Wisdom of crowds 👥

In early January, someone on prediction market site Polymarket put about $30,000 on a low-odds contract that Venezuela’s president, Nicolás Maduro, would be out of office by the end of the month. And then Maduro was captured in a US special operation. That position reportedly turned into a profit of roughly $400,000. 

After the Maduro operation, Polymarket faced backlash for refusing to pay out contracts that bet on whether the US would “invade” Venezuela. The platform argued the operation did not meet the specific contract terms, triggering complaints about arbitrary enforcement.

A prediction market is an exchange where people buy and sell contracts tied to a real-world outcome. They aggregate scattered information into a single forecast and the price is usually read as an implied probability. If “Yes” trades at 60 cents, the crowd is saying “about 60%”. Research by Wolfers and Zitzewitz suggests market-generated forecasts are often fairly accurate and can outperform common benchmarks in some settings. 

In a way, a prediction market is the simplest financial product possible: a contract that pays out if something happens. The price is a live, continuously updated number that looks a lot like capital-T truth.

Betting volumes across the space have surged recently. But there are a number of caveats: liquidity, manipulation risk, and regulatory confusion. Nonetheless, prediction markets are becoming part of the information environment that moves prices. A prediction market that implies a higher probability of tariffs or a higher probability of rate cuts, becomes a piece of public narrative. Screenshotted, shared, debated, this is what-the-market-is-saying content. And that can feed into positioning in real markets.

Intelligence service

A CIA Studies in Intelligence paper from 2006 argued prediction markets are a good way to improve intelligence work. Essentially, a ‘Standard & Poor’s 500 Index’ for internal forecasts, aggregating dispersed judgement into one continuously updated price. It claims “market-generated predictions” can act like a kind of index for belief inside the intelligence community.

DARPA’s FutureMAP programme (launched in 2001) and the Policy Analysis Market (PAM) would have allowed trading on political, economic and military indicators, even ‘conditional’ contracts that let traders express relationships between events (if X happens, what does that do to the odds of Y?). It was cancelled in 2003 after political backlash – “terrorism betting parlors” – but the CIA paper argues the experiment was mischaracterised and ended prematurely.

The most compelling use-case for prediction markets, therefore, may be as a mechanism for forcing analysts to put numbers on their beliefs and for surfacing disagreement early before it ossifies into consensus.

Odds on

Prediction markets sit at the intersection of three trends. One, people want probabilities and are over punditry. Trading apps have trained everyone to think in a buy/sell binary. And politics and geopolitics are producing more binary events than ever before, such as elections, court rulings, wars, sanctions, and central bank decisions.

This month, Polymarket was integrated into the Golden Globes broadcast, with the platform calling 26 of 28 winners. During a recent US winter storm, people were trading contracts on snowfall totals with hundreds of thousands wagered across platforms. 

South Park did an episode in September 2025 that skewered prediction markets, name-checking platforms like Polymarket and Kalshi – “you can bet on literally anything”. In suitably self-referential fashion, traders created markets about the episode itself including contracts on whether certain words or names would be said on screen. 

But prediction markets’ breakout moment was during the 2024 US presidential election. Reporting around that period put election market volumes in the billions on major venues. That created a feedback loop. The odds became news, which drew more attention, which drew more trading, which made the odds feel even more ‘real’. It also produced the era’s signature controversy: the whale problem. A single trader (or coordinated set of accounts) can be a meaningful share of liquidity in a market that the public is treating as gospel. Coverage of Polymarket’s large pro-Trump positions described how a handful of accounts dominated flow and triggered constant arguments about distortion. Big trades don’t have to change the outcome to change the narrative. If the number itself becomes part of the media cycle, moving the number can matter.

Medium is the message

A lot of market information already exists in prices, but it’s encoded. Prediction markets provide a headline-friendly output. That is catnip for journalists and social media. The issue is that simple numbers can still be messy underneath. As we’ve seen, prediction markets can be moved by a few large traders and pushed around by people trading for reasons other than Truth (hedging, activism, attention, vibes). A lack of liquidity leaves these markets prone to distortions and harder to use as reliable hedging tools.

Venues match buyers and sellers and take fees, or bake in costs via spreads, or charge per contract. Kalshi, for example, is positioned as a regulated exchange and talks about its rulebook and market structure in those terms. Polymarket says it does not charge trading fees and does not “benefit from the outcome” of markets. Users may still face network and transaction costs depending on how they interact.

The economics of the venue shape behaviour. If your platform gets paid on volume, it will cultivate the hype. If it gets paid on spreads, it will like churn. If it gets paid on listings, it will like ever more markets.

In the US, prediction markets tend to fall under the Commodity Futures Trading Commission, because event contracts look like derivatives. Polymarket has already been on the wrong side of the regulator once. In 2022, the CFTC announced an order settling charges against Polymarket for offering off-exchange event-based binary options and failing to register. The order included a $1.4m penalty and required the platform to wind down certain markets and implement more robust checks and balances. 

Kalshi says it received CFTC approval as a Designated Contract Market in 2020. Kalshi fought the CFTC over political event contracts, and a federal appeals court decision lays out the dispute over whether certain “Congress control” contracts should be blocked as gaming.

The small print

The binary contracts look simple — yes or no — until someone has to decide what counts as ‘yes’. Did an event happen ‘by a date’ or ‘on a date’? Does a leader ‘leave office’ if they flee the country or only if they resign? What counts as ‘control’ of Congress: the election result, the seat count, the swearing-in? This is where prediction markets collide with the real world. In ordinary markets, insider trading is already hard enough to tackle. Here, the underlying events can be even more sensitive – military operations, sanctions decisions, prosecutions, hostage negotiations – and the pool of people who could have privileged information can be large.

In financial markets, insider trading is usually about earnings or deal talks. In prediction markets the most valuable information is often procedural such as when a regulator will announce something, when a court ruling will drop, when a sanctions decision is agreed, when an operation is planned. Prediction markets want informed trading – that’s the whole argument economists make about why they work – but ‘informed’ can slide into ‘privileged’ very quickly. 

Lay of the land

Polymarket has faced regulatory attention and access restrictions in parts of Europe. France’s gambling regulator (ANJ) has publicly warned about Polymarket and has asked ISPs to block access. Belgium’s Gaming Commission blocked Polymarket last year. 

Meanwhile, in places like the UK and Australia, there’s a long tradition of political betting in regulated gambling frameworks. Matchbook says it plans to launch a UK-licensed prediction market platform soon, pitching it as a simpler, probability-style way to trade outcomes on sports (and potentially politics) under a Gambling Commission licence. UK regulators have decided prediction markets fall under the Gambling Commission, effectively treating them as a form of betting exchange.

Where it gets awkward is when ‘event contracts’ try to look like financial instruments rather than bets. Jake Green at Ashurst points out that if the contract references a financial benchmark or statistic, it’s likely to be treated as a financial instrument – often in the direction of a binary-option-style payoff – whereas non-financial events look much more like straight betting under the Gambling Act.

The FCA made its permanent ban on selling binary options to retail consumers effective from 2 April 2019. That makes it hard to run a UK retail product that smells like yes/no options on markets, even if you wrap it in friendlier language.

Up to code

Prediction markets are just like regular markets but with a headline-friendly output. They can be genuinely informative when liquid, well-specified, and hard to game. But they can also be brittle and far from a golden source of truth when the rules are unclear or incentives are bad. The Venezuela episode – the suspicious Maduro trade on one side, and the settlement dispute over ‘invasion’ wording on the other – is a neat reminder that how a market is built matters as much as what it predicts.

‍