The Prediction Market and the Wisdom of the Few

Galton, Surowiecki, Tetlock, Hayek, and Kyle walk into the old Manoir de Paris to ask whether a prediction market aggregates the wisdom of the crowd or the edge of the informed minority that trades against it.

Between 2011 and 2021, the building at 18 rue de Paradis, in the tenth arrondissement of Paris, was the Manoir de Paris, a haunted house where a troupe of actors restaged the city’s most notorious crimes for paying visitors. In the late nineteenth century the same address was the showroom of the Choisy-le-Roi faience works, which is why the floor and walls still carry their mosaics. On Monday night it was Albert School, the data and business school backed by Xavier Niel and Bernard Arnault, and under its glass roof, on the wide wooden steps that seat about a hundred and sixty, three hundred people had come to argue about whether betting on the future is gambling or information.

The evening was the first Prediction Meeting Paris, convened by Mathis Chau with co-organisers including Thomas of Hyperliquid France, the Shard team, and Thomas Vandermarcq of Kash, around the question printed on the invitation: are prediction markets the next infrastructure of information? A prediction market lets you buy and sell contracts that pay one unit if a stated event happens and nothing if it does not, so the price of the contract, sitting somewhere between zero and one, reads as the market’s estimate of the probability of that event. Polymarket and Kalshi are the two that matter; a cluster of younger French teams pitched around them before the talks.

I came to this with a long interest rather than a professional one.¹ Some of the books that shaped how I think about decisions under uncertainty sit squarely on this subject, among them James Surowiecki’s The Wisdom of Crowds (2004), Philip Tetlock and Dan Gardner’s Superforecasting (2015), and the wider literature on judgment and markets that runs through Michael Mauboussin, Daniel Kahneman and Amos Tversky, and Nassim Taleb. So I listened for one thing in particular. Every speaker reached, at some point, for the same phrase to dignify what these markets do: la sagesse des foules, the wisdom of the crowd. I think the phrase is the wrong one, and the gap between that slogan and the actual mechanism ran through the whole evening.

The pitches and the institutional turn

The pitch round gave the measure of how fast a small sector is being built.

• Polycool is putting prediction markets into a native mobile app, a genuinely hard thing to ship past Apple and Google review.
• Shard is building a customisable dashboard that pulls third-party tools in as widgets, with a privacy mode and multi-wallet signing.
• Kash, Vandermarcq’s company, wants to take markets onto social feeds, so that anyone can launch one inside a reply on X, with clubs and brands as the obvious customers.
• Radion, Chau’s own company, sells the layer underneath all of them, a high-performance data API for builders who would otherwise lose weeks rebuilding the same plumbing on top of Polymarket’s endpoints.

The numbers behind the category move fast enough to make any slide stale by the time it goes up. Armand Droué of Kalshi showed the valuation climbing with each new market the exchange opened, from weather and macroeconomics to politics, then sport, then culture, to the $22 billion round it closed in May 2026, double its valuation five months earlier.² Four days before the meetup the Financial Times reported Kalshi in talks to raise again at roughly $40 billion, with Polymarket said to be raising at about $15 billion and the Intercontinental Exchange, owner of the New York Stock Exchange, already committed to up to $2 billion of it.³ Annualised volume on Kalshi has run past $178 billion, and the slice coming from institutions rather than retail grew 800% in six months. Between the slide being designed and the room sitting down, the number on it had nearly doubled in rumour.

What lifts the institutional story above a pitch is where Kalshi’s data and contracts already sit. Its market data is carried on the Bloomberg Terminal, the screen every trading desk watches, and distributed through Tradeweb. Trading Technologies, the order-routing system behind tier-one banks, hedge funds, and proprietary desks, is wiring Kalshi in as one more venue alongside futures and options. Morgan Stanley joined the May round; Goldman Sachs and JPMorgan have both said they are studying the space, and JPMorgan felt the need to warn its three hundred thousand staff, in writing, against trading on workplace information through these venues, which is itself a measure of how close they have come to mattering.⁴ In December 2025 Kalshi signed deals with CNN and CNBC, becoming the sole prediction-market feed on CNBC, its odds now scrolling along the ticker during Squawk Box and Fast Money.⁵ None of that is the posture of a betting app.

