UK Gambling Black Market: What the New Banking Data Actually Shows

The argument over Britain’s gambling black market has acquired new evidence. On 28 October 2026, the Gambling Commission published an analysis connecting observed banking transactions with survey responses. Industry coverage followed on 30 October. The useful question is what the data adds to the debate—and where its boundaries lie.

The new official research release reports a strong within-sample association between detected illegal-market transactions and elevated gambling-harm scores. It also describes continuing activity with licensed operators. These findings concern people and transactions identified in the dataset, not a new national market-share estimate.

The figures that need their denominators

• 3,426 detected transactions were associated with 179 individuals. These are identified transactions, not a count of all illegal gambling.
• 45% versus 6%: the proportion with a Problem Gambling Severity Index score of eight or more, comparing the detected illegal-transaction group with licensed non-lottery gambling consumers in the same dataset.
• 91% overlap: among people with repeated detected illegal transactions, most also transacted with licensed operators. This percentage should not be described as applying indiscriminately to every illegal-market user.

For the PGSI comparison, the results section specifies 146 people in the detected unlicensed-transaction group and 12,543 in the licensed group. The groups are not mutually exclusive. The repeated-transaction chart uses 122 individuals; its base is different again.

New publication, longer observation period

The research concerns Great Britain; this article does not extend its findings to Northern Ireland. The underlying YouGov banking panel covers July 2018 to July 2026 and contains approximately 38,000 individuals. Data was shared with informed consent, anonymised and linked to other panel information, including PGSI scores where available. The Commission worked with Warwick Business School on the analysis. These materials and methods matter: a late-October publication does not mean the observed transactions all happened in October.

Why this is a longstanding debate

In November 2025 research, the Commission explained why estimating illegal gambling expenditure was difficult and said it was not ready to publish a credible overall estimate. Hidden activity, incomplete data and uncertain assumptions were central obstacles. The latest banking study adds a different kind of evidence, but its authors still decline to turn it into a market-size figure.

This distinction separates three questions that are often collapsed into one: Does illegal gambling occur? Who is affected? How large is the total market? Evidence that helps answer the first two does not automatically answer the third. A dramatic percentage can describe a selected group accurately while saying little about the proportion of all gambling expenditure.

Overlap deserves attention alongside displacement

Our analysis: a debate framed entirely around customers leaving licensed gambling can miss people who interact with both sectors. A regulated-market relationship is not, by itself, evidence that a person has no exposure to unlicensed activity. The research is a reason to investigate connected behaviours and protection gaps, rather than assign every transaction to a simple story of migration.

A second October report examines motivations

The Commission’s 20 October Consumer Voice report, conducted by Yonder, used digital intelligence, a survey and qualitative interviews to examine motivations. It identifies varied audiences and highlights access barriers, requests for personal information and attitudes to authority. It also reports that people can encounter frustrations on illegal sites without equivalent protections. This report deliberately examines a subset of people who gamble; it is not a description of everyone who gambles.

The two projects answer different questions. Banking records help document observed transactions; attitudinal research explores experiences and motivations. Their populations and measurements differ. Combining them into one headline percentage would create a statistic neither project produced.

Four editorial tests for any black-market headline

1. Population: is the claim about British adults, people who gamble, a research panel or a selected subgroup?
2. Measure: does the number count people, visits, deposits, transactions, losses or revenue?
3. Period: is the date the release date or the period observed?
4. Inference: does the research show an association, a change over time or a causal effect?

Our interpretation: these tests are more useful than treating every new percentage as a verdict on regulation. They make competing claims comparable and expose cases where a headline changes the underlying question. Industry reactions, including the 30 October coverage, are part of the debate; the study’s own definitions remain the reference for its numbers.

What the study cannot establish

The Commission’s limitations section says the sample is not nationally representative. It includes volunteers who shared banking data, with demographic imbalances. Identification covers only a subset of illegal operators, and cryptocurrency activity is outside the captured banking transactions. Some relevant payments may therefore remain unidentified.

The authors explicitly rule out using the results to estimate the illegal market’s absolute size or its share relative to licensed gambling. The design also cannot determine whether illegal-market spending causes harm. An elevated PGSI score in linked survey data is not proof that a particular website caused it.

Why correlation is not a policy verdict

Our analysis: the same observed association can be compatible with several explanations. People already experiencing harm might seek different services; harmful experiences might span multiple settings; other factors might affect both. Distinguishing these explanations requires additional evidence about timing and alternative causes. A cross-group comparison cannot choose between them.

For the same reason, this study does not isolate the effect of a particular tax change, identity check or gambling restriction. A claim that one policy necessarily caused the observed behaviour would require a suitable causal design. Neither the article nor its headline should silently supply that missing evidence.

What readers can verify independently

The Gambling Commission’s business register can be searched by business, trading name, domain or account number. Check current licence status and the relevant activities; the register retains some historical records, so a search result alone is not proof of a current permission. This is a verification resource, not an endorsement or an assurance that gambling is risk-free.

Questions the next research should answer

• How much activity is missed by the current identification method, and does missingness vary between groups?
• How stable are the observed patterns when the sample or observation period changes?
• Can future evidence establish the sequence between gambling harm and illegal-market activity?
• Which protection measures reduce harm across people’s combined gambling activity?

These are questions raised by our reading, not findings claimed by the Commission. They offer a clearer research agenda than declaring the debate settled. A useful update would disclose what changed in the evidence, rather than reuse an old percentage under a new date.

Published and source-checked 31 October 2026. This is independent analysis of published research, not a new survey or an operator recommendation. The principal study was released on 28 October 2026; its observation period ends in July 2026.

• Gambling Commission — 28 October research release
• Gambling Commission — Results and subgroup denominators
• Gambling Commission — Materials and methods
• Gambling Commission — Limitations
• Gambling Commission — 20 October Consumer Voice findings
• Gambling Commission — November 2025 market-size research context
• Gambling Commission — Business register and data definitions
• SBC News — Industry reaction, 30 October 2026 (secondary context)

Primary research is the source for the numerical findings and their qualifications. Industry reporting is linked for reaction and context. Editorial interpretations are labelled in the article. No national prevalence estimate, causal risk ratio or new survey has been calculated.