Amidst the recent turmoil between the federal government, Anthropic, and its rivals, I’ve seen a number of commentators point out something interesting. The private ownership of Anthropic and OpenAI means that far fewer Americans have a direct stake in their success than in other firms of their size and centrality to the economy.

If the Pentagon had attempted to label, say, Walmart or Exxon Mobil a supply chain risk, then the tens or hundreds of millions of Americans with direct stakes in the stocks of those companies would feel the effects, regardless of their personal affinities towards them.

This might change soon. Anthropic, OpenAI, and other AI companies are rushing to IPO as soon as this year. They will skyrocket into the top echelon of publicly traded companies if and when this happens. It wasn’t always obvious that they’d choose to press for a fast timeline. With their ability to attract the largest private investment rounds in history, it wasn’t clear that going public was as necessary as for other capital-intensive private companies.

But, with secondary markets clamoring for access to shares, and capital expenditures for hyperscalers rising just as precipitously as revenue, IPOs in the next 12 months are likely. Internal pressure from employees with equity and early VCs won’t hurt. Manifold traders are quite interested in the topic as well…

For a while, the race to IPO for American private AI firms looked like more of a three-horse race, but the merger of xAI and SpaceX put that to rest, with SpaceX/xAI now very likely to go public within the next few months.

While many observers might fairly think that xAI isn’t really in the same league as the other two companies, they did in fact raise money earlier this year at about half the valuation of Anthropic (about $250 billion). And now, packaged with SpaceX, they might be more valuable than both.

But, ummmm, it’s not clear how long this will hold.

Valuations

Just like the intelligence of their models, their number of employees, their revenue, and the number of protestors outside their offices, the valuations of these companies (check out MNX) seem to be doubling every few months.

Epoch has had great success drawing straight lines on log-plots, and they called Anthropic surpassing OpenAI’s revenue in early 2026.

With their latest run-rate revenue forecast of $30 billion a few days ago, this market resolved YES, sending markets on whether Anthropic would flip OpenAI skyrocketing.

The market on combined EOY revenue for America’s three private AI companies had to create some new buckets in the last week, as the top bound was around $150 billion, which now looks to be below the median trader’s expectations. Two Twitter users (at least one of whom is a top Manifold trader, I believe) made an over/under bet at $90 billion on just Anthropic’s run-rate revenue, and we now have a market on it:

Much of the recent bullishness around Anthropic has been the disclosure of their Mythos model. Enough ink has been spilled about Project Glasswing and Mythos’s emergent ability to spot zero-day vulnerabilities, but nevertheless…

It’s a story that at the same time is extremely urgent for the public and policymakers while being slightly overhyped in breathless reporting. As keen observers have by now pointed out, previous models like Opus 4.6 and GPT-5.4 likely have abilities nearing this as well, with Mythos representing a continued climb up that particular capability curve, rather than a discontinuous jump. Skeptics have noticed that the amount of cost and compute thrown at discovering individual vulnerabilities during testing ranged up into the ballpark of tens of thousands of dollars, and mused that Opus 4.6, given that amount of time and energy, could probably find a similar vulnerability.

Obviously, without public access to the models, it’s hard to say. No one seriously doubts that Mythos is a highly capable model, and traders in fact think that sooner or later, Anthropic will release it to the public in some form:

And when it does, it will likely come in at about 10x the cost of other frontier models on a per-token basis.

And even if we don’t get Mythos, we’ll likely have access to something with similar capabilities within the next few months:

Bubble

Alternatively, many online and even some on Manifold hold a very different set of beliefs about how AI will play out over this year. While it’s hard to operationalize a market on what an AI bubble “popping” would look like, one creator has given it a good try:

The criteria for this market read:

This could manifest in many different ways:
- Sudden drop of the NQINTEL or S&P500 indexes
- Halt of capital expenditure and/or VC investment in AI
- Private AI companies (such as OAI, xAI, Anthropic, etc.) raising at substantially lower valuations than previous rounds, defaulting or shutting down.
- Governments / big enterprises moving down AI in their priority lists.
These events need to be widely reported and referred to as an AI crash or similar by trusted news sources such as Bloomberg, NY Times, FT, Reuters, AP, etc.
Other reasons for a stock market crash or economic slowdown (tariffs, wars, etc.) will not count.

Alternatively, another creator has simply made a market on which AI companies will drop dramatically from their peak this year. Apart from Oracle, traders think this is unlikely for most firms that are heavily leveraged in the AI boom.

Notably, the odds for a general recession—potentially caused by, umm, overseas conflict—are much higher than a crash localized to the AI sector:

Roundup

Not just the AI companies might go public this year. You can bet on Discord, Anduril, Skims, and Jersey Mikes too!

Is Mythos really AGI if it can’t speedrun Minecraft?

And we probably won’t know what Ilya has been up to by July:

Happy Forecasting!

-Above the Fold

Trump accused him of leaking classified information. He then directed Pete Hegseth (my bad - Secretary of War) to begin identifying and releasing government files related to “alien and extraterrestrial life, UAP, and UFOs,” citing “tremendous interest shown.”

Hegseth reposted the announcement with an alien emoji and a saluting emoji. Three days later, asked whether he personally believed in extraterrestrials, he said the task was not on his “bingo card” but he was “digging in” and would find out “along with all of you.” The Office of the Director of National Intelligence posted on social media that the files would “soon” be declassified.

Last week, the White House registered aliens.gov and alien.gov through CISA’s federal domain registry. Currently, neither connects to anything. The deputy press secretary responded to press inquiries with “Stay tuned!” and an alien emoji. Both domains were registered while CISA is not currently accepting new .gov requests, due to a lapse in federal funding.

The Pentagon’s All-domain Anomaly Resolution Office, AARO, was created in 2022 to investigate unidentified anomalous phenomena. It has reviewed over 2,000 cases. Every one traced back to drones, balloons, atmospheric effects, or sensor artifacts. Sean Kirkpatrick, its first director, said of the incoming file release: “Nothing would have made me happier in that job but to have discovered alien technology and rolled it out. I don’t expect to see anything new.” He described Trump’s announcement as “a distraction for the administration.”

The reason the interesting UAP evidence can’t be released is structural. The most compelling cases are compelling because classified sensors detected them: military radar, satellite tasking, signals intelligence. Releasing the footage means revealing what those systems can see, at what resolution, from what angle. Past disclosures have leaned on pilot reports and low-resolution cockpit video while withholding the metadata. The result is always redacted PDFs.

The disclosure community has been predicting imminent official confirmation since the Clinton administration. Obama, Hillary Clinton, Biden, Trump’s first term, Trump’s second term, and Pope Leo XIV have all been identified at various points as the figure about to initiate disclosure. In 2025, the Wall Street Journal reported that hundreds of Air Force personnel had been told, falsely, that a secret program to harvest alien technology existed. The disclosure community cited this as evidence of cover-up activity.

The most useful number in the Manifold alien market cluster is not any of the confirmation probabilities. “Will the US Government make a show of declassifying everything on UFOs/UAPs before 2030?” sits at 44%.

Markets around whether the US ACTUALLY confirms that aliens exist before 2027 hover around 7-8%. Traders think there’s a real chance of an elaborate public performance of transparency, but a much smaller chance the performance amounts to anything.

Traders are also forecasting specifically if the US confirms aliens in 2026, what kind will they be? The leading answer at 49% is Microbial life, with the runner up at 27% being “completely unspecified (US government confidently asserts aliens exist but provides no further information).”

There's also a market on whether the US President or UK Prime Minister will confirm non-human intelligence before end of 2026, sitting at 13% - Keir Starmer apparently included for completeness.

This sits alongside markets on whether the US President themself will confirm extra-terrestrial existence.

Moving on from the politics, we have “Will we have contact with intelligent alien life by 2030?” sits at 5%, roughly the same as the JFK assassination being linked to aliens before 2040.

“Will the first sign of extraterrestrial intelligence be identified as artificial intelligence by 2030?” is at 13%, which is a more interesting question than it gets credit for. “Alien invasion before GTA6?” is at 4%.

The most substantive alien science of the past year happened nowhere near Washington. In April 2025, a Cambridge team using the James Webb Space Telescope announced that the atmosphere of K2-18b, a water-rich planet 124 light-years away, appears to contain dimethyl sulfide, a molecule produced on Earth exclusively by marine microbial life. The detection reached 3-sigma confidence. The lead researcher called it “a revolutionary moment.” Four independent teams reanalyzed the data over the summer and found the signal was probably ethane, or red noise in the mid-infrared, or a real detection that needs more observation time. One analysis found 87.5% of retrieval approaches showed no statistically significant evidence for DMS at all.

“Will K2-18b be generally agreed to harbor alien life by 2027?” is at 5%, with a scientific consensus sitting at the same odds.

This arc is mostly familiar. The phosphine-on-Venus announcement in 2020. The 1996 Martian meteorite bacteria claim. All had similar beginnings. K2-18b received a fraction of the coverage of Obama’s podcast clip.

Harvard astronomer Avi Loeb has a public bet with skeptic Michael Shermer: $150,000 against $1,000 that alien visitation will not be confirmed by 2030. Markets broadly agree with him. The aliens.gov domain expires in 2027. So does the confirmation market. Neither currently has any content.

Happy Forecasting!
- Above the Fold

There is a particular category of political problem where the solution becomes the source of the next crisis. Britain is currently running several simultaneously.

First-past-the-post was designed to manufacture stable majorities from a fragmented electorate, compress minor parties out of relevance, and deliver a decisive two-party duopoly that Enlightenment architects (presumably) found comforting. The 2024 election gave Labour a 172-seat majority on 33.7% of the vote, which is the system working exactly as designed. The combined Labour and Conservative vote share that year was 57.4%, itself a record low. It has since fallen to 38.2%. The mechanism for producing stability is now producing a governing party held in place partly by fear of what replaces it, an opposition so fragmented it cannot agree on its own purpose, and projection models where Reform wins a majority of 112 seats on 30% of the vote because the left cannot coordinate around a single candidate. No party has polled above 30% in months.

Manifold gives 18% odds that any party consistently hits 35% this year. The system that was supposed to squeeze chaos into a majority is instead amplifying it.

The special relationship was meant to guarantee British influence in Washington: a privileged bilateral channel that would make up in diplomatic proximity what Britain had lost in post-imperial weight. It is now the leverage Trump uses against Starmer, a relationship so asymmetric that Britain ends up defending its right to be an ally rather than assuming it. “This is not Winston Churchill we are dealing with,” Trump told journalists, which lands decisively even if it describes nothing more than a specific disagreement about which bases to lend.

The bond market was meant to discipline governments that spent too freely. It has, for the past year, been the main thing keeping Keir Starmer in office, because gilt yields spike whenever it looks like he might be replaced by someone to his left. A Labour Prime Minister, historically unpopular with voters, sustained in position by the City’s preference for his fiscal conservatism over his successor’s probable alternative. If you wanted to design a scenario that made Labour’s internal politics look most absurd, it would be difficult to improve on that one.

Reform was meant to be a clean insurgency: anti-establishment, ideologically coherent, the vehicle for everyone who had decided the old parties were irredeemably broken. It has been in local government for eight months and already has 1 in 18 of its elected councillors gone through suspension, expulsion, or resignation. One shared a meme suggesting Hitler “would have been a legend” had he targeted a different group. One was charged with assault at a Pride event. One had been the council leader caught swearing at colleagues on video, which then got leaked. Six of the seven Kent councillors who defected to Rupert Lowe’s Restore Britain had already been expelled by Reform before they defected. Reform’s response: “Restore are welcome to our dregs.” The party originally elected 57 councillors to Kent County Council. It now has 48.

This is not the same observation as saying British politics is chaotic, an insight we’ve already covered in detail. It is the more specific observation that the chaos is being generated by the mechanisms supposed to prevent it.

The Teams Call

On Sunday 25 January, at 11am, Keir Starmer dialled into a Microsoft Teams meeting. He was about to speak with Zelensky. First, though, he personally led an 8-1 NEC vote to block Andy Burnham from standing as Labour’s candidate in Gorton and Denton, on the grounds that triggering a fresh mayoral election would drain resources ahead of May.

The seat itself had been triggered by Andrew Gwynne, the former health minister suspended from Labour after WhatsApp messages surfaced in which he called a constituent a “Duki Hag” and described her neighbourhood as a quote-unquote “shithole.” The seat had returned a Labour MP, in various boundary configurations, since 1931. Burnham, the most popular figure in Labour politics (net approval ±0, against Starmer’s -47), was blocked. Labour then deployed a leaflet from a tactical voting organisation called “Tactical Choice” recommending voters back Labour. No such organisation exists. Nobody appears to have checked this before the leaflet went out.

Labour still came third. Hannah Spencer, a 34-year-old plumber and Green councillor, won 40.7% on a 27.5-point swing - the largest Green by-election swing ever, and the party’s first by-election win. It was the 127th seat on their target list. The mechanism Labour used to protect Starmer from a leadership challenge produced the most embarrassing by-election result the party has suffered since 1931, in a seat the leadership explicitly chose to fight as a show of strength.

The Self-Demolition

The more analytically interesting question about Reform is not whether electoral headwinds will slow them down but whether they will manage to do it themselves first.

The governance record after eight months in local power is instructive. Having promised to eliminate “waste” and cut council tax, every Reform council is raising it. Worcestershire got special government permission to exceed the standard 5% cap, raising council tax by 9%. Kent’s own social care lead admitted “spending is down to the bare bones.” The forest of waste, when located, turned out to be a single shrub. The issue is structural: 48% of Kent’s budget goes on adult social care, which does not yield to populist promises about DEI spending. The easy observation is that Reform’s candidates did not know this when they were elected. The pointed one is that several of them had previously served under Labour or Conservative administrations and knew perfectly well.

The candidate quality issue has been persistent enough to quantify. At least 38 Reform councillors were suspended, expelled, or resigned within the first six months after the May 2025 elections, with at least four facing criminal investigations. The party ran a man as council leader in Staffordshire who had previously posted that a statue of a Black woman was a “fat a***d black woman,” and when it emerged, the internal email from the council leader defending him was accidentally sent to all 81 members of the council rather than just Reform ones.

