Betting on the Blockchain: The Tax Treatment of Prediction Markets

Partner Ben Lee and Crypto Tax & Accounting Technical Director Dion Seymour discuss the evolving tax considerations around prediction markets, in Taxation.

Ben and Dion’s article was published in Taxation, 13 July 2026.

The FIFA World Cup has arrived, and with it a surge of activity on a class of platform that most UK tax advisers will not yet have encountered in practice. Prediction markets; decentralised platforms where users take positions on the outcomes of real-world events, have grown rapidly over the last two years. Polymarket, the largest, settled over $8 billion in volume during the 2024 US election cycle alone. The World Cup will almost certainly push participation to new heights. For the first time, a meaningful proportion of that activity will involve cryptoassets rather than sterling.

Clients will ask their advisers what this means for their tax position. The honest answer is that the question is harder than it looks, and the analysis sits at the intersection of three areas: gambling, cryptoasset taxation, and stablecoin accounting, that do not always speak to each other clearly. This article sets out a framework for approaching it.

What are prediction markets?

Prediction markets let users buy and sell positions on the outcome of real-world events. On Polymarket, each market poses a binary question: which team will win a match at the World Cup, will a given economic indicator hit a threshold, and users purchase yes or no “shares” whose price fluctuates between zero and $1 based on perceived probability. If the market resolves in a user’s favour, the share pays out at $1. The losing side becomes worthless.

Users can enter and exit positions before settlement, locking in wins or cutting losses by trading dynamically. The platform facilitates price discovery between participants rather than setting odds itself; it is not the house. In this respect, prediction markets function far more like a betting exchange, Betfair is the closest domestic analogue, than a traditional bookmaker.

Polymarket operates on Polygon (a layer 2 scaling solution for Ethereum). Positions are recorded using ERC-1155 tokens, smart contract-based outcome tokens that are unique to a specific event and resolution. Crucially, these are not exchange tokens in the sense that HMRC’s cryptoasset guidance uses that term. They are semi-fungible; they can enable shares in a single position (bet) to be owned by many. Therefore, they are not fungible in the way bitcoin or ether is. They are, in substance, digitally recorded rights to a particular outcome.
As HMRC guidance has limited recognition with NFTs and no mention of semi-fungible tokens, this is new territory.

The entry point for most UK users is USDC, a US dollar-pegged stablecoin. They convert sterling into USDC, fund their Polymarket wallet, and from there enter positions. Winnings are received in USDC and, at some point, converted back. That pipeline, sterling to USDC to outcome token and back, is where the tax complexity begins.

The gambling question

Many advisers will be familiar with HMRC’s cryptoasset manual at CRYPTO10450, which states that HMRC does not consider the buying and selling of cryptoassets to be gambling. A number of clients and, sometimes, their advisers take this to mean that participation on Polymarket must therefore be taxable. That reading misunderstands what the guidance states.

CRYPTO10450 was inserted to correct a misconception that originated from previous guidance: that the mere holding and disposal of cryptoassets such as bitcoin could constitute gambling and is therefore exempt. Gambling is a deliberate activity, and HMRC’s point is that the asset is not a bet; therefore, gains from it are within the Capital Gains Tax regime. The guidance explicitly acknowledges that whether any given activity constitutes gambling is a question of fact, there are no badges of gambling, it is either gambling or it is not. It does not close the door on prediction market activity being characterised as betting.

HMRC will consider the position from the Gambling Commission, where there is uncertainty. In February 2026, it published a blog on prediction markets concluding that their core aspects are akin to what in the UK would be described as a betting exchange, and that a prediction market operator launching in Great Britain would not be able to classify itself as a non-gambling product. That view supports the argument that participation on platforms such as Polymarket constitutes betting for UK purposes, and therefore falls within the exemption at section 51(1) of the Taxation of Chargeable Gains Act 1992.

There is also the question of what the outcome tokens represent. HMRC’s own guidance at CRYPTO10100 acknowledges that different tokens serve different functions. The outcome tokens issued on Polymarket do not have value independent of the event they reference; they are, in substance, a digitally recorded right to a bet outcome. This is meaningfully different from holding bitcoin, which has independent economic value. The structure of the token, not the platform it sits on, should be the primary factor in characterisation.

When the analysis shifts

Not every prediction market token is the same, and the structure of the position matters. The gambling analysis is most robust where the token has no economic value beyond its link to a specific outcome where it is, in substance, a wrapped bet. The more a token begins to resemble a financial instrument, the more the analysis shifts.

Some commentators have suggested that Polymarket positions are better characterised as contracts for difference. We do not think that holds up. CFDs are bilateral contracts between a trader and a provider, regulated by the FCA, tied to an identifiable underlying financial asset. Polymarket positions are peer-to-peer, created on a decentralised smart contract platform, outside the FCA perimeter, and settled against an event outcome rather than a price difference. The payoff mechanism is binary, not differential.