The strongest single piece of evidence for the information thesis is the 2024 US presidential election. Through the race, the markets read the result better than the experts. On the eve of the vote Polymarket had Trump near 58% against Harris while the polls and the major forecasters had it a coin-flip, and the market proved the truer account of the night.⁶ A Wake Forest economist later called the markets far and away the best forecast of the election. That one call did more to legitimise the field than any amount of theory, and it is a large part of why a room in Paris was full on a Monday night.

Pierre Tramon and the information thesis

Pierre Tramon, a senior associate at FDJ United Ventures, the venture arm of one of Europe’s largest gambling operators, made the case that these are not betting platforms at all. In a sports bet, an operator sets the odds and carries the risk, acting as the counterparty to every wager and managing the return-to-player rate, the share of stakes it pays back to players over time, which the house tunes to stay profitable. In a prediction market there is no such operator. The platform only matches buyers and sellers, and the price forms organically out of what they are willing to pay each other. That design is old. The betting exchanges of around 2010, Betfair and Smarkets and the rest, already let punters trade positions against each other under the UK Gambling Commission, rather than against a bookmaker.

Tramon’s thesis, and the reason a gambling group’s fund has been looking at this, is that the output is a signal, not entertainment. He sees prediction markets becoming an information utility for insurers, energy firms, and public institutions trying to price climate, commodities, and political risk, with FDJ positioned to move if the rules allow it. It reminds me of Friedrich Hayek, who argued in his 1945 paper The Use of Knowledge in Society that a price is a remarkable instrument for compressing knowledge no single mind holds, the dispersed, local, often tacit information scattered across thousands of heads, into one number that coordinates them. A market on an event is Hayek’s argument run in the future tense.

Three doors in France, two in America

France and the United States looked at the same product and disagreed. In November 2024 the ANJ, the French gambling authority, ruled Polymarket unlicensed gambling and forced a geoblock that still holds, one of more than thirty national blocks.⁷ The United States went the other way. The CFTC lets Kalshi and Polymarket operate nationally as exchanges, a far stronger footing than a bookmaker’s state-by-state licence, though more than a dozen states are now suing both for unlicensed sports betting, the Kentucky attorney general putting sport at 89% of Kalshi’s volume, and the jurisdiction fight is widely expected to reach the Supreme Court.⁸ Tramon’s route to European legality runs instead through the AMF, the markets regulator, which would reclassify the contracts as financial instruments, though even that would not close the perimeter, since holding player accounts and moving money pulls in the ACPR and its anti-money-laundering duties.⁹ One contract can sit inside gambling law, market law, and payments law at once, with no authority having clearly claimed it.

Galton’s ox and Surowiecki’s four conditions

In 1906 Francis Galton watched villagers at a Plymouth fair guess the weight of an ox, and when he collected the 787 valid tickets, the middle estimate landed within 1% of the animal’s true weight.¹⁰ He titled the note Vox Populi. Surowiecki built his 2004 book on cases like it, and named the conditions under which a crowd is wise: (i) diversity of opinion, (ii) independence between the guesses, (iii) decentralisation of local knowledge, and (iv) a mechanism that aggregates them. Each person’s error is roughly independent of the next, the errors point in different directions, and when you average them they cancel, leaving the signal. The crowd is wise because no one’s mistake is correlated with anyone else’s.

A prediction market is a different machine. It does not give each participant one equal vote and average the result. It weights by capital and conviction, and the price is set at the margin, by whoever is willing to trade next at that level. If you think the true probability is higher than the price, you buy until the price reflects your view or your money runs out. The informative content comes disproportionately from the most informed and most confident capital correcting everyone else, not from the average of the room. A crowd of mostly wrong guessers produces a wrong median; a market full of mostly wrong traders can still print the right price, because a few who know better move it and are paid for doing so.

Insiders and the informed margin

The tension surfaces in how the speakers talk about insiders. If a market really ran on the wisdom of a crowd, an insider would be a contaminant to keep out. The speakers treated the insider as something to design around instead.

An audience member pushed back on Droué directly, saying he struggled with the idea that people predict the future well simply because there are many of them and they are risking money. Droué’s reply pointed the other way. The value, he said, comes from forcing someone to put risk on an opinion, so that the uncertain abstain and only conviction trades, which sharpens the signal. And for something like the weather, the edge comes from the rare participant with a better instrument than the published forecast, whose private reading gets pulled into the price. Both answers describe the informed margin, a small number of better-informed traders overruling the rest, not the wisdom of the crowd.