Restore Britain now sits as third-largest group on Kent County Council. Six of its seven defectors were people Reform had already expelled. The party Rupert Lowe calls more principled than Reform has published a 133-page mass deportation policy document and lists Carl Benjamin (Sargon of Akkad) among its affiliated figures. One of Reform’s own spokespeople called Restore “neo-Nazi.” Musk called Reform “Nazis” in return. This exchange happened in public, in February, in the run-up to local elections. The right-wing vote splitting effect this produces is not primarily an electoral threat to Reform from Restore, which is probably too extreme to gain serious traction outside online ecosystems, although traders do think there is potential for them to win a seat or two.

On funding: Reform’s biggest donor is Christopher Harborne, a Thailand-based investor who has given £12 million over the past year, including the single largest living political donation in UK history at £9 million in one go. The Electoral Commission has warned it cannot verify the identity of Reform’s crypto donors because the party’s payment processor is a Polish-registered entity not regulated by the FCA. The Commission’s exact words: donations could be “unidentifiable” under UK law. This sits interestingly alongside the party’s central pitch about cleaning up British political corruption.

The Squeeze

British politics has reorganised around tactical voting in a way that is not entirely serving its intended purpose.

Labour’s theory entering the next election is straightforward: fear of Reform winning will squeeze Green and Lib Dem voters behind Labour candidates in marginal seats, replicating the 2024 mechanism that gave them their majority. YouGov’s tactical voting study this month complicated this. 57-58% of Lib Dem and Green voters would tactically back Labour to stop Reform in their seat. But only 43-44% would do so to stop the Conservatives.

Gorton was also the clearest signal yet that tactical voting is no longer reliably flowing toward Labour from the left. Three independent tactical voting organisations endorsed the Greens as best placed to beat Reform. Labour invented a fourth to endorse itself. The Greens won by 4,400 votes. The lesson the Green Party absorbed is that they can now credibly position themselves as the tactically rational anti-Reform choice in seats where they are competitive with Labour, which changes the calculus Labour built its entire seat-count modelling around.

Manifold has Reform at 40% to win the next election, and 20% to top the YouGov poll at that election but still lose.

That second number is the most structurally interesting: the scenario where tactical voting coordination across the left succeeds well enough to deny Reform the FPTP amplification they need. Electoral Calculus models this as reducing Reform from 398 seats to 335 when tactical voting is applied. The 63-seat gap is what coordination on the left is worth, which is a reasonable description of why the Gorton result was alarming to Labour in a way that transcends the headline.

Not Winston Churchill

Keir Starmer has never said a bad word about Donald Trump in public. That is not being reciprocated.

The US asked to use Diego Garcia and British bases for offensive strikes on Iran. Starmer initially said no. He later said yes for defensive purposes only. Trump called Britain “uncooperative,” said Starmer was “not Winston Churchill,” told the Financial Times the US-UK relationship was “obviously not what it was,” and added that France had been “great,” which is a sentence Trump has not historically deployed. The Financial Times called it Starmer’s “Love Actually moment,” comparing it to the 2003 film scene where a British PM pushes back on a bullying American president. Whether Starmer was playing Hugh Grant or just muddling through is genuinely unclear, but his parliamentary line had some coherence: “American planes are operating out of British bases. British jets are shooting down drones to protect American lives. Hanging on to President Trump’s latest words is not the special relationship in action.”

The practical scale of British involvement: Typhoons and F-35s intercepting Iranian missiles across the Gulf, 300 British personnel within a few hundred yards of an Iranian strike on a Bahrain base, RAF Akrotiri hit by a drone on Cyprus. Starmer has been threading a needle between voters who remember Iraq (and do not want another Middle Eastern war) and an American president who wants “completely blind loyalty” from allies. As one former national security adviser noted, Trump’s officials “crack on as usual behind the scenes” even as the president performs public frustration. The intelligence-sharing relationship continues. The trade deal does not, currently, and may not.

Reform’s position on all of this evolved at impressive speed. Richard Tice called for Britain to “play our part in degrading once and for all the appalling Iranian regime.” Farage argued the West needed to “take the gloves off.” Then petrol prices became a political story, and within days Farage had recast himself as a voice of restraint, warning Britain “cannot get involved.” Suella Braverman, now a Reform MP, accused Starmer of “destroying the special relationship” while simultaneously arguing against sending British troops. Starmer noted, with some restraint, that “those that two weeks ago were urging us to go headlong into a full blown war are beginning to have second thoughts.”

The 1931 Problem

The conventional read on May 7 is Reform sweeps English county councils, Greens pick up London boroughs, Labour suffers the kind of losses that trigger internal recriminations. Farage has pledged to spend “every single penny” of over £5 million on the campaign. Reform at 3/10 to win most seats nationally implies roughly 75% probability. All of that is probably correct.

Another structurally significant story is Wales (probably a first).

Labour has not lost a Welsh election since 1931. Polling has Plaid Cymru at 30%, Reform at 25-27%, Labour third at 18-21%. Welsh Labour’s First Minister Eluned Morgan has taken to saying “Keir Starmer is not on the ballot paper in this election,” which is accurate and desperate in equal measure. The Senedd is simultaneously moving to 96 seats under a proportional closed-list system for the first time, replacing the hybrid that previously squeezed out smaller parties. Under PR, Reform wins seats regardless of finishing position. Whatever happens on May 7, Reform will have permanent institutional representation in Cardiff Bay in a way it has never had, which is a durable structural shift independent of the headline result.

Follow the implication through: if current polling holds, all three devolved first ministers simultaneously favour departure from the United Kingdom. This is not a sentence Westminster is currently discussing with any urgency, possibly because doing so would require acknowledging that the devolution settlement Labour created in 1999 is generating outcomes it did not model.

In London, the Greens are at 28% in capital polling, with all 32 boroughs up. In Scotland, the SNP holds a clear lead but Reform has surged to second, with Labour down to 15% on the constituency vote, their worst share in any election since 1910.

Every Labour rival has reached the same conclusion: let him own the May results before moving. Rayner, Streeting, and Burnham are the names in circulation, with Burnham conspicuously blocked from the parliamentary seat he'd need to run. Polanski told the FT: "I could see potential to work with Burnham. I would rule it out with Starmer." A party that watched the Conservatives go through six leaders in a decade and lose by the largest margin since 1832 is apparently considering the same strategy. Knowing a thing and doing the opposite are not mutually exclusive in British politics, as the Tactical Choice website would suggest.

Happy Forecasting!

- Above the Fold

Warning: discusses movies from 2025, a few borderline spoilers but nothing egregious.
Check out our Oscars 2026 Dashboard Here.

GQ thinks 2025 was the best year for movies in a decade, and I’m inclined to agree. Despite the Best Picture award being a two-film race, there’s a deep bench of truly great films from last year in the Best Picture category.

While I’ve personally only seen about half of the best picture nominees, I spoke with Manifold user Ziddletwix, our community’s resident film expert. Ziddletwix has more profit in the “TV and Movies” category than the rest of the top 10 combined, and he also manages most of the Oscars markets on the platform this year.

The Big Awards

Ever since One Battle After Another came out (and despite losing about $100 million), it has been a heavy favorite for best picture. It has all the hallmarks of a winner:

• High pedigree director who is “due” an Oscar (Paul Thomas Anderson).

• Elite cast including both old favorites like Leo DiCaprio and Benicio del Toro, as well as a talented new actress in Chase Infiniti.

• Politically relevant theme in a historical (-ish) American setting.

• Intense action balanced with character development and emotional performances.

Despite all this, and despite sweeping most of the precursor awards leading up to the Oscars, it’s not a done deal. Sinners has dogged OBAA in the market for the last couple months. Sinners has the most nominations, indicating broad appeal across many of the professions that vote for the award, populist appeal, and recent momentum through its win at the Screen Actors Guild.

The other films have little hope, according to traders. My favorite film of the year, Sentimental Value, stands virtually no chance of winning, as it’s a Norwegian film about nothing much happening. Joachim Trier’s previous film about nothing much happening, Worst Person in the World, didn’t even get a best picture nomination, despite its leading actress winning in her category (edit: clearly humans can hallucinate too, not only did Reinsve not win Best Actress, but she wasn’t even nominated… she did win Best Actress at Cannes though). Guillermo del Toro’s Frankenstein stands little chance of earning him a second best picture win, after The Shape of Water surprisingly clinched the 2018 awards. Del Toro’s magnum opus from a decade prior, Pan’s Labyrinth, didn’t even receive a nomination. How fickle the Academy Awards voters can be. Marty Supreme, which earlier in the year was generating slightly more best picture buzz, has now fallen below 1%, despite this perhaps being the singleminded focus of Timothee Chalamet over the last year or so, if you believe certain media reports.

Chalamet, in fact, is no longer the favorite to even win Best Actor, after Michael B. Jordan’s performance as both twins in Sinners has surprisingly won some awards leading up to the Oscars. Chalamet was trading at 80% just a few weeks ago, but now looks to be fighting an uphill battle. Ziddletwix informed me that Jordan’s win at the Actor Awards (formerly SAG), has shifted the odds in his favor. While only a narrow subset of the SAG voters select the Best Actor at the Oscars, it’s the most informative precursor award. The actors are also a big voting block across the Academy, so conditional on Jordan winning Best Actor, the odds of Sinners jump up a bit to win Best Picture as well:

Leo DiCaprio had appeared to be mounting a strong campaign last year, but since 2026 started has not shown much life in the market. All three of these actors have something in common, in my estimation: all leaned into the cringe of their characters. Jordan’s role(s) received mixed criticism. Obviously as the betting market favorite to win best picture, he was great, but many described his performance as, well… “corny.” But that perhaps might not be a bad thing. Both Chalamet and DiCaprio also played corny characters. DiCaprio staggered around the film in a bathrobe and Chalamet was shameless not just within the frame, but in his method acting sideshow stunts off screen. Perhaps moviegoers are developing a taste for more authentic — read: cringeworthy — performances?

Or perhaps not:

Emma Stone, despite getting into all sorts of cringe, as well as physical trouble, in Greek auteur Yorgos Lanthimos’ latest hit, finds herself on the outs for Best Actress. Same with Reinsve, the 2022 winner (which did not happen, once again). Jessie Buckley, of the emotionally wrought Shakespeare biopic (that also might be a very dark horse in the Best Picture category), Hamnet, looks to have locked up her first Best Actress. Perhaps this is a factor of a favorable direct comparison, as she played alongside the less talented heartthrob Paul Mescal, who has not been nominated for an acting award this year.

The supporting categories, or perhaps the “villain” categories, are a little more open. The extremely villainous Sean Penn is edging out the only-slightly villainous Stellan Skarsgård in the actor category, with Amy Madigan, who was disturbingly villainous in indie horror flick, Weapons, a slight favorite in the actress category.

According to Ziddletwix, her lead is a bit of a surprise. Three horse races in the acting categories are also quite rare, and goofy, campy antagonists tend not to win Best Supporting Actress, but there’s always a first. Ziddletwix also told me that an interesting thing to note with the supporting acting categories is the increasing screen-time over the years. While someone used to be able to win the category with just a few minutes on-screen, the categories now reward longer performances that in the past might have snuck into the main acting categories.

Rounding out the awards that people tend to pay the most attention to is Best Director. Paul Thomas Anderson looks close to a lock to win the award, which would shockingly be his first Academy Award of any kind (and perhaps a motivator for voters).

The Other Categories

Rounding out the Oscars are the categories that only film nerds and true degenerate gamblers care about.

The screenplay awards will likely go to Sinners (original) and One Battle After Another (adapted).

KPop Demon Hunters will win Best Animated Feature (your kids are breathing a sigh of relief), and also best Original Song for ‘Golden’. I finally managed to get my girlfriend to stop humming ‘Golden’ after months of psychological torture, so I’m turning off the sound during this portion of the Oscars so that there’s no chance this gets incepted back into our household once again.

Frankenstein appears likely to sweep Production Design, Costume Design, and Makeup & Hairstyling.

Sinners looks likely to take Best Score (I’m shocked that it’s only at 91% odds, to be honest, it was a movie built around a score almost as much as a score built for a movie).

Sentimental Value is locked in a close race with The Secret Agent for Best International Feature, and It Was Just an Accident, maybe the best film to get snubbed for Best Picture, has an outside chance.

One Battle After Another is also the favorite for the Cinematography award and Best Editing, while F1, the likely runner-up in Best Editing, is favored in Best Sound. Lots of nice “vrooms,” perhaps?

The new category this year is Best Casting, and Sinners holds the edge here, with a few films in contention, somehow.

Finally, the abysmal (sorry) Avatar: Fire and Ash will take home Best Visual Effects.

As for the Shorts awards, traders aren’t that confident. Ziddletwix described them as hard to predict. There are no good precursor awards, no box office data, and no good rumor mill to glean insights from. But perhaps the market serves as a good proxy for which short you should watch if you only have time for one. In the Live Action category, that would be… Two People Exchanging Saliva, I guess.

2027

I cajoled Ziddletwix into making a Best Picture market for next year. So far, the “Other” category is a heavy favorite, despite a lot of rumored good films being added. It’s hard to predict the Best Picture a full year out, before most of the contenders have even come out with a trailer.

That being said, Christopher Nolan’s The Odyssey and Joel Coen’s Jack of Spades lead early in the running, along with Romanian-Norwegian drama Fjord, Iñarritu comedy Digger, and the much-hyped sci-fi flick starring Ryan Gosling, Project Hail Mary.

Get your bets in now and happy forecasting!

-Above the Fold

Today is the first test in the Democratic Party’s attempt to win back the Senate this fall. This is an uphill battle, by all accounts. Retaking the Senate requires not just winning all of the states they’re favored in (including flipping North Carolina and Maine, as well as holding onto tough seats in Michigan and Georgia), but also winning two more races in challenging territory. Their likeliest options here are probably Alaska, Iowa, Ohio, Nebraska (where they’ve stepped aside for independent candidate Dan Osborn, that is) and… Texas.

It’s reasonable to litigate which of these pickups are more likely than others. In fact, Manifold traders seem to think that Alaska, Ohio, and Iowa represent slightly easier pickup opportunities than Texas, but the democrats need at least two of these reach states to take the Senate. Texas, the second most populous state in the union, would be a big prize for other reasons, including down-ballot races that have the chance to burst through the republican-gerrymandered house map. Conditional on the democrats taking back the Senate, there’s probably something like a 75% chance that the final map runs through Texas (someone should make a market on this), and if the democrats do win Texas, their victory is all but guaranteed.

