While American regulators, legislators and prediction market operators have spent the past eighteen months fighting over what prediction markets are, the UK Gambling Commission stated its position in February.
Brad Enright, the Commission’s Director of Strategy, published the UKGC’s view without ambiguity. Prediction markets are gambling products in Great Britain. Any commercial product meeting the legal definition of gambling under UK legislation must be licensed and regulated by the Commission. The only established exception is spread betting, which falls under the Financial Conduct Authority. Prediction market operators who have argued, with considerable legal and lobbying resource behind them, that their products are financial derivatives sitting outside gambling regulation have no version of that argument readily available in the UK. Operating without the appropriate licence is a criminal offence.
Whether that position holds is a separate question. Polymarket and Kalshi have not accepted regulatory classification they disagreed with in Nevada, Massachusetts, New Jersey, or anywhere else. Their operating model is built around jurisdictional challenge. Given that track record, a legal challenge to the UKGC’s position in a UK court is probable rather than merely possible. The Gambling Act 2005 defines betting on different statutory grounds to the US Commodity Exchange Act, and no UK court has yet ruled on whether a prediction market contract constitutes a bet under English law. That ruling is coming.
The governance question matters regardless of how it resolves. An operator that builds robust integrity infrastructure while the classification is settled is better placed if it holds. An operator that builds it proactively is also better placed if the classification gets challenged, because a demonstrated governance record is the strongest available evidence that the product can be operated responsibly. Either way, the governance work needs to happen. The debate over classification is a reason to accelerate it, not to defer it.
So the harder question, which has barely been asked, is what responsible prediction market operation in the UK actually looks like.
What the UKGC Position Actually Requires
A prediction market operator holding a UKGC betting intermediary licence is subject to the full weight of the Licence Conditions and Codes of Practice. That includes, among other things, the integrity obligations under LCCP 15.1.2.
The requirement is precise. Licensees must as soon as reasonably practicable provide the Commission with any information from whatever source that they know relates to or suspect may relate to the commission of an offence under the Act. Licensees who facilitate the making or acceptance of bets between others on sporting events governed by sport governing bodies listed in Schedule 6 of the Gambling Act must report to those governing bodies directly. That information must be submitted to the Sports Betting Intelligence Unit using a standardised template developed in consultation with the betting industry.
This is a mature and specific reporting architecture built over two decades of regulated sports betting operation. The standardised SBIU template exists because the Commission determined that informal reporting was insufficient. It reflects a settled understanding of what actionable integrity intelligence looks like, what data fields it requires, and how it needs to be structured to support investigations.
It maps onto fixed odds and exchange betting on established sporting events. It does not map onto prediction market contract data, binary yes/no outcomes on a much broader range of events, traded by a participant population with very different characteristics to the regulated sports betting customer base. That gap is structural, and it is where the governance work actually begins.
The Exchange Problem
Prediction markets are exchange markets. That matters more than most commentary on them has acknowledged.
In a fixed odds sportsbook the operator is the counterparty to every bet. The customer wins or loses against the house. The operator’s trading desk observes every position because every position affects the operator’s book. Suspicious behaviour is visible by default because it creates commercial exposure.
On an exchange the operator is the infrastructure. Customers trade against each other. The operator facilitates every transaction but is financially neutral on the outcome. That neutrality is the exchange’s commercial advantage, with no book risk and theoretically tighter prices, but it creates a fundamental monitoring problem. The operator has no natural commercial incentive to scrutinise individual positions because no individual position affects the operator’s P&L.
The lay facility compounds this materially. An exchange customer can sell an outcome as well as buy it, acting as the bookmaker and taking the other side of the market. This means the full range of manipulation strategies available in financial markets is available on a betting exchange. Wash trading, coordinated position building before an information event, account clustering to obscure beneficial ownership: these are documented patterns in thin markets with motivated participants.
It requires customer classification infrastructure built from trading knowledge, which means understanding who the sharp money is, what normal behaviour looks like for each account segment, and when a pattern deviates from classification in a direction that demands investigation. That classification work is invisible in most of the regulatory debate around prediction markets. It is absent from the CFTC guidance. It does not feature in the academic literature on prediction market accuracy. It is the operational foundation on which meaningful integrity monitoring is built.
