When a UK-licensed operator applies a stake cap between midnight and 04:00, its offshore mirror sites pick up roughly 19% of the displaced handle within the following fortnight. That figure comes from affiliate tracking data covering eleven white-label brands across the 2024–25 football season, and it holds reasonably steady whether the cap is set at £2, £5, or a percentage-of-deposit rule. The mechanism is not subtle: players who hit a friction point at 3am rarely stop gambling, they simply change the domain in their address bar.

What the 4am cap actually is

Nobody in the UK licensed market calls it a "4am cap" officially. The term has grown up around two overlapping practices. The first is voluntary: a handful of operators, mostly those with a retail or heritage brand behind them, apply lower maximum stakes to casino and slots products during overnight hours, typically £2 to £5 per spin, on the reasoning that late-night play correlates with chasing losses and impaired judgement. The second is regulatory pressure dressed as product design. Since the Gambling Commission's 2023 consultation on financial risk checks and the subsequent 2024 pilot, several operators have introduced time-of-day stake limits to demonstrate duty-of-care credentials ahead of any formal requirement.

The specifics matter. A £2 cap on a 96.2% RTP slot at 02:00 is not a meaningful deterrent to someone staking £1.50 a spin. It bites hardest on the cohort staking £10 to £50 a spin — the players who generate a disproportionate share of gross gaming yield and who, by definition, are the ones most likely to have a mirror account already open in another browser tab.

Why the cap is porous by design

Stake caps apply at the account level, tied to a licence, a domain, and a payment method. They do not apply to the player. A cap on brand.co.uk has no bearing on brand-mirror.com, even when both are operated by the same parent company under different licences — commonly Curaçao, Anjouan, or until recently a Malta or Kahnawake arrangement. The player's KYC on the offshore entity is usually lighter, the deposit limits are self-set rather than imposed, and the stake cap simply does not exist.

The 19% figure and how it's calculated

The 19% is a net migration estimate, not a gross one. Affiliate networks track click-throughs on mirror links using sub-IDs that persist across the player's first deposit. Comparing the fourteen days before a cap introduction with the fourteen days after, across the eleven brands, the mirror-attributed first-time depositor count rose by a median of 19.3%, with a range from 11% to 34%. The wide range is instructive: brands with a strong retail presence and older player bases saw less migration; digital-native brands with younger, mobile-heavy audiences saw more.

Two caveats worth stating plainly. First, the sample is small and affiliate-reported, which means it skews toward players who arrive via comparison sites and bonus aggregators rather than direct. Second, some of the migration would have happened anyway — offshore mirrors grow on their own trajectory, and disentangling the cap effect from baseline drift requires a control group that the data does not cleanly provide. The 19% is best read as an upper-middle estimate, not a precise measurement.

What is not in doubt is the direction. Every operator that has introduced an overnight cap and subsequently reviewed its mirror traffic has found an increase. The question operators ask privately is not whether migration happens but how much of it they can recapture through "win-back" emails sent the following morning.

The player-side economics

For a player staking £20 a spin on a 96.5% RTP slot, the expected loss per hour at 400 spins is £280. Cap that stake at £2 and the expected hourly loss falls to £28 — a tenfold reduction in theoretical loss, which is precisely the point. But the player who was staking £20 was not doing so to lose £28 an hour. They were doing so for the variance, the volatility, the possibility of a £4,000 hit on a bonus round. A £2 cap does not give them a safer version of that experience; it gives them a different, slower, less interesting one. The offshore mirror restores the original product.

This is the uncomfortable core of the policy. Stake caps work on players who are ambivalent about gambling and stop when friction appears. They do not work on players who are determined, and those are the players the policy is aimed at. The 19% is the visible portion of that failure — the portion that leaves a trail through affiliate links. The invisible portion, players who already held offshore accounts or who move to crypto casinos with no affiliate relationship at all, is larger and untracked.

The crypto complication

Crypto-only casinos, which hold no UK licence and accept no GBP cards, have become the default destination for a subset of capped players. Deposits via USDT on Tron settle in under a minute, KYC is often a self-declaration below certain thresholds, and there is no stake cap, no affordability check, and no self-exclusion register that reaches them. GamStop does not cover offshore operators, and while some crypto casinos voluntarily integrate with self-exclusion schemes, uptake is patchy. A player who self-excludes at 03:00 on a UK site can be staking £50 a spin on a crypto mirror by 03:10.

What the Commission can and cannot do

The Gambling Commission has been explicit that it cannot regulate offshore operators directly. Its leverage is indirect: licence conditions on UK operators, payment blocking via the Financial Conduct Authority and UK Finance, and domain blocking through ISPs under section 97A of the Copyright, Designs and Patents Act — a mechanism designed for piracy, awkwardly repurposed. Payment blocking is the most effective lever, and it is precisely why crypto rails matter. A card block is trivial to enforce; a USDT transfer from a self-custody wallet is not.

The Commission's 2024–25 enforcement activity has focused on UK licensees' marketing and affiliate relationships, with several six-figure fines for operators whose affiliates promoted unlicensed sites. That closes one door. It does not close the mirror sites themselves, which increasingly operate without affiliate programmes at all, acquiring players through Telegram groups, Discord servers, and word-of-mouth among communities that treat the UK licence as a quality mark on the product they are deliberately avoiding.

Where this leaves the overnight player

The policy question is whether a stake cap that pushes 19% of affected handle offshore is a net harm reduction or a net harm transfer. If the players who migrate are the ones most at risk, and the offshore environment offers them less protection than the UK one they left, the cap has moved the problem rather than solved it. If the players who migrate are largely recreational and the cap successfully deters the genuinely vulnerable from staking beyond their means, the 19% is a acceptable cost.

Nobody currently has the data to answer that, because the offshore side of the ledger is invisible by construction. What is measurable — the 19%, the affiliate sub-IDs, the fourteen-day windows — describes only the migration that leaves a trail. The question worth asking is not whether the cap works, but whether a policy whose success depends on players staying put can survive contact with a player who has already decided to leave.