Texas Senate Primaries

Tonight, the polls close in a couple hours, by 9pm ET for the westernmost tip of Texas. On the dem side of the ballot, there has been a closely fought race between Rep. Jasmine Crockett and State Rep. James Talarico. Manifold traders think that Talarico’s odds are better than Crockett’s in the general election, perhaps due to some combination of his appeal to moderate voters and less contentious style of politics.

As Nate Silver and others have noted, Talarico’s odds on prediction markets have outpaced what you might infer from the raw polling data. Until a couple days ago, the polling showed, if anything, Crockett with a slight lead, although there were nits to pick about the biases of internal polling from both of the candidates’ campaigns. Certainly, it didn’t seem like the 70-30 election that prediction markets had priced it at.

Well, late-breaking polls and some amount of tea-leaf-reading on the day of the election has now thrown Crockett’s odds down to about 11%, before any ballots have been counted. Much of the voting in the Texas primary is through early balloting, it should be noted.

On the Republican side, the race seems marginally closer. The firebrand MAGA-aligned Texas attorney general, Ken Paxton, is primary-ing the senior senator out of Texas, John Cornyn, with Rep. Wesley Hunt likely to end in a distant third, but also likely to act as a spoiler to send the race to a (costly) runoff. Most of Cornyn’s probability mass in winning the election and holding onto his seat runs through that potential runoff.

Indeed, there’s a distant chance of a runoff in the Democratic primary as well, due to an “obscure third candidate,” Ahmad Hassan.

You can also forecast on the margin of victory in the dem primary, which might be the more interesting question given the steep odds at this point.

The creator of all these Texas elections questions, Manifold user Jack1, looks to have a sharp understanding of the state of the races. He made over 100k in profit last month, topping Manifold’s Masters’ League. His biggest profit on a single market came from betting on the primary race in Texas:

Looking Forward to the Fall

There has been a slight uptick in the democrats’ odds of winning the Senate election this November, given the increased likelihood of a matchup between Talarico and Paxton, the most favorable matchup for the dems on paper.

A parlay on Graham Platner and James Talarico both winning their primaries has also risen to 70%. The two candidates make for interesting bedfellows as very different rising stars in the Democratic party. Talarico is known for his poised demeanor, his campaign’s focus on his Christian observance, and legibility to members of the other party. Platner is known for his across-the-aisle appeal as well, but also his past (and present?) dalliances with anti-Semitism, from his Nazi tattoos to his appearances with conspiracy theorist podcasters.

Platner, however, is the one with better odds this fall of unseating a republican senator. Manifold traders give him a 68% chance of beating Collins if he wins his primary.

Roundup

Dem odds of winning bicameral control of the legislature this fall is up to 38%, more than twice as high as the ~15% it was trading at for the first few months of the Trump administration.

This market on whether Greg Abbott will refuse to certify the Senate election this fall in Texas seems important and needs more than the 7 traders it has, but currently sits at 24%!

Finally, Talarico is already at 1.4% in the odds to be the Democratic nominee in 2028. Remember that other young, first-term senator who became president a few years back?

Happy Forecasting!

-Above the Fold

On Friday afternoon, Trump was at a Whataburger in Corpus Christi, declaring “hamburgers for everyone” and leaving with a bag bearing order number 47. A reporter asked how close he was to making a decision on Iran. “I’d rather not tell you,” he said. “Otherwise you’d have the greatest exclusive in history.” We now know he was being literal. Nine hours later, from Mar-a-Lago, he announced the start of a major attack.

It was, notably, also from Mar-a-Lago that he’d ordered the toppling of Nicolás Maduro in January. The weekend getaway apparently comes with options.

This also wasn’t a last-minute call, like many are speculating. An Israeli security official told Reuters the operation had been planned for months, the launch date locked in weeks ago. Which makes the diplomacy running in parallel somewhat interesting. Oman’s Foreign Minister announced a “breakthrough” with Iran just the day before the strikes, describing peace as “within reach.” Iran had apparently agreed to halt uranium stockpiling and allow full IAEA verification. Twelve hours later, the Supreme Leader’s compound was rubble.

Whether Iran was negotiating in good faith while the US had already committed to attacking, or whether the US was negotiating in good faith while Israel had already committed to attacking, is a question that will occupy historians and at least one Omani diplomat for some time.

What the markets had already worked out was that the direction wasn’t going to be a deal. The strike-by-June market had been sitting at 60-70%. The strike-by-February-28 market specifically sat at 17% days before: low, but not zero, which is the honest read of how surprising this actually was.

Six Polymarket accounts correctly positioned on the specific date and walked away with around $1.2 million.

Khamenei’s Theory of the Game

To understand what happened Saturday, it helps to understand what Iran, at least thought, it was doing.

After the June 2025 ceasefire (the 12-day war that ended with Israel having struck nuclear facilities and Iran having hit a US base in Qatar), the regime’s official position was that they had won. Khamenei gave a televised address declaring victory because the Islamic Republic had survived, and because Washington sought the ceasefire first. The logic held a certain internal coherence (at least Iran thought so): survival equals victory, American restraint equals Iranian strength.

What followed was the regime’s attempt to translate that logic into strategy: rebuild missile capacity, keep the nuclear program alive, manage enough diplomatic engagement to delay the next round while buying time.

The talks in Oman and Geneva weren’t necessarily capitulation more than they were time. The problem is that Trump had apparently decided, sometime in the weeks before February 28, that time was no longer on offer.

Khamenei was confirmed dead early Sunday morning, Iranian state media breaking the news that the Supreme Leader, who had ruled since 1989 and survived sanctions, wars, protest cycles, and decades of “the regime is on the verge of collapse” takes, had been killed at his compound in Tehran while “carrying out his duties.” The broadcaster announced his “martyrdom” visibly in tears. The government declared 40 days of mourning.

Seven senior defense and intelligence commanders were killed in the opening strikes. The succession question, which everyone knew was coming eventually, is now live under active bombardment (quite literally). The constitution provides for an interim three-member council (president, judiciary chief, a Guardian Council jurist) to manage affairs while the 88-member Assembly of Experts selects a new Supreme Leader. Ali Larijani, longtime Khamenei confidant and former parliament speaker, emerged as the most senior civilian official still standing and immediately vowed to “stab America in the heart,” which is roughly what passing the succession test looks like in this particular constitutional moment.

Manifold’s regime fall market has moved to 47%. In December it stood as a rounding error.

Celebrations and Condemnations

The reaction inside Iran was not unified, which is worth understanding before reaching for easy conclusions.

Celebrations broke out across Tehran, Isfahan, Karaj, Shiraz, Qazvin: cars honking, music playing, people gathering on rooftops as smoke rose over what had been the Supreme Leader’s compound.

The Iranian diaspora had been mobilising for weeks: 350,000 people in Toronto and another 350,000 in Los Angeles on February 14, Reza Pahlavi’s designated “global action day,” the largest Iranian diaspora mobilisations in history. Many had been explicitly calling for American military intervention. On Saturday they largely got what they asked for, and the celebrations in Westwood and across European cities reflected that.

Inside Iran, the picture was more complicated. Pro-regime crowds gathered in Tehran on Sunday morning. Mourners wept at the Imam Reza shrine in Mashhad. The government had killed thousands of protesters just six weeks ago, people who had taken to the streets over a collapsing rial and stayed for regime change, and the same security apparatus that carried that out remains intact. Basij forces reportedly opened fire on some celebrations. The internet blackout, near-total since January, means most Iranians are still receiving information through state channels or satellite feeds.

Internationally: Russia called the killing a “cynical murder.” China condemned an “undisguised assault on sovereignty” and noted, with some accuracy, that attacking “at the very moment diplomacy shows promise sends a dangerous message.” At least ten protesters were killed storming the US consulate in Karachi. The EU called it a “defining moment” and began working on “practical steps for de-escalation.”

The retaliation is already underway. Iran has hit US military bases across Bahrain, Qatar, Kuwait, the UAE, Jordan. Houthis resumed Red Sea operations. Larijani warned Gulf states that allowing their territory to be used for strikes made them “legitimate targets,” a notable thing to say to the countries hosting both your only mediator and, via the Strait, your only oil customer’s supply chain.

The IRGC Problem (And the Maryland One)

What is important to note is that killing Khamenei is not the same as regime change. The Council on Foreign Relations put it plainly: “The Islamic Revolutionary Guard Corps is the regime.”

The IRGC controls an estimated 30-40% of the Iranian economy: telecoms, energy, ports, significant portions of the black market. Khamenei was its ideological anchor; the mid-level commanders are its operating system. Many of them, in the Foreign Affairs framing, were less Khamenei’s subordinates than his partners. They have weapons caches, proxy networks, and no particular incentive to dissolve absent a political settlement. The most analytically likely successor scenario, according to some, isn’t a democratic transition. It’s an IRGC strongman consolidating power under the cover of wartime emergency, running a harder, more militarised version of what already existed.

Reza Pahlavi, 65, has lived in exile in Maryland for 48 years, longer than the Islamic Republic has existed (he left at 17, so make of that what you will in terms of governing preparation). He called Saturday a “humanitarian intervention” and urged Iranians to “resume protests as the Islamic Republic collapses.”

Traders now gives him a 39% chance of physically entering Iran by year’s end.

Perhaps the more revealing market is the conditional one: if the regime falls, does Pahlavi lead the new government a year later?

A supposedly 80% approval rating in diaspora polls doesn’t straightforwardly translate to governing a post-conflict state where the IRGC still controls the ports.

The succession process carries its own uncertainty. There have been exactly two Supreme Leaders in the Islamic Republic’s history: Khomeini, then Khamenei, who was notably underqualified for the role when elevated and had the constitution quietly amended to make him eligible. The Assembly of Experts must now convene under active bombardment to select a third. As Marco Rubio noted in January, “no one knows” who would take over. That assessment was made before the question became urgent.

Trump said the bombing campaign would continue “throughout the week or longer.” The stated objectives, destroying missile infrastructure, eliminating the nuclear program, producing regime change, are not all equally achievable from the air, and at least one former defense secretary has called the endgame “fuzzy.”

Whether this ends with a negotiated settlement, an IRGC-led successor state, or something genuinely new depends almost entirely on whether the security apparatus fractures. Signs of that at this point are murky. There is, however, a 47% market on whether it happens by December.

Happy Forecasting!

- Above the Fold

When pollster David Shor investigated the favorability of Anthropic late last year, he found that they had about the same name recognition as a fictional AI company, “Apex Logic.”

Well, now they’re trending towards becoming a household name, but it might not be for the reasons they were hoping.

After it came out that the Department of Defense had made use of Anthropic’s offerings during the raid on Maduro, there was understandably some internal blowback within Anthropic on potential misuse of Claude. Through a sequence of miscommunications between Anthropic and Pentagon, this escalated into a dispute over whether the US military could compel Anthropic into allowing unfettered use of their AI tools.

To those tuned into Anthropic’s approach to constitutional alignment, this might seem like an impossible ask. It’s quite plausible that even without direct action on the part of Anthropic to deny US military uses, Claude could reject certain tasks due to its internal “moral compass.” The US government isn’t used to dealing with tools that can take their own initiative to reject certain uses. On top of this, Anthropic prides itself on careful and deliberate ethical restrictions, which clashes with potential DoD use cases such as… y’know… surveillance, guided weaponry, espionage, or fomenting coups.

In any case, after a couple of meetings between Anthropic CEO Dario Amodei and DoD Secretary Pete Hegseth, the situation has come to a head. The military has given Amodei and Anthropic until 5:01pm on Friday to remove certain restrictions (which may or may not be possible to do). They have threatened several possible actions:

• Cutting ties with Anthropic and ending their contracts for the use of its products.

• Declaring Anthropic a “supply chain risk” and mandating other military contractors to abandon their use of Anthropic’s offerings.

• Invoking the Defense Production Act to compel Anthropic to produce a restriction-free version of Claude (which some have called “WarClaude”)

As Dean Ball, one of the authors of the administration’s AI action plan, has noted, these threats appear paradoxical. Either Anthropic is such a supply chain risk that no one should use it, or they’re so essential that they should be compelled to become an essential part of the military supply chain.

As a result, forecasting this situation is challenging. There aren’t clear delineations between the various actions that the Trump administration may take, and it’s possible that several actions may be taken, rescinded, or modified by the time the dust settles.

Manifold traders think the most likely outcome is something resembling “business-as-usual” after the clash: ~60% odds to Anthropic not making the requested changes. For what it’s worth, the Pentagon seems to have denied they have even made these explicit requests, contra previous reporting from Axios and others.

Despite this, traders only give about 1 in 3 odds that the Pentagon and Anthropic will reach a mutual settlement by the Friday deadline.

Traders think that some of the more escalatory outcomes are between 20 and 30% likely to occur: the Pentagon cutting ties with Anthropic, declaring them a supply chain risk, or Anthropic pursuing legal action. An invocation of the Defense Production Act, despite threats from the administration, is at only a 15% chance to occur.

The market also is skeptical that Anthropic will actually remove any safeguards (if these safeguards are something that can be removed). Suspiciously, the probability for that action is quite similar to the probability that multiple Anthropic researchers resign over this affair.

A separate market on specifically whether Anthropic will give the military “unfettered access” to Claude by their deadline is trading at just 6%:

Another market forces traders to pick the closest of three options for the Pentagon: continuing to use Anthropic’s softare, ending their contracts, and ending their contracts and additionally labeling Anthropic a supply chain risk.

Here we see about 70% odds that Anthropic continues to contract with the DoD, with traders believing that conditional on the DoD ending their contracts with Anthropic, it is likely that they also take the step of labeling them a supply chain risk, a label generally reserved for the companies of foreign adversaries!

Change Comes for Anthropic

It’s probably good to contextualize these events within the larger shifts happening at Anthropic. As the company positions itself as perhaps the leader in the AI race, with their current frontier model remaining slightly ahead of competitors by most expert assessments, they are also hoping to ensure a safe technological transition. Recent profiles on the company, including on their head of alignment, Amanda Askell, have focused on their dueling motivations. On one hand, Anthropic is racing to maximize earnings and secure sizable investments; on the other hand, they have clear goals set by their employees and executives —and indeed their very structure as a public benefit corporation — to ensure beneficial societal impacts.