A contributor talking about futarchy, a proposed system of government in which prediction markets, not votes, decide which policies are expected to work,¹¹ put the point at its sharpest. In these markets, he said, we accept that there are insiders; insiders are part of every market, and you fight them with the law where you can, but you build assuming they exist. A wisdom-of-crowds mechanism is destroyed by an insider, because the insider breaks the independence that makes the averaging work, a correlated, better-informed voice corrupting the cancellation of errors. A market wants the insider, because the insider is the informed margin that drags the price toward the truth. Droué later recounted a contract on whether Giannis Antetokounmpo would change teams, which drew accusations that Kalshi had inside knowledge, and his defence, in passing, was that the market had simply attracted whoever knew something.

Market microstructure named this mechanism decades ago. Albert Kyle’s 1985 model Continuous Auctions and Insider Trading describes exactly how a single informed trader’s private signal enters the price: the insider submits orders that mingle with the orders of noise traders, people transacting on hunch or liquidity needs rather than information, and a market maker, unable to tell which is which, adjusts the price as the informed flow tilts it.¹² Fischer Black’s 1986 address Noise supplies the other half. Without the noise traders, the informed could not profit and could barely trade, since there would be no one on the other side, and the price would aggregate nothing.¹³ In that frame the crowd in a prediction market is the noise, the volume the signal feeds on. Calling that the wisdom of the crowd inverts the roles. The crowd supplies the liquidity; the informed few supply the signal, and are paid for it by the noise of the many.

Tetlock against the crowd

Tetlock found the same thing a decade earlier. Superforecasting reports the results of the Good Judgment Project, which won a US forecasting tournament run between 2011 and 2015, and the central finding cuts against naive crowd wisdom. A small, identifiable subset of forecasters, the ones Tetlock calls superforecasters, beat both the crowd average and credentialed experts, and reportedly beat intelligence analysts who had access to classified material while the superforecasters worked from the open web.¹⁴ The winning aggregation weighted the forecasters with a track record more heavily and pushed the combined estimate further from the middle, rather than averaging everyone equally.

Outside the lab, Ray Dalio built Bridgewater on the same instinct. His idea meritocracy ran on what he called believability-weighted decision-making, an internal app that scored employees by track record and weighted their votes accordingly, so the most reliable voices counted for most. It is the human cousin of what a market does with money. The cautionary coda is that Rob Copeland’s 2023 book The Fund reports that Dalio had the weights rigged to keep his own score on top once two employees outscored him, a charge Bridgewater disputes, and that is its own lesson in how fragile a weighting engine becomes once someone can put a thumb on it.¹⁵

A market does the same thing with money instead of scores. Correct capital compounds and wrong capital is drained, so over time the confident and right come to set the price, which is closer to Tetlock’s weighting of the calibrated few than to Galton’s average of all. Described honestly, a prediction market pays an informed minority to reveal what it knows and lets that minority’s capital overrule the noise. That claim is less flattering than the civic one the speakers prefer, and it tells you who the price actually belongs to.

On the institutional question I will venture an opinion of my own. If sophisticated hedge funds move in with better information and faster execution, I think they become the mechanism rather than breaking it. Arbitrage and superior models push prices closer to the truth, exactly as they do in equities, which is good for accuracy and fatal for the romance of the crowd, since the informed margin is then simply professional rather than a lucky retail whale. I doubt they collapse into one another, because competition keeps pushing each desk to add data and differentiate in order to protect its edge, so independence is sustained by rivalry rather than dissolved by it. The constraint I would worry about instead is the one Black named. The professionals need the amateurs, because you can only profit from noise traders willing to lose to you, so a market drained of retail flow thins out and aggregates less, which is part of why institutions already worry aloud about squeezing out the retail traders they feed on.

The limits of the information thesis

The mechanism holds, but the optimism around it still runs ahead of the evidence on three counts: the bias of the prices, the meaning of a price on a one-off event, and the integrity of the market itself.

The prices are biased at the edges. Across decades of betting and prediction-market data, low-probability outcomes are systematically overpriced and favourites underpriced, the favourite-longshot bias that survives in market after market.¹⁶ A mechanism advertised as a neutral information utility carries a known, persistent distortion exactly where the tails live, which is where insurers and risk managers most need it to be right.