What this means for advisers is that the characterisation question requires looking beneath the platform’s labelling to the substance of what has been acquired. A position on whether England will win the World Cup, tradable on a blockchain, is structurally a bet, regardless of what the token is called. A position on whether the S&P 500 will close above a given level might be closer to a binary option, and a different analysis would be required.

The same logic applies across the broader prediction market sector, which has expanded well beyond Polymarket to include a range of platforms covering sports, finance, and macro events. The starting question is always: what does this token actually represent, and does that representation align more closely with a bet or with a financial instrument?

The hidden tax events

Even where the gambling analysis is robust and winnings from Polymarket are non-taxable, there is a separate set of tax events that many clients will be generating entirely unintentionally: disposals of stablecoins.

Most Polymarket users fund their accounts in USDC and receive their winnings in USDC. They treat it, functionally, as cash. For UK tax purposes, it is not. USDC is a cryptoasset. Each acquisition and disposal of USDC is potentially a chargeable event, giving rise to a capital gain or allowable loss. Where sterling is converted into USDC to fund an account, that acquisition has a base cost. When USDC is used to enter a position, there is a disposal of the USDC, at whatever sterling value prevails at the time, against that base cost. When USDC is received as a winning payout and subsequently converted back into sterling, there is a further disposal.

For a user who is active on Polymarket through the World Cup, funding the account, trading positions, receiving payouts and rolling the USDC balance into new positions, the number of chargeable events can be very large. The gain or loss on each stablecoin disposal will typically be small (USDC is, by design, stable relative to the dollar), but sterling/dollar movements mean that the sterling equivalent gain or loss is not zero. Clients who assume they have no tax exposure because they have been gambling may be surprised to discover a capital gains computation lurking in their wallet history.

This is not a problem unique to prediction markets. It arises any time stablecoins are used as a medium of exchange rather than held as an investment. Prediction markets are simply a context in which it is especially likely to catch clients out, because the gambling framing makes them assume the whole activity is outside the tax system.

If not gambling: the trading and miscellaneous income analysis

Where the gambling characterisation does not hold, either because HMRC successfully challenges it or because the facts of a particular client’s activity take it outside the exemption, the analysis moves to trading or capital gains, with income tax taking priority.

The starting point is the well-established principle from Graham v Green [1925] 9 TC 309: betting and gambling do not, of themselves, constitute a trade. Rowlatt J’s formulation remains relevant; a bet is merely an irrational agreement that one person should pay another on the happening of an event. The fact that a client is systematic, skilled, and profitable does not change the analysis. In McMillan v HMRC [2020] UKFTT 0082, the First-tier Tribunal found that sophisticated, research-driven sports betting, well beyond the average gambler’s skill, was not a trade, and HMRC was unable to identify what the trade actually was.

If HMRC were to argue that a Polymarket participant is trading, the question it must answer is: what is the trade? The obvious candidate would be acting as a bookmaker. But on any sensible analysis, a Polymarket user cannot be a bookmaker in this case, as it is Polymarket that creates the markets and sets the odds unilaterally. There are no customers, the user is a punter, however sophisticated.

A high-frequency participant who actively trades positions, buying and selling before settlement rather than simply holding to resolution, may attract more scrutiny on the badges of trade. The number of transactions, the short holding periods, and the profit-seeking motive are all present. But as Manzur v HMRC [2010] illustrates, those factors alone are not determinative. The absence of customers, external financing, and commercial structure all point away from trade.

For users who cannot rely on the gambling exemption, perhaps because the activity is structured in a way that takes it outside section 51(1), or because corporate entities are involved, the default analysis is likely to be capital gains on the disposal of the outcome tokens, with USDC stablecoin disposals also giving rise to separate computations as described above.

There is a further wrinkle for clients who receive referral bonuses from platforms. Polymarket and similar platforms offer referral programmes that reward users for bringing in new participants. These are not gambling winnings. Whether they are taxable depends on whether they constitute income arising from a service. Our view is that, in most cases, passively sharing a referral link does not amount to a service performed for reward, there is no contract, no obligation, no commercial activity, and the receipts are better characterised as promotional incentives rather than taxable income. The facts matter, and a client who has structured referral activity more commercially will sit in a different position.

A practical framework

When a client presents with Polymarket or prediction-market activity, the adviser’s initial questions should be: which events did they bet on, and which tokens did they use? The gambling analysis is strongest for users betting on sports and other real-world events using outcome tokens that have no value independent of the resolution. HMRC’s apparent current position, that all of this is just a series of crypto-to-crypto disposals, does not adequately engage with the nature of what outcome tokens represent, and it is a position worth challenging.
Even for clients who have a strong gambling argument, the stablecoin layer requires separate attention. The fact that winnings are not taxable does not mean the USDC used to earn them is irrelevant. Advisers should obtain wallet history, identify every USDC acquisition and disposal, and compute the sterling-equivalent capital position for each.

The sector will only grow. As prediction markets expand into new events, new assets, and new structures, the tax questions will grow with them. The World Cup is a convenient hook for raising this with clients now, before positions are opened and records are lost. For those with existing activity, the first step is always the same: get the data, understand the structure, and only then proceed to analysis.

---