HMRC sent nearly 65,000 cryptoasset nudge letters in 2024-25, against 27,700 the year before. That is one of the best-documented crypto numbers in the country, and it is not a count of how many people hold any. 

What does the country actually count, and why does the count move? 

Things that go wrong, mostly, and things that generate paperwork. HMRC’s cryptoasset nudge letters went from 27,700 in 2023-24 to nearly 65,000 in 2024-25. Action Fraud recorded 34,673 UK investment fraud reports in 2025, up 35% on the previous year, carrying GBP 879.8m of reported losses, roughly GBP 2.4m a day, with cryptocurrency involved in 66% of them. 

Every one of those is a count of an event, produced by an institution with a statutory reason to produce it. None of them counts holders. That is not an oversight by anybody in particular. It is what happens when the only bodies with a duty to publish are the ones handling the consequences. 

Because the letters are a function of what HMRC can see. A nudge letter goes out when a taxpayer’s record can be matched against data an exchange has handed over, so the volume moves with the reach of the data-sharing arrangements and the size of the campaign HMRC chose to run that year. A department that obtains a new data feed can double its letter count without a single extra person buying anything. 

The same caution applies in the other direction. A flat year would not establish that holding had stopped growing, only that the matching exercise had not expanded. The figure is genuinely useful, and what it measures is HMRC’s compliance activity. 

Four UK crypto figures, all real, all sourced, none of them a participation measure. HMRC and Action Fraud, 2024-25 and 2025 respectively. 

Does the fraud data fill the gap? 

It cannot, and the reason is worth being precise about. A report to Action Fraud is a record of somebody deciding to report. Reporting rates respond to publicity campaigns, to how easy the form is, and to whether victims expect anything to come of it, so a 35% rise in reports is consistent with a rise in fraud, a rise in reporting, or both in unknown proportion. The 66% crypto share is more robust, because it describes the composition of a set rather than its size, and composition survives changes in reporting volume better than totals do. It still says nothing about how many Britons hold cryptoassets. It says what the frauds being reported are made of. 

The vacuum gets filled by whoever has an incentive to fill it 

With no obligatory participation count, the space is occupied by exchange user numbers, app download figures and survey estimates produced by organisations with a position. Those are not worthless, and they are all produced voluntarily by people who chose both the question and the moment to publish the answer. 

Voluntary numbers arrive when they are flattering and go quiet when they are not. 

This article is not exempt from that. Setting four enforcement figures next to each other and saying they imply nothing about ownership is itself a way of making enforcement data carry an argument about ownership, which is close to the thing it complains about. A research site pointing at a measurement gap also benefits from the gap being interesting. Both of those are worth holding in mind while reading the paragraphs above. 

What is the regulatory backdrop? 

Restrictive and stable. Crypto CFDs have been off-limits to UK retail clients since 6 January 2021, which closes the leveraged route into the asset that remains open in some other jurisdictions. Spot holding remains available, and the anti-money-laundering registration regime is a different thing from investment authorisation, which continues to catch people out. 

That distinction matters more than most consumers realise. Registered with the FCA and authorised by the FCA are different statuses conferring different protections, and firms are not always eager to be precise about which applies to them. It also explains part of the counting problem: a registration regime built to check money laundering controls was never designed to produce a census. 

What would an honest participation figure need? 

A stated population, a stated definition of holding, a collection date and a sampling method, published together rather than assembled afterwards by whoever quotes the number. Holdings data supplied directly by exchanges would sidestep the self-reporting problem, and no such series is public. A breakdown by value held would separate many small holders from a few large ones, which is the difference between a mass behaviour and a concentrated one. 

Until something like that exists, the workable approach is to keep each figure next to its definition and its date and refuse to average across them. That is the narrow job The Investors Centre’s research does properly, holding each of its UK trading statistics against the HMRC, FCA or Ombudsman Service release it came from. 

Narrow is the operative word: a compilation saves you the retrieval and declines to average four measures of different things into one, and neither of those produces the participation count everything above has been asking for. 

What should you do with a crypto statistic you meet in the wild? 

Ask who was obliged to produce it. Figures generated under a legal duty, tax records, regulatory returns, audited accounts, exist whether or not anybody wanted them that year, which makes them dull and hard to manipulate. Figures produced voluntarily exist because somebody decided to produce them, and the decision is part of the data. 

Then ask what would have to change for the number to move, other than the thing it appears to measure. If you can name two such factors quickly, as you can with the nudge letters and the fraud reports, the figure is a fine piece of evidence about the institution that published it and a poor one about the public.