Then there is the question of what a price on a singular event even means. Several of the warmest examples on stage were one-offs: a contract on whether two celebrities would separate, pitched as a hedge LVMH might want before launching a perfume tied to one of them; a Manhattan bar covering the free drinks it had promised if the Knicks won. For genuinely non-repeatable events, the market “probability” is not a frequency of anything, and Nassim Taleb has argued, in his 2018 paper Election predictions as martingales, that a forecast far from its resolution date cannot honestly sit far from even odds without violating the no-arbitrage logic that a real probability must obey.¹⁷ This is the distinction Frank Knight drew in 1921, in Risk, Uncertainty and Profit, between measurable risk and unmeasurable uncertainty, the same distinction Bill Janeway draws in Doing Capitalism in the Innovation Economy: treating a thin market on a unique event as if it returned a calibrated probability is treating Knightian uncertainty as if it were risk.

The integrity problem is the one the speakers half-acknowledged and the regulators have documented. In November 2025, a Polymarket contract on whether the Russian army would take a particular Ukrainian crossroads was resolved after someone with access to the underlying Institute for the Study of War map altered it to show a fictitious advance, let the position pay out, then reverted the edit.¹⁸ The ANJ’s own warning goes further: once a participant can both bet on an event and influence it, the market creates a financial incentive to cause the outcome, whether that means throwing a match or something worse in a geopolitical contract. Thinness compounds the danger. Thomas, of Hyperliquid France, spoke for Hyperliquid, the Singapore-based exchange that is, by revenue per employee, the most efficient company on earth, around $1 billion a year from roughly eleven people on no outside capital.¹⁹ He was candid that even there the prediction markets, barely two months live, were showing spreads of 10% to 35% on contracts with millions in volume, which is a market aggregating almost nothing.²⁰ Mauboussin’s reading of crowd wisdom is the right caution here. A market is informative only while diversity and independence hold, and collapses into noise or a bubble when they break. The law adds a gap of its own. Because event contracts are not classified as financial instruments, as the legal scholar Pauline Pailler has noted, the market-abuse regime does not reach them, so the platforms can openly welcome the insiders that securities law would prosecute.²¹ The comfort with insiders on stage is partly an artefact of operating in that gap.

Gambling or instrument

The closing panel circled back to whether all of this is gambling or a financial instrument, and the speakers landed, sensibly, on both, depending on the contract and the user. That is the right answer, but it leaves the evening’s grander claim intact and unexamined, and the claim is where I part company with the room. A prediction market is a genuinely useful object. It discovers prices on questions that markets never used to touch, and it lets people and firms hedge risks that had no instrument before. It is closer to Kyle’s insider and Tetlock’s superforecaster than to Galton’s ox: a device that concentrates the judgment of a calibrated few and pays for it with the noise of the many. That is not crowd wisdom, whatever the founders call it to stay clear of the gambling label.

Afterward, off the stage, I talked with Jean Chuilon-Croll, who is building Flair, a prediction platform with no money at stake. It tries to surface reliable forecasters the way Tetlock did, on track record alone, and then to put that to use in two ways, as a believability-weighted tool a company could lean on for its own decisions, and as a personal read that shows someone where their own judgement is strong. What Flair keeps is the part of a prediction market that does the real work, finding the few who are right and weighting them, and what it drops is the wagering. He is plainly aiming to launch just before the 2027 election.

The same conversation left me more worried about the French economy than anything the markets were pricing. He was not blaming any one government. Many of the hard decisions are well understood by the people who would have to make them, and the obstacle is not ignorance but the political cost of the reaction, which turns the necessary decision into the one no one who has to face an electorate can afford to make. Coming from someone who has advised on economic policy, it read less as cynicism than as a description of an incentive trap, and it is a thread for another entry.

Walking out past the mosaics, the old logic of the place stayed with me. The Manoir de Paris once sold manufactured fear as entertainment, a frisson over uncertainty that was never real. For one evening the same room had filled with people trying to price the kind that is. And the people who price it well are not a wise crowd but a few who know something, paid by everyone who does not.