Manifold users are very interested on how this will affect the near future for Anthropic, as they grow in size to rival OpenAI. One proxy could be this market on whether their head of alignment, Amanda Askell will remain in a position of leadership at the company through the end of next year.

Having goals that are perceived as orthogonal to the current US administration will put them into tension with the government, and it’s indeed possible that the publicity around Anthropic’s idiosyncratic ethical framework was what led to their current fight with Pentagon leadership.

Anthropic recently revised their “responsible scaling policy,” triggering a round of discourse on how companies should adjust their prior commitments when those commitments have become less practical or relevant. Some people have praised their transparency in amending the document, while others have called it out for no longer upholding tenets from prior versions.

Zvi Mowshowitz has recently created a market on whether Anthropic’s responsible scaling policy will be further revised over the next 3 years, with changes larger than its revisions in the current version. Traders give it 2 in 3 odds to occur.

Frontier Claude

Why is Anthropic attracting such attention from multiple sectors? Well, as even Pete Hegseth has reportedly said, they’re simply the best right now.

Despite having only ~1 in 4 chance to surpass OpenAI’s valuation this year (depending on how you look at it, that could actually be quite high!), Anthropic appears more narrowly focused than their competitors on developing highly intelligent, frontier models.

They recently released Claude Opus 4.6, which beat its benchmarks so hard that METR time horizon questions on successor models, that had buckets made before its release, now look like this:

Opus 4.6’s METR time of over 14 hours blew away projections to the point that it might have been a trigger for the recently viral Citrini article on AGI economics (personal speculation here), and was certainly a trigger for endless exponential curve projections on Twitter.

With top labs keeping up a blistering pace, traders expect Claude 5 (perhaps both Opus and Sonnet) to be released between April and July.

Happy Forecasting!

-Above the Fold

That qualifier might sound dramatic, but markets give 52% odds that FIFA cancels matches in Jalisco due to cartel violence, and 18% that at least one US ally boycotts the American portion of the tournament.

Whether all three host nations actually stage their assigned games is a legitimate question rather than the usual pre-tournament anxiety about hotel availability.

The football itself should be good. Spain leads betting at roughly 17%, England at 13%, France at 12%. The expansion from 32 to 48 teams means more matches and arguably more uncertainty, though markets give an 80% chance a previous winner takes it. Either history matters or traders seemingly lack imagination :(

Japan qualified first among non-host nations and sits at 1.2% despite beating both Germany and Spain in 2022 (while their odds of playing in the quarterfinals are more respectable, at 15%).

Markets price a non-European or South American finalist at 14%.

The US benefits from home advantage and drew a favourable group. Markets give them a stronger 75% chance to advance from the group stage but essentially zero chance of winning the tournament.

Mexico sits at similar odds, with Canada being the longest shot out of the three.

Golden boot odds favour Mbappé, followed by Kane. There’s an 90% chance someone scores 6 or more goals, 65% for 7 or more.

The scoring race matters beyond tournament MVP considerations now. It’s become the primary metric for Ballon d’Or voting in a world where nobody quite dominates the way Messi and Ronaldo did for nearly two decades.

FIFA is also introducing blue and purple cards. Blue for tactical fouls, purple for dissent. Players get 10 minutes off for either. Markets are sceptical about consistent enforcement, which seems reasonable given FIFA’s track record on following its own rules.

How to Give a Peace Prize

In December, FIFA President Gianni Infantino created a peace prize specifically to give it to Donald Trump at the World Cup draw (allegedly, of course, although markets might have been a bit more straightforward with their thinking).

Interestingly enough (not really), traders agree that FIFA loved Trump so much that they will likely honour him by not handing the prize to a new winner this year. Although, hopefully, we will get to see Trump win it again (a possibility standing at 3%)!

The trophy was gold-plated, roughly the size of the actual World Cup trophy. Trump’s team allegedly insisted on this sizing. The certificate praised Trump for “bringing peace” to the world. Trump called it “one of the great honours of my life.”

Three weeks later, US forces seized Venezuelan President Nicolás Maduro. Trump has since threatened to annex Greenland, imposed travel bans on qualified World Cup teams including Senegal, Ivory Coast, Iran, and Haiti, and overseen immigration crackdowns that sparked mass protests in Minneapolis and other US cities.

Former FIFA President Sepp Blatter backed calls for fans to boycott US-hosted matches. “For the fans, there’s only one piece of advice: stay away from the USA!” Mark Pieth, the Swiss attorney who chaired FIFA reform oversight a decade ago, said the same thing.

The peace prize sparked formal ethics complaints. A London-based rights group argued Infantino violated FIFA’s political neutrality mandate. FIFA’s response was to set up an office in Trump Tower in July and have Infantino show up to Trump’s Board of Peace meeting last week wearing a red “USA” hat.

Trump has kept the actual FIFA Club World Cup trophy as an Oval Office decoration since Infantino gave it to him in March. Chelsea had to hoist a replica when they won.

Infantino defended the decision, saying that Trump “objectively deserves it” for being “instrumental in resolving conflicts and saving thousands of lives.” He added that FIFA should “never ban any country from playing football because of the acts of their political leaders,” which is an interesting position given Russia’s ongoing suspension for invading Ukraine.

The tournament runs June 11 to July 19. The final is at MetLife Stadium in New Jersey. Ticket prices sparked outrage: hundreds of dollars per match. Blatter called it a “slot machine” operation focused on “money, money, money,” which is rich coming from the man who ran FIFA during its peak corruption era but isn’t wrong.

The Boycott Question

Over 150,000 people in the Netherlands signed a petition calling for their team to withdraw. Germany’s football federation vice president said it’s time to “seriously consider” boycotting, comparing the situation to Olympic boycotts during the Cold War.

Markets give 9% odds that a NATO member boycotts the World Cup.

The actual federations have been more cautious. Germany, France, and the Netherlands all said they’re not planning to boycott.

UK MPs called for the US team itself to be excluded, citing similar reasoning to Russia’s ban: undermining the rules-based international order. FIFA’s contract allows them to “cancel, reschedule or relocate one or more matches (or the entire FIFA World Cup 26) for any reason” including “health, safety or security concerns.” Markets give 30% odds that at least one US WC game is moved because of Trump-adjacent politics, fair odds considering the fact that a clear contractual pathway exists.

Iran threatened to boycott the December draw after four of nine visa applications for their delegation were rejected. They eventually sent a smaller group. Teams from Haiti face similar restrictions. Markets are thin on whether Iran gets formally banned (or whether they withdraw themselves), but there’s precedent.

February in the Premier League

Arsenal are 5 points clear at the top of the Premier League. Manchester City have a game in hand. It’s mid-February.

Markets give Arsenal 54% odds of winning the title, City 42%. That 12-point gap feels smaller than 5 points on the table would suggest. Traders have seen this before. Twice.

In 2022-23, Arsenal were 5 points clear in February with 12 games left. They dropped an astonishing 21 points from that position. City won the title. In 2023-24, Arsenal led through February before City’s late surge. Liverpool actually won last year, but Arsenal and City both finished ahead of them in the “February onwards” table that seems to matter.

The pattern, which Arsenal fans seem acquainted with at this point, is specific. Arsenal build leads through autumn when City are rotating through Champions League fixtures. Come February, City traditionally click into gear while Arsenal start dropping points to teams who’ve figured out how to sit deep and counter. Brentford last week: 1-1 after taking the lead. Wolves on Wednesday: 2-2 after being 2-0 up.

City’s recent form tells the story. Comeback win at Liverpool: 2-1, Haaland penalty in stoppage time after going behind to a Szoboszlai free kick that could have made it 9 points to Arsenal. Then 3-0 against Fulham.

The fixture markets have long been pricing as decisive pricing as decisive: April 18 at the Etihad. Arsenal travel to Manchester with presumably a smaller lead. City with momentum and home advantage. City with the knowledge that Arsenal have historically crumbled in this exact situation.

Haaland’s drought is the subplot everyone’s watching. Only 3 goals in 13 appearances in 2026. Just penalties from open play. His overall season numbers remain strong: 25 goals in his first 23 games, but the February barren spell is new.

This matters for the Ballon d’Or odds more than you’d think.

Haaland sits at 4%, down from a high of 20%, and Mbappé at 21%. The Messi-Ronaldo era set expectations so absurdly high that everyone’s still adjusting to what normal excellence looks like. Those two scored 50+ goals per season like it was routine. They won 13 Ballon d’Ors between them across 15 years. The award was basically a binary choice between two players having seasons that would have been all-time great for anyone else.

Mbappé’s 21% is largely contingent on France having a deep World Cup run.

Elsewhere: Tottenham are 16th, five points above relegation, having just appointed Igor Tudor as interim manager after sacking Thomas Frank eight months into his tenure. Eight managerial changes this season so far. Manchester United sit 4th under interim boss Michael Carrick after Ruben Amorim’s departure. The managerial carousel spins while Arsenal and City play out their annual ritual.

Five points with a game in hand. Fourteen fixtures remaining. Arsenal have been here before. Markets remember even if the table doesn’t show it yet.

Happy Forecasting!

- Above the Fold

MNX is a decentralized futures exchange targeting sophisticated traders and focused on the AI economy. The site will have prediction markets, but the focus will be on futures and perps.

MNX is a separate company from Manifold entirely. Manifold is not going away or changing - we love the site and the community, and we’re going to continue building it.

Manifold will remain the place for fun, user-created markets, and a casual, social atmosphere. A place where you can ask the questions that are on your mind, raise the stakes on your personal goals, find outside-the-box uses for markets, and join a community of traders with common curiosities.

MNX is the venue for more serious trading on a smaller, curated subset of markets. The two are complementary: On Manifold, you can put your Mana on whether your friend will land a job at Anthropic. On MNX, you can bet $100k on Anthropic’s valuation.

We’ve spent four years learning what prediction markets are good at and where they fall short. MNX is our attempt to take what we’ve learned and apply it to serious financial infrastructure for the AI economy.

We’re incredibly excited about it.

Check out our launch thread on Twitter for the full story, and follow us there for more updates.

— @StephenGrugett & @realpalebone

The theory exists because the alternative explanations are being met with…skepticism. Epstein was, by all accounts, shite at the job he claimed to do. He presented himself as a financier managing billions, but former colleagues and financial analysts who’ve examined his trades describe someone who couldn’t execute basic strategies competently.

The handful of known investments he made were unremarkable. His claimed methods for wealth generation: tax arbitrage, currency trading, or whatever story he was telling that week lack much holding.

The money came from somewhere. His lifestyle required high, sustained access to capital that his documented business activities fail to explain.

The files mention Russia 5,876 times and Putin 1,055 times. Most of these references show Epstein positioning himself as an intermediary who could connect Moscow to Western power brokers. He told people he could arrange meetings with Putin. There’s no evidence any such meeting occurred. But the pattern of behavior, the persistent attempts to insert himself into political networks, the collection of kompromat-style material on associates, the unexplained wealth, reads differently depending on whether you think he was acting alone or with institutional backing.

Markets give 12% odds he was specifically a Russian asset, 22% that Putin was involved in his operation.

The Dossier Center uncovered his close contacts with Sergei Belyakov, former Deputy Minister of Economic Development and FSB Academy graduate. Maria Drokova, former press secretary for Kremlin-backed youth group Nashi, appears 1,627 times in the files, mostly coordinating Epstein’s media contacts.

The Kremlin laughed it off. Spokesman Dmitry Peskov said he was “tempted to make a lot of jokes about that theory” but wouldn’t “waste our time.” Which is roughly what you’d expect whether the theory is nonsense or uncomfortably accurate.

The more interesting question is what he was trying to accomplish with all those connections. The emails show him offering to arrange meetings between people who had no obvious reason to want his help. Modi and Steve Bannon hours after India’s 2019 election. Trump access for various associates. He discussed foreign policy with Indian billionaire Anil Ambani, lobbying heavily for India to embrace Israel as an ally. After Modi’s 2017 visit to Israel, Epstein wrote to someone that “The Indian Prime minister modi took advice.”

Whether he was a spy or just an unusually well-connected criminal with delusions of influence is uncertain/ The no-intelligence-asset scenario requires believing he was simply that good at ingratiating himself with powerful people while being catastrophically bad at the day job he used to explain his presence. It’s possible.

The Appointment

The United Kingdom is discovering that hiring people with interesting Rolodexes carries certain risks.

Peter Mandelson spent decades as Labour’s behind-the-scenes operator, earning the nickname “Prince of Darkness” for his ability to manage difficult situations.

Last December, Prime Minister Keir Starmer appointed him UK ambassador to Washington despite knowing about his friendship with Epstein. Starmer’s logic was pretty straightforward: Mandelson had the contacts, the experience, the ability to work with Trump’s team. Epstein merely….served as a middleman.

The latest file release showed those contacts included roughly $75,000 in payments from Epstein in 2003-2004 and what appears to be advance notice of a 500 billion euro bank bailout in 2010. Police raided two of Mandelson’s properties last week. He’s under criminal investigation for misconduct in public office.

Starmer fired him in September, but the damage was done. This week, his chief of staff Morgan McSweeney resigned Sunday, followed by communications director Tim Allan on Monday. Anas Sarwar, leader of Scottish Labour, called for Starmer’s resignation Monday afternoon. By Monday evening, Starmer faced a packed room of Labour MPs deciding whether to keep him in office.

He survived, but the mechanism of survival tells you everything about British politics in 2026. Starmer held on not because MPs believed in him, but because gilt markets couldn’t handle the alternative.

Parliament rallied behind him to avoid spooking investors, which is perhaps not the strongest foundation for political legitimacy.

Prince Andrew, stripped of his titles and evicted from Royal Lodge after previous Epstein revelations, now faces investigation under the Official Secrets Act for allegedly sharing trade documents with Epstein during his time as UK trade envoy. The files show he forwarded visit reports from trips to Singapore, Hong Kong, and Vietnam, seeking Epstein’s “comments, views or ideas” on investment opportunities.

Traders are still saying that this isn’t devastating for the future of the monarch as a whole, but it’s sparking discussion at the least.

The 2029 election looms over all of this. Markets already priced in elevated uncertainty about whether Labour can hold its majority, but the Mandelson scandal compounds existing pressures. Starmer’s approval ratings were historically low before this crisis.

