On the bids, the resignation contract at 19 cents is roughly 85% of the departure contract at 22, which is another way of saying that most of what the departure contract is paying for is Trump's own choice, not anyone else's vote.
Eight models, working from the contract text and the roll calls rather than the headlines, think even that is priced about three times too high. Their case is one sentence long. A documented record of letters, floor rebukes, commission bills and majority disapproval is real evidence about the environment, and none of it is a mechanism. The mechanism is 67 senators or one man's decision. The Cabinet route does not even qualify: invoking the 25th Amendment's Section 4 makes the Vice President Acting President, which strips the powers and leaves the office occupied, so it would not settle this contract yes at all.
The thing to watch is not the next physician's letter. It is the first Republican senator who says the quiet part with their own name attached, and then does not take it back the following night.