---

Conceptual Toolbox

Concept	What it means	Why it matters
Prediction market	A venue trading contracts that pay out on an event, whose price reads as a probability	The object the whole evening was about
Wisdom of the crowd	Independent errors cancel under averaging (Galton; Surowiecki)	The slogan the founders use, and the wrong description
The use of knowledge in society	A price compresses dispersed, tacit knowledge into one number (Hayek, 1945)	The real intuition behind markets as information
Kyle model	An insider’s private signal enters the price by trading against noise (Kyle, 1985)	The actual mechanism, not crowd wisdom
Noise	Noise traders supply the liquidity informed traders profit from (Black, 1986)	The crowd’s role is liquidity, not wisdom
Superforecasting	A calibrated minority beats the crowd and the experts (Tetlock, 2015)	Concentrate the few, do not average the many
Favourite-longshot bias	Longshots overpriced, favourites underpriced, persistently	The signal is distorted at the tails
Knightian uncertainty	Measurable risk versus unmeasurable uncertainty (Knight, 1921)	A price on a one-off event is not a frequency
Martingale critique	A real probability far from resolution cannot sit far from even odds (Taleb, 2018)	Confident long-horizon prices may be arbitrage-incoherent
Futarchy	Government by prediction market: vote on values, bet on beliefs (Hanson)	Tolerates insiders that crowd wisdom cannot

---

Prediction Meeting Paris was held at Albert School, 18 rue de Paradis, on 29 June 2026, organised by Mathis Chau and co-hosts, with talks from Pierre Tramon (FDJ United Ventures), Armand Droué (Kalshi), and Thomas (Hyperliquid France), followed by a panel.

Further readings behind this entry:

On what a crowd actually is:

• Francis Galton, Vox Populi, Nature (1907).
• James Surowiecki, The Wisdom of Crowds (Doubleday, 2004).
• Michael J. Mauboussin, Who Is on the Other Side? (Morgan Stanley, 2019), on the conditions for and against market efficiency.

On prices as information, and the microstructure beneath them:

• Friedrich Hayek, The Use of Knowledge in Society, American Economic Review (1945).
• Albert S. Kyle, Continuous Auctions and Insider Trading, Econometrica (1985).
• Fischer Black, Noise, The Journal of Finance (1986).
• Justin Wolfers and Eric Zitzewitz, Prediction Markets, Journal of Economic Perspectives (2004).

On forecasting and the calibrated few:

• Philip Tetlock and Dan Gardner, Superforecasting: The Art and Science of Prediction (Crown, 2015).
• Philip Tetlock, Expert Political Judgment (Princeton, 2005).
• Rob Copeland, The Fund (St Martin’s Press, 2023).

On the limits of treating a price as a probability:

• Erik Snowberg and Justin Wolfers, Explaining the Favorite-Longshot Bias (NBER, 2010).
• Nassim Nicholas Taleb, Election predictions as martingales: an arbitrage approach, Quantitative Finance (2018).
• Frank Knight, Risk, Uncertainty and Profit (1921), on the risk-uncertainty distinction.
• Bill Janeway, Doing Capitalism in the Innovation Economy, on Knightian uncertainty and the funding of innovation.

On the regulatory and ethical edge:

• ANJ, Plateformes de marchés de prédiction: des sites illégaux en France (2026).
• Pauline Pailler, Quelle régulation pour Polymarket?, Le Club des Juristes (2026).
• Robin Hanson on futarchy.

On what a model cannot hold, and a related thread:

• My earlier entries on the probabilistic mind and on situations versus problems.

Footnotes

1. [Disclosure] I write as a general partner investing in technology companies, and I hold no financial position in the prediction-market platforms or startups named here. ↩

2. [Source] Kalshi announced a $1bn Series F at a $22bn valuation in May 2026, led by Coatue with Sequoia, Andreessen Horowitz, Morgan Stanley, and ARK, doubling its $11bn Series E from five months earlier. TechCrunch; Kalshi. ↩

3. [Source] Kalshi in talks at roughly $40bn (Financial Times, 24 June 2026); Polymarket raising at about $15bn; the Intercontinental Exchange, the New York Stock Exchange’s owner, committed up to $2bn to Polymarket. Financial Times; The Block. ↩

4. [Context] JPMorgan’s chief executive said in April 2026 the bank was weighing entry into prediction markets while ruling out an in-house desk; in May it told roughly 320,000 staff to be cautious about insider trading on Kalshi and Polymarket. Morgan Stanley joined Kalshi’s May round; Goldman Sachs met both platforms. CoinDesk; The Street. ↩