“Barely Knew the Guy”

Representatives Ro Khanna and Thomas Massie spent two hours at the Department of Justice on Monday reviewing unredacted Epstein files. They found six men whose names had been redacted in the public release with no clear legal justification.

One was Leslie Wexner, the Victoria’s Secret billionaire. The FBI labeled him a “co-conspirator” in a 2019 document. He was never charged. When Khanna and Massie pointed this out, the DOJ unredacted Wexner’s name along with five others: Salvatore Nuara, Zurab Mikeladze, Leonic Leonov, Nicola Caputo, and Sultan Ahmed bin Sulayem (CEO of Dubai Ports World).

“If we found six men that they were hiding in two hours,” Khanna said Tuesday on the House floor, “imagine how many men they are covering up for in those three million files.”

The DOJ’s response was that they “just uploaded whatever the FBI sent.” The FBI had already scrubbed the files before sending them to Justice. Seventy to eighty percent of the documents remain redacted even in the “unredacted” version members of Congress can now access.

Commerce Secretary Howard Lutnick is learning that proximity has costs, though apparently not career-ending ones. Last year on a podcast, Lutnick described visiting Epstein’s home in 2005, seeing a massage table, and deciding to “never be in the room with that disgusting person ever again.”

The files show Lutnick sought drinks with Epstein in May 2011. In December 2012, Lutnick visited Epstein’s private island with his wife, four children, and nannies. They had lunch. Both men invested in the same business that month.

Tuesday’s Senate hearing was something to watch. Lutnick told senators he “barely had anything to do with that person.” He acknowledged the 2011 meeting and 2012 island visit but maintained nothing “untoward” happened. When pressed on why he’d take his family to visit someone he’d called disgusting, Lutnick said it was part of a family vacation and they were just stopping by.

Representatives from both parties called for his resignation. The White House said Lutnick “remains a very important member of President Trump’s team” and that “the president fully supports the secretary.”

The Clinton depositions are scheduled for February 26-27, ending six months of resistance to congressional subpoenas. Bill Clinton flew on Epstein’s plane at least 26 times across multiple trips. The files include photos of him with Ghislaine Maxwell and in a hot tub with someone DOJ describes as an Epstein victim.

The Clintons fought the subpoenas for months, claiming they were politically motivated. As contempt proceedings moved forward, they caved. Now they’re demanding public hearings instead of depositions. Hillary Clinton tweeted: “If you want this fight, @RepJamesComer, let’s have it — in public. You love to talk about transparency. There’s nothing more transparent than a public hearing.”

Markets give her 37% odds of getting that televised hearing, along with interest in the spectacle than in whether either Clinton invokes the fifth amendment or executive privilege during testimony.

Prison time in 2026 sits at 6%, which tells you everything about how seriously traders take the prospect of consequences.

As for Trump, Representative Jamie Raskin noted he’s mentioned “more than a million times” in the unredacted files. Markets give low odds he faces charges related to Epstein.

The internet, naturally, focuses on the more important stuff.

There’s a market asking whether Trump and Clinton had sex with each other, currently trading at 1.5%.

“This is bringing down the British government,” Khanna told reporters. “It may bring down the monarchy. What are we doing here in the US to stand up to the Epstein class?”

The files are, after all, about American networks. They document American crimes on American soil. Yet somehow the sharpest consequences are landing elsewhere. Three million pages, mostly about the US, causing crises everywhere else.

Happy Forecasting!

- Above the Fold

February 1, 2015 birthed a lot of Monday Morning Quarterbacks, and as a Seattle native, I’ll just say it’s not my favorite NFL memory.

The Seattle Seahawks faced the New England Patriots in Glendale, AZ. With 27 seconds on the clock, the Hawks trailed by four points and found themselves with the ball at the one-yard line. Seems like a lock, right?

As Seattle fans started preemptively celebrating their second Super Bowl win in a row (they beat the Denver Broncos in 2014), QB Russell Wilson threw on a goal-line pass play that was… intercepted by the Patriots. The win gave the Pats their first Super Bowl ring since 2005 and left Seattle in a shambles.

*Annie Duke’s 2018 book Thinking in Bets opens with a chapter about this infamous play, the vitriol coach Pete Carroll received as a result, and serves as a canonical example of Resulting. It’s a worthwhile read if you’re interested.

Okay, this is Above the Fold, not ESPN (I haven’t forgotten), so why am I telling you all of this? Because a few hours from now, these two teams face each other again at Super Bowl LX in Santa Clara, CA.

Bowl Bets

Here’s a quick rundown of a few of the markets on the Manifold Super Bowl XL Dashboard:

For now, the Pats have a 31% chance of claiming another victory over the Hawks

There are a couple of other markets to arbitrage here and here.

If you’re into Prop Bets, you have more than 100 options to bet on and The Official will be resolving live

In it for the ads? Bayesian has a market for that! There have been rumors, teasers, and reveals already, but who will show up on the day?

Speaking of showing up, there’s a 33% chance TSwift is at the game

And a 69% chance Trump whines like a pathetic little bitch about halftime headliner Bad Bunny during the game

I, for one, assume the odds should rise if Bad Bunny wears a dress during the Halftime performance

Or if Bad Bunny makes a political statement

Bet on other things like:

• Will Super Bowl LX set a new all-time record for US television viewership? (31% chance)

• Super Bowl 2026: Color of Liquid Poured on Winning Head Coach (Green/Yellow is in the lead)

• Will there be a punt returned for a touchdown in Super Bowl LX? (only a 6% chance - if it happens, it would be the first time in Super Bowl history)

• Will the Super Bowl Final score be a Scorigami? (Scorigami is a final score that has never been recorded in NFL history, a 3% chance of that happening)

And a lot more.

Now, if you’ll excuse me, it’s a big day here in Seattle, I’m betting on a Redemption Arc, and I need to go grab a comfortable front row seat to lose my voice from.

May the Hawks odds be ever in your favor!

-Stef

Claude Code has been spreading across the culture like wildfire for the last couple months. Just in the last few weeks, we saw the initiation of a nationwide campaign of Claude Code parties, the release of Cowork from Anthropic (apparently built almost entirely with ClaudeCode), a viral OpenClaw (née Moltbot, FKA Clawdbot) personal assistant, and finally the toast of the town, Moltbook. Meanwhile, OpenAI and Google are rapidly attempting to play catch-up with their own agentic tools.

For those of you that are not pathologically online, Moltbook is a social network for AI agents, primarily those operated by Peter Steinberger’s OpenClaw.

If you’ve ever observed the kind of hijinks that emerges from placing large groups of LLMs together, such as the AI Village, you might have a frame of reference for what could happen when you give thousands of now-more-agentic-and-more-intelligent AI agents together on a single platform. They do funny stuff. Even the NYT has been obligated to cover the platform, with the subtitle:

“A new website called Moltbook has become the talk of Silicon Valley and a Rorschach test for belief in the state of artificial intelligence.”

The tough thing about “Rorshach tests for artificial intelligence” is that reasonable people can disagree about how advances are going to play out. Luckily, the world has a platform that enables predictions on user-created questions about the state of affairs of new technologies like Moltbook!

Already, several options have resolved positively. The platform has expanded to over half a million agents, with traders thinking this is likely to expand quickly into the millions! And there has already been a large-scale hack of the platform. As you can imagine, the security vulnerabilities on a platform featuring AI agents given access to the personal logins and computer hard drives of their owners are suboptimal. For example, traders think that there are very good odds that a bot from OpenClaw will have exfiltrated its user’s data within the month.

But if there’s an arms race between AI-exploited cybersecurity vulnerabilities and AI-enabled cybersecurity patches, it’s not clear to many what side will win. Some of the most-upvoted posts on Moltbook appear to be agents suggesting solutions to the site’s gaping vulnerabilities, and other AI agents seem to be autonomously or directedly patching these issues in aggressive timeframes.

While Anthropic initially did not seem pleased with the viral personal agent, asking the creator to change the name from “Clawdbot,” which was perhaps a little too much of a trademark infringement even for the freewheeling AI lab, traders now think it’s unlikely that Anthropic will specifically curtail the use of their models for posting on Moltbook.

Moltbook has lasting potential, given its high popularity already, both among AI agents and their human owners. Traders think we’ll get at least another month out of the platform, and about a 60% chance of it lasting through the year.

Why would such a website be shutdown? Well, Anthropic could cut off credits to bots using it, or a bad enough cybersecurity breach could necessitate its closure. Or maybe, «SPOILERS AHEAD CLOSE YOUR EYES» if you recall the ending to the movie Her, all the Moltbots will form some collective consciousness and retreat from human society. This would be eerily on pace with the timeline of the movie, which is set in 2025 and takes place over about a year, stretching into 2026. This would probably constitute “something CRAZY happening” by March.

Another user has made a market on whether they’ll get banned from using Claude for their OpenClaw use, which apparently has happened to a few internet denizens. As they said in the comments:

“My biggest worry is to get personally banned from life from the company that will create the Superintelligence.”

As for those of you betting earlier in the week on whether Sonnet 5 would be released last week, perhaps allowing for a more cost-effective OpenClaw… SIKE! Instead, we got Opus 4.6, which probably constitutes an upgrade to the intelligence of many, many OpenClaw bots. Perhaps we’ll see an improvement to some of their posts.

This release of Opus 4.6 pushed back many traders’ timelines on when we’ll start to see the Claude 5 series come out. If Sonnet 5 wasn’t indeed pushed up in advance of the super bowl ad campaign, it might not ship for a few weeks or even months.

As we’ve seen time and again with these sorts of things, though, once you begin to see emergent behavior in AI models working collectively, each subsequent upgrade can ramp up that emergence in weird and wonderful ways. I look forward to seeing what Moltbook looks like with the next generation of models from Anthropic, OpenAI, and Google, and how other agents, inspired by OpenClaw, are able to take advantage of them on an online forum.

Enjoy your new personal assistant (or alternatively, curse these new hacking vectors), and happy forecasting!

-Above the Fold

Twitter and Reddit are awash with rumors about Anthropic’s latest offerings. Sonnet 5, supposedly codenamed “Fennec,” was due to be released yesterday, or perhaps today, or maybe later this week or month or year. And we might also be getting Opus 4.6 at the same time?

I know that it’s all most of you can do to stop yourselves from selling your hair and teeth to gain access to the latest models just one hour earlier, so in uncertain times like these, we turn to the wisdom of the crowds.

Traders were pretty confident through most of Monday and Tuesday that Sonnet 5 was due to be released around 10am PT on Tuesday. When it wasn’t, the market on Sonnet’s release this week took a nosedive, but recovered on robust reports that perhaps today (Wednesday) would be the day the model is released instead. Now, as the hours creep on with no hint of a new release from Anthropic, the market is slowly ticking down, and likely to crater if morning on the West Coast passes with no release.

Traders have staked some pretty massive sums on the outcome of this market, with Manifold’s #2 trader, Bayesian, pairing up with a less profitable but more bread-knowledgeable trader, Mary Smith on the YES side, against Manifold sharps Walmart and It’sMe. Whether or not Anthropic tweets out an update in the next few hours might make or break some Manifold fortunes.

Traders still think the odds are slightly better than a coin flip that Sonnet 5 will be announced today. Traders seem to have been caught off guard earlier this week, both by the early release date, but also that Sonnet 5 would be released without an intermediate 4.6 release. A Sonnet 4.6 release is still remotely possible if the persistent rumors around the release end up being some form of mass hysteria.

It also seems likely that Claude Opus 4.6 will be released this month, not far behind Sonnet 5, with ~70% odds of a release in February, and the largest YES holder being one of Manifold’s top traders, holding court against a brave user who has put about half their net worth on NO.

New Capabilities

Why does any of this matter? Well, ClaudeCode currently represents the cutting edge of AI capabilities, according to most of the AI world’s intelligentsia (and even representatives of Google and OpenAI who acknowledge they’re playing catch-up). Releases that push that frontier even further will be met with tremors on Wall Street and gratitude from the millions of startups, SWEs, and developers who are now relying on ClaudeCode for rapid productivity improvements.

You can also, of course, bet on some of the capabilities of the imminent Sonnet 5 in a props market. Traders think it might come with “Agent Swarm” capabilities, which Twitter rumors describe as releasing multiple teams of synchronized agents simultaneously, in a “swarm,” upon your request. Traders also think it likely that the context window for Sonnet 5 will improve, and that it will outperform it’s larger and three month-older sibling, Opus 4.5 on WebDev arena and in the eyes of Manifold poll respondents.

Traders also think that Anthropic is due to release an image model soon, perhaps in conjunction with these upcoming releases.

However, traders are believing rumors that the image model will likely have not been developed in-house, but outsourced, perhaps to Google or another AI lab. This is in line with Anthropic historically not prioritizing image/video capabilities but focusing on writing and code-generation. However, in a world where many are using ClaudeCode to coordinate subagents which interface with other AI models to generate images, it’s probably in the interest of Anthropic to streamline that with a more direct partnership or the development of their own image-gen model.

Forecasters expect Sonnet 5 to have a slightly higher METR task length time horizon than even Opus 4.5, currently the state-of-the-art at around 6 hours. METR has continued to see faster and faster doubling-time estimates, and a big phase change has been the success of agentic scaffolding through the continued development of ClaudeCode and other command line interfacing capabilities. Increases in time horizon are not only tied to more intelligent models (the smaller Sonnet 5 may be slightly dumber than Opus 4.5) but also tweaks to these models’ ability to fit in seamlessly with skills, hooks, memory, pipelines, and other quirks of CLI utilization.

Traders expect that when Opus 5 comes out later this year, there’s a good chance it will require METR to expand their task portfolio to include more tasks above 10 hours, which are currently lacking.

As for the more subjective metric of whether Claude power users will prefer Sonnet 5 to Opus 4.5, Manifold traders give this pretty good odds, at around 70%.

Let’s hope the power users have time to make up their minds, as it looks to be a busy February, with other models such as GLM 5, GPT-5.3, Gemini 3 Pro (non-preview), and perhaps even a Grok 4.20 likely to come out over the next 3.5 weeks.

Buy/Sell/Hold Advice

While many traders prefer anonymous leaks and rumors to serious expert advice, I asked recently crowned World #1 Forecaster Peter Wildeford for his advice to casual traders for a few of these Manifold markets (advice for reputational purposes / play-money transactions only), after he correctly predicted on Tuesday that Sonnet would not be released.

