Methodology note: we test median withdrawal speed, not minimum or best-case. Minimum times are marketing. Median times reflect what a verified player making a routine-sized withdrawal actually experiences. Tests use amounts below the operator's standard review threshold to avoid security-hold variance. All tests were conducted on weekday afternoons UTC, outside of major sporting events. Network confirmation time on chain is excluded from operator processing time.

The ranked results

• 1. Razed: median 2m40s. Razed has led this table since we first tested it in Q3 2025. The operator processes withdrawals through an automated system with no human-review layer for standard verified-account amounts. It holds only a single Curaçao CGA licence, which means no multi-jurisdiction compliance overhead slowing the cashier. There is a cost to that simplicity: Razed's dispute escalation path is shorter than Stake's or BC.Game's, but for withdrawal speed the structure is an advantage.
• 2. Stake: median 4m10s. Stake the operator processes at speed commensurate with its scale. The operator handles substantially more withdrawal volume than any other platform we cover, and its automated review layer is well-calibrated to flag genuine edge cases without pausing routine cashouts. Stake drops below Razed primarily because larger-platform security checks add a processing layer absent at a newer, smaller operator.
• 3. BC.Game: median 45m. BC.Game's median is longer than the top two but has been broadly stable over three quarters of testing. The operator supports over 165 cryptocurrencies, which adds routing complexity to the cashier. For USDT on TRC-20 and BTC, times are at the faster end of the 45-minute median; for altcoins, times vary more. The BC.Game review notes which currencies process fastest.
• 4. Roobet: median 38m, improved from 60m in Q1 2025. This is the most notable movement in the table. Roobet has cut its median withdrawal time by 37% over twelve months. The operator overhauled its cashier backend in mid-2025, and the results are now visible in actual processing times rather than just support-page promises. Roobet is no longer in the bottom tier on this metric.
• 5. Betplay: median 52m. Betplay's Lightning Network integration produces near-instant results for Lightning-specific withdrawals, but Lightning withdrawals represent a subset of total cashouts and the Lightning category is excluded from our median to keep comparisons like-for-like. Standard on-chain BTC withdrawals from Betplay land in the 40–65 minute range.
• 6. Gamdom: median 61m. Gamdom is mid-table and has been for three consecutive quarters. Processing is consistent rather than fast. The operator's user base leans toward players in long sessions who are not urgently withdrawing after a single session, which may explain why cashier speed has not been a competitive priority.
• 7. Duel: median 74m. Duel's quirky low-rollover positioning has attracted a specific player profile, and withdrawal speed has not historically been a featured differentiator. Our tests show consistent mid-to-long processing times with low variance; the 74-minute median reflects predictable rather than particularly slow cashier operations.
• 8. Rainbet: median 93m. Rainbet remains the slowest in the group. The operator's sportsbook-plus-casino hybrid model involves a more complex cashier with fiat-equivalent balance handling that adds processing steps. The 93-minute median is an improvement on the 110-minute figure we recorded in Q3 2025, but still places Rainbet in a distinct tier below the other seven operators.

Roobet cut its median withdrawal time by 37% over twelve months. That's the kind of improvement that shows up as a genuine competitive advantage in a category where 40 minutes versus 60 minutes is a real difference for players managing short sessions.

What the numbers don't capture

Median processing time is the right central measure, but two other factors matter and aren't in the table above. First, tail risk: how often do withdrawals get flagged for manual review, and how long does that take? Razed's automated system is fast on routine cashouts but has fewer escalation options when something goes wrong. Stake's larger compliance team handles edge cases more systematically. Second, currency breadth: BC.Game's 45-minute median reflects the full range of 165+ currencies. Players using only BTC or USDT will see faster processing at BC.Game than the median implies.

The full methodology and currency-specific times are in the withdrawal speeds guide. For a direct side-by-side of all eight operators on this and other metrics, the comparison tool is filterable by withdrawal speed tier. And for a summary ranking that weighs withdrawal speed alongside game selection, bonuses, and licence quality, see the best crypto casinos ranking, where withdrawal speed contributes roughly 15% of the overall score weighting.

The withdrawal speed table has been stable at the top and bottom for two quarters. Razed leads, Rainbet trails. The middle cluster (BC.Game, Roobet, Betplay, Gamdom, Duel) is genuinely competitive and changes quarter to quarter. Roobet's improvement is the most substantive development in the last year. Whether Rainbet's slow trajectory of improvement continues into Q2 2026 will determine whether it remains in a category of its own at the bottom of the table or joins the middle group.

The mechanics of crash games have not changed since the category emerged. A multiplier starts at 1.00x and climbs until it crashes at a point determined by a provably fair algorithm seeded before the round begins. Players cash out before the crash or lose their stake. The operator's edge comes from instant-bust rounds: rounds that crash at exactly 1.00x before anyone can act. In a game with 1% house edge, approximately 1% of rounds end this way. That's the entire mathematical structure of the game; it's not complicated.

The RTP picture in 2026

The current confirmed RTPs for the major originals are: Stake Crash at 99% (1% house edge), BC.Game Crash at 99% (1% house edge), Roobet Crash at approximately 96.5% (3.5% house edge). Gamdom's crash-style originals vary by title but generally sit in the 97–99% range. These figures are verifiable through the provably fair system built into each game (the seed, hash, and crash point can be independently confirmed after each round, which is a higher standard of transparency than certified-RNG slots, where you take the certification on faith. Our provably fair explainer covers the verification workflow in plain terms.