Critically, that classification work is a byproduct of trading knowledge. It is built from years of observing how markets actually behave, understanding what sophisticated participants look like in practice, and developing the pattern recognition that distinguishes legitimate sharp activity from something that needs a closer look. Automated surveillance tools can surface flags. Sorting signal from noise requires human intelligence that regulation cannot mandate and technology cannot replicate.
The FIFA Case Study
On 2 April 2026 FIFA announced a landmark multi-year partnership with ADI Predictstreet, naming the startup its first official prediction market partner for the 2026 World Cup.
ADI Predictstreet received its Gibraltar Betting Intermediary licence on 26 March 2026, seven days before the announcement. The company had no publicly accessible platform at the time the partnership was announced.
Within days of the announcement, one of the company’s senior executives became the subject of allegations from India’s Securities and Exchange Board relating to insider trading connected to the Adani Group, with trades worth over $900,000 allegedly executed on advance knowledge of a $2 billion investment. These are allegations at this stage. The legal distinction matters. The governance question it raises does not disappear because of it.
FIFA, which Louis Weston’s 2025/26 LawInSport annual integrity review identified as facing unprecedented integrity risks at the 2026 World Cup, selected a prediction market partner that was three weeks old, had no live product, and is now answering questions about the regulatory history of its senior leadership.
The partnership announcement stated that the platform would incorporate real-time monitoring of suspicious trading activity and structured information sharing and reporting systems. These are the right words. Whether the infrastructure to deliver them exists in an organisation three weeks old is a question that a functioning governance framework answers before a partnership is signed.
What Responsible Operation Actually Requires
The UKGC’s position on prediction markets creates a window in which a licensed operator can define what responsible governance looks like before detailed regulatory standards are specified. The Commission has classified the products. It has not yet set out the operational standards it expects beyond the existing LCCP framework. The operator that builds the governance architecture first is creating the reference standard against which every subsequent entrant will be measured.
That requires four things that most current operators have not built.
Customer classification infrastructure grounded in trading knowledge. The SBIU reporting obligation requires operators to identify suspicious activity before reporting it. Identification requires knowing what normal looks like. Normal is defined by customer classification, which means understanding the participant population well enough to recognise when behaviour deviates from it in a direction that suggests manipulation. In a B2B model, where an exchange engine sits beneath multiple operator frontends, this classification work needs to happen at the exchange level. The data flows through a single infrastructure. The monitoring belongs there.
Automated behavioural surveillance calibrated to binary markets. Traditional integrity monitoring was built for continuous price markets where suspicious movement has a time dimension. Binary prediction market contracts resolve in a single step. Cluster analysis of correlated positions, timing correlations between position changes and external information events, network analysis of connected accounts: these are the tools, and they need to be built into the exchange infrastructure from the outset.
A human intelligence layer above the automated system. Automated surveillance generates flags. Determining whether a flag represents genuine manipulation, legitimate sharp activity, or statistical noise from a thin market requires operational market knowledge. This is the accumulated expertise of people who have run trading books and managed integrity monitoring in practice. It is where technology and regulation both run out of answers.
Independent board-level governance with UKGC-credible oversight. The 2020 UKGC sanction of Triplebet, operator of Matchbook, was partly the product of inadequate governance architecture around exactly these monitoring and classification obligations. A prediction market operator seeking to demonstrate that its governance framework is genuinely independent needs board-level oversight from someone whose specific expertise covers the integrity monitoring obligations the business is under.
The Operator That Builds This First Defines the Category
Kalshi and Polymarket have invested heavily in arguing that their products sit outside gambling regulation. That argument has no settled pathway in the UK, and may face a direct legal test here before long given both operators’ history of contesting classification through litigation.
The operators that waited for the UKGC to specify what responsible gambling looked like in the early 2000s found themselves responding to enforcement actions. The operators that led, building monitoring infrastructure, engaging with the Sports Betting Intelligence Unit, participating in the frameworks that became the LCCP, are the operators that defined what responsible operation looks like in the UK today.
The operator that builds the framework first, whether the UKGC classification holds or faces challenge, is the one best positioned to shape the outcome. That is not a compliance argument. It is a commercial one.