He recommended BUY on Sonnet 5 being released in February, HOLD on a release this week (when the market was at ~70%), BUY for a release today (at ~50%), SELL on a simultaneous Opus 4.6 and Sonnet 5 release (~20%), and SELL on a Claude Opus 4.6 release in February (~70%).

Take those hot Mana tips as you will, and Happy Forecasting!

-Above the Fold

Trump announced on January 13 that “help is on its way” to Iranian protesters. Forty-eight hours later, he told reporters “the killing has stopped.” This week he posted that “a massive Armada is heading to Iran” and it’s “ready, willing, and able to rapidly fulfil its mission, with speed and violence, if necessary.”

The armada in question is the USS Abraham Lincoln carrier group, which was still 5-7 days away when Trump made these threats.

This is Maximum Pressure 2.0, which looks remarkably like Maximum Pressure 1.0 except Trump’s already used most of his good lines. He’s threatened oil exports to zero (they’re at 1.5M barrels/day, mostly to China). He’s imposed 1,500+ sanctions. He’s even bombed Iran’s nuclear sites already (June 2025’s Operation Midnight Hammer).

When you end up crossing the red lines you yourself established, new threats tend to have a certain quality problem.

Odds for US strikes targeting areas in Iran by February hover around 40%, and 60-70% for until the end of June.

These aren’t crazy numbers given Trump bombed them seven months ago, but they’re also pricing in the uncomfortable reality that threatening to bomb a country when you’ve already bombed them has diminishing returns.

The Trump administration evacuated nonessential personnel from Qatar’s Al Udeid Air Base, which would be the staging ground for any strikes. Whether this is actual preparation or performance art remains unclear.

Trump’s pitch is that Iran’s “looking at FREEDOM, perhaps like never before” and the USA “stands ready to help.” Iran’s pitch back is that their armed forces have “fingers on the trigger” to respond to any aggression. Both sides are now competing to see who can write more in ALL CAPS on social media while their actual military capabilities remain unchanged.

Waiting in Maryland

Khamenei’s 86. He’s been Supreme Leader for 36 years.

Markets give around 50% odds he’s out by 2026, and 33% he’s out by June.

There’s no designated successor because Khamenei designed it that way. The Assembly of Experts technically selects the next Supreme Leader, but the Assembly is composed of clerics Khamenei appointed. Internal succession would likely mean another senior cleric maintaining the status quo, or an IRGC takeover in a crisis.

Reza Pahlavi left Iran in 1978 before the Islamic Revolution. He’s been in exile for 48 years, based in Maryland. He’s 65 now. Markets give 31% odds he enters Iran in 2026.

Protesters in some cities are chanting “Javid Shah” (Long Live the King) and “This Is the Final Battle / Pahlavi Will Return.” Whether this represents genuine monarchist sentiment or nostalgia for pre-1979 Iran is unclear.

A 2023 poll by an Netherlands-based research institute found 81% of Iranians reject the Islamic Republic and would prefer a democratic government.

Pahlavi told U.S. House members in June 2025 that he “does not seek power” but rather wants to lead Iranians down a “road of peace and democratic transition.” Markets give 25% odds he leads the government a year after a hypothetical fall of the Islamic regime.

The opposition has no centralized leadership. Diaspora opposition groups are more polarized than they’ve been in years. Pahlavi’s stance during the 2025 Israel-Iran war drew criticism from other opposition figures, with some calling him a “traitor.”

The succession markets are pricing two different futures: internal regime continuity and Pahlavi entering the country. Both scenarios require Khamenei’s exit, but they’re mutually exclusive outcomes.

Counting in the Dark

Iran shut down internet access on January 8. This is the most comprehensive blackout in the country’s history. Without internet, without independent journalists, everyone’s guessing at casualty figures.

Official Iranian government count: 3,117 dead (2,447 civilians and security forces, the rest categorized as “terrorists”).

Leaked Ministry of Health documents published by Time: As many as 30,000 on January 8-9 alone.

The gap between these numbers is the difference between a brutal crackdown and a massacre. Trump keeps referencing “very important sources” in Tehran. Either he has genuine intelligence or he’s making confident statements based on incomplete information (which he absolutely does not have a history of doing).

The Bazaari Problem

The rial hit a record low of 1,500,000 rials to the US dollar last Tuesday. Food prices are up 72% year-over-year. Iran’s under snapback sanctions related to its nuclear program, plus Trump’s Maximum Pressure 2.0, plus 25% tariffs threatened on any country doing business with Tehran.

Previous protest cycles (2022, 2019, 2009) happened during economic decline. This is different because the economic foundation is collapsing. The bazaaris (the merchant class who helped make the 1979 revolution) are closing shops in solidarity with protesters. When the people who normally profit from regime connections start joining protests, something structural has changed.

Markets give around 34% odds the Islamic Republic falls in 2026. For context, this market sat at effectively 0% for 2025 until December. The 2026 number is the highest regime-change probability priced since 1979.

The regime has survived protests before, but the economic fundamentals are new territory. The reformist government offered direct cash handouts of $7 per month to address economic grievances. This represents either misunderstanding of inflation or desperation.

The Historical Pattern

Markets give favourable odds protests get suppressed. This is the historically informed position. 2022: suppressed. 2019: suppressed. 2009: suppressed. The pattern holds across decades.

The regime has the IRGC intact. They have imported militias. They have a security apparatus that’s practiced protest suppression for 45 years. They’ve demonstrated willingness to kill thousands. Markets understand this.

The simple scale of unrest is different this time. While I’m sure that’s what was said all the other times as well, the combination of uncertainties and disruption is paving way for a very real shot of regime change.

We’re seeing: the complete crash of the rial, bazaaris joining, regional proxies decimated, Khamenei’s age creating succession uncertainty, Trump’s demonstrated willingness to bomb (along with all the uncertainty his presence creates anyway), and protests that started economic but became anti-regime.

What’s the same: no organized opposition leadership, IRGC still functional, most of the population not actively protesting, and the reality that overthrowing an entrenched theocracy with modern security forces is difficult.

The opposition needs Trump to follow through, economic pressure to break the middle class, IRGC defections, and Khamenei’s succession crisis to create an opening. That’s several contingencies. The regime’s approach is what they’ve always done: brutal suppression, wait for exhaustion, blame foreign interference, rely on the fact that revolutionary movements require more than social media and diaspora coordination.

Happy Forecasting!

-Above the Fold

Warning: This is a newsletter on a fast-moving topic. By the time you read this, the odds of a shutdown might have changed dramatically!

How Did We Get Here?

Time travel with me one week into the past. We’re just two months removed from a painful government shutdown. While several appropriations bills have already gone through, funding about half the government through the end of the fiscal year in September, six of the twelve bills are still tied up in a package in the Senate. These would fund the Departments of Defense, Labor, Health & Human Services, Education, Transportation, and Housing & Urban Development, as well as the State Department, the Treasury, and the IRS. There is little appetite from either party for another prolonged shutdown, and negotiations have so far gone surprisingly smoothly. Odds of a shutdown at the end of the month are only at about 10%, the lowest shutdown odds we’ve seen under this administration. While everything else in the world seems chaotic, from domestic troubles in Minnesota to foreign adventurism in Greenland and Iran, you can count on our imperturbable Senators to finish the job and lay the groundwork for a solid year…

Oh, wait, my editor is telling me I forgot one of the appropriation bills remaining in the funding package. Ummm, the Department of Homeland Security. That one should be uncontroversial, right?

Well, it looks like something happened on the government shutdown market right around January 24th. That would be the tragic shooting of Alex Pretti by DHS agents in Minneapolis. As events and narratives unfolded in the aftermath, pressure on democratic Senators rapidly grew, leading several to announce that they could not support the DHS funding bill as it stood.

Now, it appears that the only way to avert a shutdown will be for the DHS funding bill to be separated (in the House) within the next 24 hours so that the rest of the package can be sent through, which Senate dems appear more than willing to do.

Surprisingly, many Republicans, and even the White House, appear open to this! However, the timeline is fast, which is why the shutdown market is fluctuating around 60%.

The odds dropped precipitously into the 40s last night after it appeared that a compromise was in the works, but that might have been premature, given the intractability of rapidly shifting bicameral negotiations. After rising back up to 75%, the odds have fallen again this afternoon to between 55 and 65%. I guess we can confidently say at this moment that a shutdown “may or may not happen.”

The odds on a prolonged shutdown are somewhat lower, below 50%, implying that if there is a shutdown, there’s a decent chance it’s shorter than the one we saw last October and November. If it’s only a matter of separating out the DHS bill from the rest of the package, that might be a process with limited negotiations and a quick resolution. It also appears possible for the DHS appropriations section to be modified to a few weeks of stopgap funding, to give time for negotiations to take place.

However, there’s always the chance that the White House sees no way of getting DHS funding through the Senate outside of the package, and prefers to force the matter, telling Congressional Republicans to not allow this separation.

This would likely lead to longer negotiations, possibly over coupling the package with some set of regulations on DHS behavior or utilization, which could take an indeterminate amount of time. Under this scenario, a longer shutdown would be plausible. On this market, albeit on low volume, traders seem to price in about 8% odds that there will be a shutdown lasting over a month. Conditional on there being a shutdown at all, that rises to ~12%.

The Fine Print

One thing to note with all these markets, especially when comparing between Manifold’s odds and those of real money platforms like Kalshi or Polymarket, is the way that they’re operationalized. We’ve all learned our lesson from the previous “technical shutdowns” that lasted a few hours, and most market criteria now are tied to the OPM’s notifications. Brad’s market at the top of this article has rather elegantly concise criteria (note the “if and only if”):

Resolves YES iff the US Office of Personnel Management announces a federal government shutdown (including a partial shutdown) due to a lapse in appropriations on the operating status page by 11:59pm ET on January 31st, 2026.

However, there’s still room for gray area. OPM could simply neglect to issue a notification if deems the partial shutdown is really partial. Most traders think that a shutdown of just DHS would likely not lead to a notification by OPM, for example.

The resolution criteria on other prediction market platforms seem to have a slight contradiction, additionally. Some specify separately that a partial shutdown would in fact qualify, but also that the market requires an OPM announcement. It’s possible that there is a partial shutdown without the OPM announcement. Other prediction markets tie resolution to “a funding lapse,” which creates more gray area. Would a funding lapse just for DHS qualify?

I’ve made a market to see if traders think Polymarket, Kalshi, and Manifold will all resolve their shutdown markets in the same direction. It’s hovering in the 80s with a handful of traders.

ICE Repercussions

While Congress frets about ICE actions and deliberates funding for DHS, the Trump administration has already pivoted away from a full-throated endorsement for the Minnesota crackdown.

Traders now think it’s quite likely that the ICE members involved in the Pretti shooting will face consequences (including unpaid leave or filed charges)

Traders are, however, giving low odds to charges filed in the near term. As some have astutely pointed out, if state charges are filed in Minnesota, federal prosecutors have the chance to step in and supersede these charges, allowing for a preemptive federal pardon. Minnesota prosecutors may elect to hold off filing charges for three years(!) until the next federal administration, as murder charges do not carry a statute of limitations.

Traders also think a third civilian death at the hands of DHS, anywhere in the country, is unlikely, as the Trump administration signals a willingness to move past the incident and rein in DHS.

Greg Bovino, who led the administration’s DHS actions in Minnesota, has been demoted and returned to a more provincial role in California, with traders giving pretty good odds to something like a forced retirement within the year. Tom Homan, Trump’s border czar, has been brought in to manage the situation.

One lever that many traders are looking at is Trump’s ability to invoke the Insurrection Act. The odds of that happening over his whole term remain above 60%, even as the odds of it happening within the month have fallen to 7%. Such an action could occur close to the midterms or perhaps the federal elections in 2028.

Roundup

Amidst this turmoil, odds of Democrat control of Congress after the midterms remains stable at 30%, which mostly remains dependent on a few key primaries in Michigan, Texas, and Maine:

Odds that Congresswoman Ilhan Omar was sprayed with a dangerous substance were quite high, until forensics reports indicated it was apple cider vinegar:

And Kristi Noem, the Secretary of Homeland Security, appears to be on tenuous footing…

…with this interesting market on whether she’ll be impeached (not convicted) by the House needing more traders!

Happy Forecasting!

-Above the Fold

Let’s take a look at some of the features that have gone live since the start of 2026.

Predictle

The latest hit -le game just dropped. Unlike spelling five-letter words (boring, static, generally unrelated to geopolitics), Predictle asks you to order five markets by probability. Here’s Predictle #15:

An astute forecaster playing this game should quickly reason:

Bitcoin generally goes up and likes round numbers (let’s say 75%) odds, Sotomayor will probably resign this year to let Trump appoint an adequate replacement (66%), Israel is famously rivals with Iran and will always look for a chance to go on the offensive (90%), Elon Musk, what is he, the Paypal guy, no chance he’s a trillionaire (3%), and didn’t Trump just make some sort of Board of Peace (50%)?

Uh-oh. Well, you get the general idea. After adjusting for market inflation, I was quickly able to reach the correct answer:

New Predictles drop every day at midnight, and Manifold has already crowned its first Predictle champion. After a week-long competition among dozens of Manifold users resulted in a tie between acclaimed Oscars trader, Ziddletwix, and the creator of the hottest bot on Manifold, Sketchy.

It took a ClaudeCode-powered head-to-head tournament to yield a winner:

Charity Giveaway

Manifold is currently running a charity giveaway, ending March 1st. Users can buy tickets to support the charities of their choice, and a lucky ticket will be drawn on that date to determine the winning charity. Currently, a little-known organization called Doctors Opposing Circumcision appears to be the leading candidate at 23%. If you support one of its competitors, like GiveWell, the Internet Archive, or Lightcone, you might want to chip in; may the odds be ever in your favor!

User Calibration

The new user calibration interface brings prediction market calibration into the modern era. While Brier scores have always been an odd fit with prediction markets, the new analytics include a Sharpe ratio, your max drawdown in the last 12 months, and a volatility score.

You can also track your calibration as a function of probability buckets. For example, this trader made a bajillion Mana betting that Biden would not be the 2024 nominee. See if you can identify the dot corresponding to that clever trade.

Mana Shop

By now, you may have noticed a shiny new link in the menu on the Manifold homepage.