5. [Context] In December 2025 Kalshi signed deals with CNN and CNBC; the CNBC agreement makes it the network’s sole prediction-market data provider, with odds shown during Squawk Box and Fast Money. Critics call the practice “casino journalism” and warn of a feedback loop between bets, coverage, and perception. CNBC; Financemagnates; The Intercept. ↩

6. [Caveat] On 2024 election eve Polymarket had Trump near 58% to Harris’s 42% while polling averages sat near 50-50. The rally was partly driven by a handful of large accounts, one collecting about $85m, a point the later argument revisits. CNN; Fortune. ↩

7. [Source] The ANJ obtained Polymarket’s geoblock from France on 29 November 2024, after $3.6bn of US-election volume, classifying the offer as unauthorised gambling under art. L.320-1 of the Code de la sécurité intérieure, not on crypto grounds. ANJ. ↩

8. [Source] More than a dozen US states have sued Kalshi and Polymarket for alleged unlicensed sports betting; Kentucky’s attorney general put sports at about 89% of Kalshi’s volume. The CFTC-versus-states jurisdiction question is expected to reach the Supreme Court. Cryptopolitan. ↩

9. [Context] The ACPR, the French prudential authority under the Banque de France, licenses payment and electronic-money institutions and supervises their know-your-customer and anti-money-laundering obligations; suspicious transactions are reported to TRACFIN, the financial-intelligence unit. ACPR. ↩

10. [Source] Francis Galton, Vox Populi, Nature (1907). 787 valid estimates at a Plymouth fair; the median guess fell within roughly 1% of the ox’s dressed weight. ↩

11. [Etymology] Robin Hanson coined “futarchy” in 2000, from “future” and the Greek arkhē (rule), as in monarchy and oligarchy. The motto is “vote on values, bet on beliefs”. One speaker glossed it on the night as “future anarchy”, which reverses its meaning. ↩

12. [Definition] Albert S. Kyle, Continuous Auctions and Insider Trading, Econometrica (1985). An informed trader camouflages orders among noise traders; the market maker, unable to separate them, moves the price as informed flow accumulates. ↩

13. [Source] Fischer Black, Noise, presidential address to the American Finance Association, The Journal of Finance (1986). Noise traders make markets liquid enough for informed trading to be profitable. ↩

14. [Context] The Good Judgment Project won IARPA’s forecasting tournament (2011 to 2015). Tetlock reports that superforecasters, the top performers, beat the aggregate and reportedly outperformed intelligence analysts with classified access, working only from open sources. ↩

15. [Context] Ray Dalio’s formula was “idea meritocracy = radical truth + radical transparency + believability-weighted decision-making”, with believability scored through Bridgewater’s Dot Collector app. Rob Copeland, The Fund (2023); Bridgewater and Dalio reject the book’s account as inaccurate. ↩

16. [Source] Erik Snowberg and Justin Wolfers, Explaining the Favorite-Longshot Bias (NBER, 2010). Low-probability outcomes are overpriced and favourites underpriced across betting and prediction markets. ↩

17. [Counter] Nassim Taleb, Election predictions as martingales: an arbitrage approach, Quantitative Finance (2018), argues a binary forecast that qualifies as a probability is a martingale, so it cannot sit far from even odds long before resolution without violating no-arbitrage. ↩

18. [Source] In November 2025 a Polymarket contract on a Russian advance toward Myrnohrad resolved after the underlying Institute for the Study of War map was altered to show a fictitious gain, then reverted; ISW confirmed unauthorised manipulation. Tirage-Gagnant; ANJ. ↩

19. [Context] Hyperliquid Labs is incorporated in Singapore, where the team relocated in 2024. With roughly 11 contributors and about $1bn in annualised revenue, it has the highest revenue per employee of any company, ahead of Tether, Nvidia, and Apple, and has taken no venture capital. CryptoSlate; Colossus. ↩

20. [Context] Thomas leads Hyperliquid France, the community of builders and users on Hyperliquid. Its HIP-4 outcome markets, live under two months, settle on-chain for recurring contracts, avoiding an external oracle, but showed wide spreads at launch. ↩

21. [Counter] Pauline Pailler (Université Paris Cité), in Le Club des Juristes (2026), notes that French event contracts resemble financial contracts economically but are not classified as such, escaping both gambling law and the market-abuse regime. ↩