A newspaper editor texts Polly Curtis to ask who will be prime minister by Christmas. She replies "dunno." He tells her she is off his clever list. That exchange, recounted in the Guardian this week, is the perfect miniature of where British political analysis has ended up. The interesting question has been swapped for the bettable one.
Curtis, who runs Demos, is arguing that the bettable question is a distraction. Voters are not giving up on Starmer, or on Labour, or on this particular cabinet reshuffle cycle. They are giving up on democracy itself. And while she writes that, traders on Polymarket are doing the thing she is warning against: pricing the surface.
What the market is actually betting on
The "Starmer out by...?" market on Polymarket exists because the leadership question has become a tradeable event, not because anyone thinks the answer matters in the way it used to. It is a ladder of dated contracts, each resolving Yes if Starmer ceases to be prime minister before the deadline, and right now the order book leans hard toward exit rather than survival. A separate "Next UK Prime Minister in 2026?" market prices the successor question, where Andy Burnham leads the field and Wes Streeting has slipped to the margins. These are clean binary questions with calendar deadlines, which is exactly what prediction markets are built to price.
The trouble is that Curtis's argument cuts underneath every one of those contracts. Her point is that the prime minister's identity at the December 31 deadline is roughly orthogonal to the underlying political problem. If trust in democratic institutions keeps eroding at the pace the focus groups suggest, the country becomes ungovernable regardless of who is at the despatch box. That is not a position a binary market can express.
Which is the recurring limit of how prediction market odds work on questions like this. The market gives you a probability for the named event. It does not give you a probability for the framing being correct.
The Burnham-Streeting noise versus the Curtis signal
The last six weeks of UK political markets have been dominated by leadership-substitute trades. We covered the live version of this in the Streeting standoff, where Polymarket traders spent days re-pricing the health secretary's odds on every leaked WhatsApp. Those bets live in the separate "Next UK Prime Minister in 2026?" market, and the field has moved on without him: Andy Burnham now leads it after his Makerfield run, while Streeting has drifted to the margins. The volume is real. The signal value is thinner than the volume suggests.
Curtis's Demos paper, published the same day as her column, frames the problem as a broken relationship between state and citizen. Her metric is not approval ratings. It is the share of voters who believe the system itself can deliver. That number, on the polling she cites, has been sliding for years across multiple governments of multiple stripes. Reform's polling rise, the Green bounce, the focus-group fatigue: they are symptoms of the same underlying erosion, not separate stories.
Markets are quite good at pricing symptoms. They are bad at pricing the meta-question, partly because nobody knows how to write the contract. "Will UK democratic trust recover by 2030" has no resolution source. "Will Starmer be out by December 31" resolves on a Wikipedia edit. Guess which one gets the order book.
What a hope-loop contract would even look like
Curtis's most interesting move is the suggestion that a doom loop has a mirror image. A hope loop, in her framing, would be a self-reinforcing cycle where institutional reform produces visible improvements which produce renewed trust which produces appetite for further reform. It is the same feedback dynamic running in reverse.
This is the kind of claim that sounds soft until you try to operationalise it. What would a tradeable hope-loop contract look like? You would need a measurable proxy. Turnout in the next general election above some threshold. Trust-in-government polling crossing back above a benchmark by a fixed date. Some constitutional reform package passing by a deadline. Each of those is closer to the underlying question than "who is PM at Christmas" but further from the kind of clean resolution criterion that draws liquidity.
This is the chronic mismatch between prediction markets and polls on questions of political legitimacy. Polls measure the mood directly, badly. Markets measure the events that mood produces, sharply but narrowly. Neither gives you the thing Curtis is actually pointing at, which is the rate of change of democratic trust itself.
The honest editorial take
Curtis is right that the leadership parlour game is consuming attention that should be on the institutional question. She is also right that the honest answer to "who is PM at Christmas" is closer to "dunno" than the order book admits, because the variables driving it are too coupled and too contingent for confident calls. And note which way the book leans: the money is on an exit before year-end, not on a steady hand riding out the storm. That is the surface symptom of exactly the erosion Curtis is describing. The Polymarket price is a useful summary of crowd opinion on a specific calendar event. It is not a forecast of British political stability, and treating it as one is the analytical mistake she is warning against.
The more useful read of the current market is contrarian in a quiet way. If Curtis's framing is correct, then the leadership contracts are pricing the wrong question with too much confidence on both sides. The right trade is probably not in the leadership market at all. It is in whatever venue ends up listing the next general election outcome, where the doom-loop dynamic, if it is real, will eventually have to resolve into seats.
iPredicta is tracking the UK political book across Polymarket and the regulated venues, including the "Starmer out by...?" ladder, the "Next UK Prime Minister in 2026?" market, and the longer-dated general election markets where the underlying trust question will eventually price. The exit and successor contracts are on the watch list, but they are not the ones we think will matter most.
Frequently asked questions
Why do prediction markets price leadership questions so heavily?
Leadership contracts have clean binary resolution, short time horizons, and obvious news catalysts, which is exactly what draws liquidity. Deeper questions about democratic trust or institutional legitimacy do not resolve cleanly on a fixed date, so they rarely attract enough volume to produce a meaningful price.
Is the Polymarket "Starmer out by...?" market a reliable forecast?
It is a reliable summary of where money is willing to sit on one specific question: whether Starmer leaves office before a given date, with the December 31, 2026 contract trading around 72% Yes as of early June 2026. It is not a forecast of broader political stability, voter trust, or election outcomes, and reading it as such overstates what the contract can do.