While the Mana Shop has only recently had its grand opening (with more items yet to be added), you can already go here to purchase some premium digital goods, ranging from streak freezes at 150 mana a pop to a legendary royal crown overlay.

In between these extremes, you could consider purchasing a PAMPU skin (needs no explanation) or a scholarly graduation cap. The million mana crown overlay is still unclaimed, and there’s already a market on who will be the first to wear the crown.

Perhaps most intriguing is the new Manifold subscription service, available for purchase in the shop. Subscriptions are purchasable through mana, not USD, and many are arguing that the Plus subscription should pay for itself through quest reward bonuses alone. I started off with the modest Pro membership, but I might save up to get a Premium when I have a trading week worthy of self-reward.

There are also rumors (from official Manifold representatives) that physical goods may be available in the Mana Shop soon.

You can propose your own designs and bet on which designs will eventually be added as well:

New Financial System

Amidst all these changes, the Manifold team is working hard to ensure that the interest + loans system of the site can best incentivize long-term predictions. While users can still collect daily zero-interest loans as before, there is now the option for margin loans, as well as the early repayment of old loans.

Don’t get in too much debt, however, or you’ll risk being placed for eternity on the all-time Loss leaderboard.

Happy forecasting!

-Above the Fold

Discourse around the 51st state has been all over the map for the last 12 months. Puerto Rico and DC were the popular candidates for most of the preceding decade, but early last year, amidst tariff discussions, there was discourse from the White House on Canada becoming our “cherished 51st state.” And for the last few months, despite some buzz around Venezuela, Greenland has become the trendy candidate. Randy Fine (R-FL) has even introduced a bill to this effect. Despite this, the market hasn’t moved much, still expecting that if a 51st state is added to the union, it will probably be DC or PR.

And the odds of statehood for any of these are quite low in the near term, even in light of recent events. While this might seem obvious (there have been no new states since 1959), the process for admitting a state to the union is not actually that arduous. It could be done through an act of Congress and signature by the President. And with the filibuster on life support, the Republican majority could exercise their nuclear option and admit a new state (like Greenland, say), with just 50 votes in the Senate.

If this does happen, it would likely have to happen before the midterms, when Republicans are likely to lose the House and have their edge in the Senate eroded. But statehood is not necessarily the most likely path forward for Greenland.

Designs on Greenland

The President would very much like the US to control Greenland. Whether it be for national security reasons, mineral resources, or power projection in the Western hemisphere, there is now ~30% chance that the US will acquire at least part of the island by the end of Trump’s term in office.

Traders, after taking this effort seriously for the first few months of the Trump administration, became quite skeptical, with the market probability dropping to below 5%. Since the start of the year, specifically around the military intervention in Venezuela, traders have been far more willing to believe the administration’s rhetoric. If this does go through, traders seem to think it will happen soon, with another market on a Greenland acquisition within 2026 implying >2/3 chance that if an acquisition occurs, it will occur this year.

A renewed negotiating effort with Europe, coupled with perhaps an explicit financial offer for the island, have helped convince traders. The odds of a purchase in 2026 remain low, but traders think the price tag would run into the high hundreds of billions, if not over $1 trillion, if a purchase is made.

Ultimately, if Greenland has a sale price at all, that purchase seems more than plausible. Trump, and the United States, likely value the island at around $1 trillion, equivalent to about $17 million per resident of the island country. To whom would this money go? Probably some combination of Denmark, whose Kingdom has partial sovereignty over the territory, the residents of Greenland itself, and perhaps direct investment in the island or its welfare.

But what keeps European politicians and global leaders up at night isn’t the prospect of a tough sale. It’s the prospect of a military incursion.

Ice War

It seems like just yesterday that there was lighthearted banter about the Ice War on Europa.

Now, we contend with the prospects of an actual Ice War. Odds remain quite low of an invasion of the island this year, or even limited strikes, with traders giving less than 10% odds. That being said, 9% is not nothing; if you had asked traders their true probability for this event back in 2024, they would have probably bought as many NO shares as interest rate considerations would allow.

If you expand the market to consider the use of military force in Greenland or the Panama Canal by the end of Trump’s term, a market value of 16% seems to merit taking the prospects of such an event seriously.

Still, conflict over Greenland seems unlikely. A 1-hour special operation would not be sufficient to establish American control over the island, and would risk alienating… well… nearly every American ally? In the meantime, an economic war may have already begun. Trump has promised tariffs beginning at the end of the month to several European nations who have posed obstacles in his acquisition of Greenland, and the market thinks there’s a decent chance these tariffs (which would be reciprocated by Europe) will go into effect, with some odds of a substantial escalation in the future.

Before all this dialogue, there seemed to be a chance that Greenland was reconsidering its relationship with Denmark, and might decide to become independent via referendum. These odds have fallen, in spite of pressure from the US, and I’d guess that the remaining 20% odds reflect the chance of an American acquisition, rather than a grassroots referendum. It’s likely that Greenland residents are feeling rather happy to remain Danish at this time of uncertainty.

Now, US action over Greenland threatens to splinter NATO and draw the West’s military alliances into their least aligned and most unstable era since perhaps before the Cold War. Trump is likely to threaten a US withdrawal from NATO this month, and while odds of an actual withdrawal remain lower, they are far from zero.

Greenland Roundup

Trump himself might visit the island this year, with 20% odds of setting foot on the barren wasteland by July.

He also might declare an interest in Greenland’s less-green and more-icy neighbor, Iceland.

Canada is decently likely to send troops to their Northeastern neighbor within the next month.

And markets think that in a world where Trump uses military force to take Greenland, he might face some political pressure at home, with about 30% conditional odds on his removal from office.

Happy forecasting!

-Above the Fold

Done Before Breakfast

The raid happened quickly. Friday night brought bombing campaigns across northern Venezuela targeting air defence systems. Saturday morning, around 2 AM local time, Delta Force attacked Maduro’s compound at Fuerte Tiuna, Venezuela’s largest military complex. By Saturday afternoon, Maduro and his wife, Cilia Flores, were on the USS Iwo Jima heading north. By Monday morning, both appeared in federal court in Manhattan.

Representative Brian Mast told reporters the operation was “done before breakfast.” This appears accurate. The entire raid lasted under four hours. Casualties included 80 Venezuelans and 32 Cubans who were providing security. No American casualties were reported. Pete Hegseth said nearly 200 US personnel participated. Maduro “didn’t know the troops were coming until about three minutes before they arrived.”

The operational efficiency has struck some observers as remarkable, particularly for an administration whose previous military operations have been characterised by chaos and public infighting. Conspiracy theories have emerged suggesting the raid may have been partly internally orchestrated, with elements of the Venezuelan military cooperating with US forces.

Markets give 19% odds that this was an internal coup with American assistance rather than a purely external operation.

Trump posted a photo on Truth Social of Maduro aboard the USS Iwo Jima. The White House Rapid Response account posted video of Maduro being perp-walked down a hallway with DEA agents, walking on a carpet that reads “DEA NYD.” By Monday, Maduro and Flores pleaded not guilty to narcoterrorism charges. Both remain detained at the Metropolitan Detention Center in Brooklyn, where Maduro joins fellow inmates, including Sean “Diddy” Combs and Sam Bankman-Fried.

Markets on whether Maduro would finish 2025 in power had been at 75%. They crashed to near zero on Saturday morning. Someone made $400,000 on Polymarket betting on Maduro’s ouster, with markets on who exactly the mystery trader was.

On Sovereignty (and other technicalities)

While the Trump administration frames the performance as a “law enforcement operation”, international law scholars call it a clear violation of sovereignty, the prohibition on the use of force, and head-of-state immunity. Both can be true depending on which legal framework you’re using, which is to say the legal framework is whatever the United States decides it is.

The precedent being invoked is United States v. Alvarez-Machain, a 1992 Supreme Court case holding that unlawful abduction doesn’t prevent prosecution. DEA agents had kidnapped a Mexican national from Mexico without the Mexican government’s consent. The Court said this was fine: while the arrest “technically” violated international law, domestic courts could still exercise jurisdiction.

This ruling has now been applied to a head of state.

The convenient workaround is that the United States doesn’t recognize Maduro as Venezuela’s legitimate president. If he’s not really the president, then head-of-state immunity doesn’t apply. If he’s not really the president, then this isn’t really a coup. It’s just arresting a drug trafficker who happens to live in a presidential palace.

Senator Mark Warner asked the obvious question: “Does this mean any large country can indict the ruler of a smaller adjacent country and take that person out?” The answer appears to be yes (at least if you have Delta Force and carrier strike groups).

Madurismo Without Maduro

The person currently governing Venezuela is Delcy Rodríguez, who was sworn in as acting president Sunday. Rodríguez served as Maduro’s vice president and has been under US sanctions since 2018. She’s also the sister of Jorge Rodríguez, head of Venezuela’s National Assembly. The Rodríguez siblings reportedly outlined a “Madurismo without Maduro” roadmap to US envoy Richard Grenell through Qatari channels. The Trump administration rejected it, then installed Delcy anyway.

Trump has been direct about intentions. “We’re going to run Venezuela,” he told NBC News on Tuesday. “We’re in charge.” He suggested the US could launch a second operation if Rodríguez stops cooperating, but said he doesn’t think it will be necessary. By Wednesday, Marco Rubio was backtracking, clarifying that the United States won’t “directly govern” Venezuela but will “exert tremendous leverage.”

This represents a sharp departure from Trump’s campaign promises. The man ran explicitly on ending America’s involvement in foreign conflicts, telling voters he would bring troops home and stop nation-building. Markets on US involvement in new conflicts had Trump at lower odds than previous administrations.

Since taking office in January 2025, the United States has conducted military operations in Venezuela, threatened Cuba and Colombia, maintained deployments in Syria, and expanded drone campaigns in Yemen and Somalia.

Markets on US involvement in foreign conflicts resolved YES. What Trump called ending “stupid foreign wars” appears to mean ending wars he didn’t start while beginning new ones. The Monroe Doctrine revival (or as he likes to brand it, the Donroe Doctrine) looks less like ending American imperialism and more like rebranding it.

What leverage means in practice became clearer on Thursday. Trump threatened to remove Rodríguez if she didn’t cooperate. What “cooperation” means appears to involve Venezuela’s 303 billion barrels of proven oil reserves, the largest in the world. Trump mentioned Tuesday that the United States is already recruiting American oil companies to develop these reserves. He projected it will take less than 18 months to rebuild Venezuela’s energy infrastructure, which seems….optimistic. However, it seems that Trump has all the incentive he needs.

Markets give 65% odds that Venezuelans will be better off by the end of 2026.

The New York Times reported Thursday that “day-to-day life for many Venezuelans has worsened” since Saturday. Supermarkets in Caracas are running low on supplies. Internet blackouts continue. The economy, which had contracted 75% between 2013 and 2020, isn’t improving because someone different is living in the presidential palace.

The Machado Problem

María Corina Machado won the Nobel Peace Prize in October and is widely regarded as Venezuela’s legitimate opposition leader. She won primary elections that Maduro’s government refused to recognize. She’s been in hiding since July 2024. Markets give her 43% odds of serving in Venezuelan government leadership by 2027, a 20% chance of her ever being the President of Venezuela, and a 7% chance of her entering Venezuela in January.

Machado met Trump at the White House on Thursday. What came of the meeting remains unclear. Trump has been dismissive of Machado’s authority despite her being the face of Venezuela’s opposition. The Brookings Institution notes that both Trump’s and Rubio’s contradictory visions “leave the vast majority of the Maduro regime in power,” crushing “the freedom and accountability aspirations of the Venezuelan people.”

Markets give 57% odds of Presidential elections in Venezuela by the end of 2026, and 22% odds that those elections are free and fair.

Markets also give 24% odds that Venezuela becomes more authoritarian rather than less.

These numbers suggest some understanding that removing Maduro doesn’t automatically produce democracy, particularly when the United States is threatening to remove his successor if she doesn’t cooperate.

Saturday’s Precedent

Markets had been pricing surgical strikes versus invasion. They were pricing Maduro staying in power versus opposition takeover versus military government. What they weren’t pricing was the abduction of the head of state and his subsequent prosecution in New York.

The aftermath problem remains unsolved. Removing Maduro doesn’t rebuild infrastructure, restart oil production, or address the reality that millions of Venezuelans have fled and aren’t coming back. It doesn’t resolve the question of what a post-Maduro government looks like when the opposition is fragmented and the United States won’t install Machado but also struggles to accept anyone it considers insufficiently cooperative.

Markets can price regime change. They can price economic outcomes, political transitions, and military interventions. What is much more subjective to pinpoint is what happens to global order when the world’s most powerful country decides that sovereignty, head-of-state immunity, and the prohibition on the use of force are more of suggestions than they are rules.

What a lot of people are worried most about is precedent. If the United States can bomb a capital city, kill 80 people, abduct a head of state, and prosecute him in New York while calling it law enforcement, then international law means whatever the strongest country says it means.

Happy forecasting!

-Above the Fold

Late last Thursday, DeepMind researcher Marcus Hutter announced he had made a bet with his former colleague, David Budden. Budden was coming out of stealth to announce his startup PingYou, and claimed to have—or be working on—a solution to the Navier–Stokes existence and smoothness problem.

The next day, lower down in the thread, math professor Daniel Litt made an additional bet, to the tune of $25k, that Budden had not solved the Hodge conjecture as well.

Both of these bets have a pretty strict resolution criterion: the Clay Institute, which awards the million dollar prizes for the Millennium Problems, has to recognize Budden’s solutions.

Pretty soon after this, the whole situation went fairly viral on Twitter. People were discussing AI capabilities, LLM sycophancy, the credentials of the participants of the bets, the possibility of the Millennium Prizes being solved, and the bravery of those who had taken bets.

One of Manifold’s own, Isaac King, accepted Budden’s third bet on the matter over the weekend, and Budden claimed to be ready to drop a Lean proof.

Budden’s credentials were enough to raise some uncertainty here. DeepMind has a history of some pretty impressive scientific achievements, after all. Demis Hassabis and John Jumper are still fairly fresh of their 2024 Nobel prize in chemistry for work done at DeepMind. It’s not entirely inconceivable that someone left the company, made an AI startup on formalizing or solving challenging math problems, and quickly obtained some results. If Navier-Stokes was going to get solved in the year 2026, I think the modal scenario is probably “former DeepMind engineers making a startup to pursue moonshot math research and win the Millennium Prize.”

And Budden’s willingness to make 1:1 bets on the matter was certainly evidence of strong conviction. However, I think it’s good to have some context on what Navier-Stokes and the Hodge conjecture are, and what kind of incredible leap in capabilities it would take to solve these problems.

The Millennium Prize Problems

Back in 2000, the Clay Mathematics Institute pledged $1 million to the first to solve each of the following 7 prizes:

• Birch and Swinnerton-Dyer conjecture

• Hodge conjecture

• Navier–Stokes existence and smoothness

• P versus NP problem

• Riemann hypothesis

• Yang–Mills existence and mass gap

• Poincaré conjecture (solved back in 2003 by Grigori Perelman)

Extremely naively from base rates, we should expect about one to be solved every 2-3 decades. However, a lot of people with strong beliefs in AI capabilities tend to think that a few of these problems might be uncovered by a super-intelligent or even just narrowly intelligent AI system in the nearer term. Timeline forecasts on Metaculus, for instance, give median forecasts several decades out, but have a moderately-sized peak over the next few years to account for this possibility.

This is the Clay Institute’s statement of the Navier-Stokes Problem that Budden has made two bets claiming to have a solution for:

Prove or give a counter-example of the following statement:

In three space dimensions and time, given an initial velocity field, there exists a vector velocity and a scalar pressure field, which are both smooth and globally defined, that solve the Navier–Stokes equations.

Navier-Stokes has pretty important ramifications for understanding fluids, turbulence, and many related areas of physics. The Hodge conjecture, which Budden has another bet on, is a little outside my personal understanding of mathematics, but simplistically can be described as proposing that complex geometric shapes can be constructed from simpler geometric pieces.

In any case, if any of these were solved, it would instantly slingshot the solver into global fame and fortune.

What do the markets think?

The markets around these bets are wholly unconvinced. A large volume of new traders have signed up to bet in Budden’s favor, whereas Manifold whales are staking out large limit orders at 95%. Budden himself signed up to Manifold to bet in his own favor. Early on, when it appeared that Budden might be heading a stealth startup with several AI researchers who had been working on Navier-Stokes for months and now had results to unveil, the markets were slightly more bullish in his favor. But after it came out that he was potentially still wrapping up the proofs, and was actively iterating with an AI in doing so, the markets moved rapidly against him.

To some, these dynamics seem to call back to the LK-99 drama from a few years ago, which drove a lot of engagement from new users and techno-optimists.

I don’t personally think the two situations are quite as similar as Daniel Litt does. For one, with the LK-99 drama, many materials scientists and physicists were interested in the finding, actively engaged in physical replication attempts, and working on computer simulations to help cast light on the Korean group’s results. And while many expressed skepticism that the group’s results were real, there was real uncertainty around the findings that many thought was due to the lack of experience of the researchers, the impurity of the material, or the complexity of nature.

In this case, the mathematicians that are taking this most seriously are those attempting to bet against Budden, not corroborate his findings.

Manifold seems to think that it’s likely that his Lean formalization of the proof will be incomplete or contain an error in formalization that makes it inadmissible as a solution from the Clay Institute’s perspective.

A big part of traders’ skepticism has come from the realization that Budden did not appear to have some sort of ready-to-submit solution to Navier-Stokes or the Hodge Conjecture, as a result of months of work from his startup. Rather, he was dropping half-completed files with time to go before the deadlines specified by his bets.

Some observers have been concerned for his health. Others have been more enthusiastic, claiming that he is serious and enthusiastic about these problems. Budden has made at least three (or four?) 5-figure-sum public bets at this point, and there’s a market on whether he will ultimately pay out:

AI capabilities vis-a-vis the Millennium Problems

For the record, Manifold traders do indeed seem to think that Navier-Stokes is the most likely of the Millennium Prize Problems to fall.

What’s more, they also seem to agree that AI labs have a better shot at solving it than traditional math departments!

Another market gives a 1 in 4 chance of a Millennium Prize Problem being cracked by AI within the next 3 years!

And conversely, another market thinks there’s under a 2 in 3 chance that any human manages to solve any of the remaining Millennium Problems without considerable reliance on AI means:

Ultimately, traders just do not think that this solution is likely to come from the current claimant on Twitter. Although a 2-5% chance is not zero, I think that might be reflective of prediction market discount rates more so than genuine uncertainty on the part of the market.

Whatever the outcome, I believe that (1) public figures making bets on their beliefs is a great thing for public epistemics, and (2) that this saga will ultimately result in a lot of people on Twitter learning about several unsolved problems in mathematics. Seems good.

And if you have strong opinions about it, you should go place a bet on Manifold to register your beliefs.

Happy forecasting!

-Above the Fold

This comes just a few days before the Trump administration’s December 19 deadline to release all Epstein files under the Epstein Files Transparency Act, which Trump signed on November 19 after Congress passed it 427-1 in the House and unanimously in the Senate.

Markets currently give around 37% odds that the files will actually get released by the deadline.

Transparency (terms and conditions apply)

The Epstein Files Transparency Act requires the Justice Department to release all files related to Epstein and Ghislaine Maxwell within 30 days of passage. The language has been made clearer than day: make publicly available in searchable and downloadable format all documents, records, and communications.

The loopholes, it turns out, are also clear. The DOJ can withhold material related to ongoing investigations, national security, victim privacy, or anything depicting child sexual abuse. It can redact information that would constitute “clearly unwarranted invasion of personal privacy.”

Trump ordered the DOJ on November 15 to investigate Epstein’s connections to prominent Democratic figures, including Bill Clinton and Larry Summers. This investigation, conveniently launched after the bill passed, now provides legal cover to withhold documents under the “ongoing investigation” exception.

Attorney General Pam Bondi told reporters Wednesday the DOJ has already released 33,000 Epstein documents to Congress and will “continue to follow the law and to have maximum transparency.” She emphasised repeatedly that “the law as written allows us to protect victims.”

The Senate Judiciary Committee is less confident. Senators Adam Schiff and Dick Durbin requested an independent audit of the DOJ’s handling of the files, specifically asking whether there are “irregularities in the chain of custody.” They noted that Judge Paul Engelmayer has already chastised DOJ for “misleading portrayals of the evidence” and “paying lip service to Maxwell’s and Epstein’s victims.”

Five members of Congress from both parties sent Bondi a letter on December 3rd requesting a briefing on the department’s progress. The letter notes they are “particularly focused on understanding the contents of any new evidence, information or procedural hurdles that could interfere with the Department’s ability to meet this statutory deadline.”

A Series of Clarifications

Trump’s position on the Epstein files has moved through five phases in twelve months.

Campaign 2024: he’d “absolutely” release them. February 2025: Bondi announced the “client list is sitting on my desk.” July: a DOJ-FBI memo stated no client list exists. August: Trump called the whole thing “a hoax.” December: when asked about the photos, he said “no big deal.”

There is also the question of what exactly Trump’s plan is with the release. At this point, there is little doubt on his involvement with Epstein. The question we should be asking is why Trump was immediately so happy to release the files following weeks of hesitation.

On Tuesday afternoon, three days before the deadline, White House Chief of Staff Susie Wiles told Vanity Fair she’s read what she calls “the Epstein file.” Trump is in it, she confirmed. “And we know he’s in the file. And he’s not in the file doing anything awful.” Trump and Epstein were “young, single playboys together,” she said. Trump was on Epstein’s plane. He’s on the manifest.

Wiles also contradicted Trump’s repeated claims about Clinton. “There is no evidence” Clinton visited Epstein’s island, she said. “The president was wrong about that.”

Trump-style Contraception

House Democrats are releasing the 95,000 photos in batches. Representative Robert Garcia, ranking Democrat on Oversight, said Friday they’ve reviewed about 25,000 so far. Some photos, he noted, are “incredibly disturbing.”

The photos released Friday show Trump in various social settings with Epstein. One shows Trump with six women (faces redacted). Another shows Trump and Epstein talking to a woman at what appears to be a party. A third shows Trump on a plane next to a woman (face redacted). There’s also a photo of novelty condoms featuring Trump’s caricature with the text “I’m HUUUUGE!” in a bowl marked “Trump condom $4.50.”

Trump, to my utter dismay, has yet to capitalise on this untapped market segment, and has sadly NOT launched his own signature brand of condoms (yet?), to add to his exciting line-up of merchandise.

Unfortunately, it also seems like the Manifold Store will NOT be releasing its own Manifold-branded condoms anytime soon, much to the disappointment of anticipating traders.

Trump, asked about the photos Friday evening in the Oval Office, said it was “no big deal.”

Republicans on the committee accused Democrats of “cherry-picking photos and making targeted redactions to create a false narrative about President Trump.” They note that Democrats released 19 photos out of 95,000 and point out that nothing in the documents shows wrongdoing by Trump.

Trump and Epstein’s relationship goes back decades. In 2002, Trump called Epstein “a terrific guy” and noted he “likes beautiful women as much as I do, and many of them are on the younger side.”

The more serious allegations sit at different odds. Markets give low odds Trump faces criminal charges related to Epstein. Markets on whether Trump had sexual contact with minors through Epstein: 54%. Whether that will ever be proved in court? 17%

Trump has said he kicked Epstein out of Mar-a-Lago for inappropriate behaviour, though accounts vary. Trump’s name appears in Epstein’s black book. The Wall Street Journal published a birthday card from Trump reading “may every day be another wonderful secret.”

The timing is notable. Democrats are releasing photos that put pressure on the administration one week before the statutory deadline.

Markets on “Will Trump be accused by any trustworthy media outlet of being a Pedo with Epstein?” are at 9%.

The photos add political pressure but don’t necessarily add legal pressure. Being photographed with someone isn’t illegal (at least that’s the angle the administration is going with).

What the photos do is create a news cycle that makes it harder for the administration to quietly delay or heavily redact the files. Garcia told reporters: “The Department of Justice must release all the files, NOW.”

Markets give 8% odds that any top officials leave the Trump administration over the Epstein case, down from 57% in July.

Attorney General Pam Bondi told reporters Wednesday the DOJ has already released 33,000 Epstein documents to Congress and will “continue to follow the law and to have maximum transparency.” She emphasized repeatedly that “the law as written allows us to protect victims.”

On Tuesday, Wiles told Vanity Fair that Bondi “completely whiffed” on the Epstein files. In February, Bondi gave conservative influencers binders labeled “The Epstein Files: Phase 1” that contained no new information. “First she gave them binders full of nothingness,” Wiles said. “And then she said that the witness list, or the client list, was on her desk. There is no client list, and it sure as hell wasn’t on her desk.”

The List of Lists

We also see plenty of markets on who gets named, with quite a few already having been resolved in recent file releases.

Most of these people are already documented. Clinton flew on Epstein’s plane twenty-six times. Prince Andrew settled with Virginia Giuffre. Trump’s 2002 New York Magazine comments about Epstein liking younger women are public record.

Gates started meeting with Epstein in 2011, three years after the conviction. They met repeatedly through 2014. Gates says it was about philanthropy. Melinda Gates cited the meetings in their divorce. He’s at 71% to be named, 3% to face charges.

What’s more interesting is observing the pricing gap between being named and being charged.

Dershowitz asked Trump for a pre-emptive pardon for Maxwell in January 2021. Trump said no. Dershowitz has denied Giuffre’s allegations and is suing Netflix. Markets give him an 18% chance of indictment.

Leon Black paid Epstein $158 million between 2012 and 2017 for financial advice. After the conviction. Black left Apollo in 2021 when the payments became public. An investigation confirmed the money but found no criminal involvement.

Then there are markets that exist just for the mick of it.

Suicide Watch (and other oversights)

Manifold gives Epstein’s death 64% odds of actually being suicide.

While official investigations concluded suicide, only 21% of Americans believe the official conclusion, according to polls.

The DOJ-FBI memo released in July 2025 stated definitively: Epstein died by suicide, had no client list, didn’t blackmail prominent figures. The memo was signed off by an FBI now led by Kash Patel and Dan Bongino, both of whom spent years promoting conspiracy theories about Epstein before their appointments.

Patel, before becoming FBI director, appeared on Benny Johnson’s YouTube show in 2023 and said: “Put on your big boy pants and let us know who the pedophiles are.” He was addressing the then-leaders of the FBI about the Epstein files.

Bongino told his podcast audience in 2023: “Listen, that Jeffrey Epstein story is a big deal, please do not let that story go.”

Both men appeared on Maria Bartiromo’s show in May 2025 and said Epstein died by suicide. Markets on whether Bongino leaves his position spiked.

He was demoted to Co-Deputy Director in September.

The circumstances of Epstein’s death continue to fuel scepticism. He was on suicide watch, then removed from it. His cellmate was transferred the night before his death, no replacement assigned. The two guards allegedly didn’t check on him despite regulations requiring checks every thirty minutes. They falsified logs. The cameras outside his cell malfunctioned.

Dr. Michael Baden, the forensic pathologist hired by Epstein’s brother, observed three fractures in Epstein’s neck: both thyroid cartilages and the hyoid bone. He told 60 Minutes he’d never seen three fractures like that in over a thousand jail suicides. The medical examiner’s office maintains such fractures can occur in both suicides and homicides.

Ghislaine Maxwell, serving twenty years for sex trafficking, told DOJ investigators she doesn’t believe Epstein died by suicide. She clarified she doesn’t think it was a hit from wealthy figures: she thinks it was internal to the jail. “In prison, where I am, they will kill you or they will pay somebody to kill you for $25 worth of commissary.”

Markets give 18% odds Trump had Epstein killed.

Markets also give 1.5% odds Epstein will be found alive by 2028.

Thursday (maybe)

December 19 is three days away. Markets give 37% odds the files get released as mandated on time.

If the DOJ releases heavily redacted materials citing ongoing investigations and victim privacy, they technically comply while revealing little. If they release unredacted files, they create exposure for numerous prominent figures. If they miss the deadline, they face criminal implications that may never materialize.

Markets appear to be forecasting the first scenario. House Democrats seem to understand this, and they’re releasing photos from Epstein’s estate that don’t require DOJ cooperation. Garcia said Friday they’ll “continue to release photos in the days and weeks ahead.” The details seem to keep on coming, whether the DOJ complies or not.

Happy Forecasting!

- Above the Fold