The difference between 99% and 96.5% RTP matters over volume. At $200 wagered per session, a 99% RTP game costs a player $2 in expected losses; a 96.5% game costs $7. Over 50 sessions, that's $100 versus $350. The 2.5 percentage point gap is not cosmetic. It's also why choosing between operators on crash games is not just a branding decision: the underlying RTPs vary by enough to meaningfully affect long-term expected return.

Crash games with 99% RTP compete directly with blackjack on basic strategy for the lowest house edge in online casino gaming. Unlike blackjack, the fairness is player-verifiable.

What maturation looks like

The 2020 hype phase was characterised by several things that have since faded: opaque crash point algorithms, streams designed to make $50,000 profits look routine (they weren't), and a general absence of RTP disclosure. Operators who ran crash games with 94% or 95% RTPs could do so without players having clear grounds for comparison. The provably fair community, initially small and technical, has since produced accessible verification tools and the RTP data is now widely published.

The maturing of crash games also shows in the betting behaviour data. The average multiplier target at major platforms has converged toward 2x–5x for most players, which is consistent with risk-adjusted expected value optimisation rather than lottery-style moonshot chasing. The $5,000,000 maximum win advertised by Roobet Crash exists, but is not the operative reason anyone should be playing crash games. The operative reason, if there is one, is that a 99% RTP provably fair game with a 1% house edge is among the cheapest casino games available by that metric.

The four originals compared

BC.Game Crash and Stake Crash are the closest comparisons: both run at 99% RTP, both use provably fair verification, both offer an auto-cashout function and a live game history panel. The user interface at Stake is marginally cleaner. BC.Game's crash game benefits from the ecosystem of BCD rakeback accruing on every wager, which modestly lowers the effective house edge for players who actually earn it through wagering. Roobet Crash, at 96.5%, has a higher house edge but offers a maximum bet of $2,000 per round and maximum win of $5,000,000, a more volatile profile for players actively seeking large multipliers. Gamdom's originals, including their crash-adjacent games, sit in the 97–99% band depending on title and reflect the platform's skins-era heritage of volatility tolerance.

The crash game strategy guide at this link works through the mathematics of target multiplier selection and bankroll management in detail. The short version: target multiplier does not change your expected return per dollar. What it changes is variance. Targeting 1.5x produces frequent small wins and slow bankroll erosion. Targeting 50x produces long losing streaks punctuated by large wins. Both cost the same in expectation. Players who understand that are playing a different game from players who think high multipliers represent better value.

The crash game category is not exciting in April 2026. It's predictable, well-documented, and mathematically legible. That's the most positive thing you can say about a casino game category, and it wasn't true three years ago.

The old Curaçao regime was simple to the point of parody. Four master licence holders (Antillephone, Curaçao eGaming, Gaming Curaçao, and C-island) could issue sub-licences to effectively anyone who passed a basic due-diligence check and paid the annual fee. Oversight was nominal. The Curaçao Gaming Control Board, the nominal regulator, had limited enforcement resources and even more limited appetite to use them. The result was a jurisdiction that served primarily as a legitimacy signal: enough to get a casino listed on comparison sites, not enough to guarantee meaningful player protections.

What LOK actually changed

LOK abolished the master/sub-licence structure entirely. As of December 2024, the only body that can issue a gaming licence in Curaçao is the new CGA. Every new operator must apply directly. Every existing operator that wanted to remain current had to apply for a LOK transitional licence before the NOOGH (the old ordinance) lapsed. Applications for B2C operators reopened in mid-March 2025, with B2B providers following in May and June.

The new requirements are substantively different from the old ones. Operators must establish a physical office in Curaçao. Within four years, they must employ at least one key person locally (in addition to a managing director); within five years, at least three. Operators with annual Gross Gaming Revenue above €50 million face twice-yearly financial audits and additional AML reserves. KYC procedures must now comply with FATF and Caribbean FATF standards, with suspicious transaction reporting through the goAML platform. Due diligence fees run €140 to €270 per individual assessed.

Knowing whether an operator has converted to the new CGA licence or is still on a grandfathered approval changes what dispute-resolution route is actually available to you.

The substance requirements are where the reform bites most. An operator that was a letterbox entity under the old regime now has to have actual humans in Willemstad. That raises operating costs and, in theory, raises the floor for who can hold a licence at all. Smaller sub-licence holders that couldn't or wouldn't meet the new criteria have either migrated to other jurisdictions (Malta, Gibraltar, Isle of Man) or are still operating under extended NOOGH approvals while hoping the extension politics play out in their favour.

The two-tier market for players

When you're reviewing a crypto casino in 2026, the Curaçao question is no longer binary (licensed or not). It's tripartite: CGA-licensed under LOK, still on an extended NOOGH approval, or operating without any valid licence at all. These have meaningfully different implications for players.

A LOK-licensed operator has submitted to CGA audits, met the physical presence requirement, and is subject to an enforcement framework with actual teeth: the CGA can revoke licences, impose fines, and refer cases to Curaçao's public prosecutor. That's materially stronger than what the old board offered. An operator still on a grandfathered NOOGH approval is, by definition, in a transition state: the licence is technically valid but the operator has not yet met the new standards. And an unlicensed operator is exactly what it sounds like.

For practical purposes, the most important check is whether a casino displays a current CGA certificate rather than a historical sub-licence number from one of the four old master holders. Sub-licence numbers formatted as [four digits]/JAZ or similar were issued under NOOGH and should be treated with appropriate scepticism; they may still be valid under the extension, or they may have lapsed. The operator's licensing page should state which entity holds the licence and which regulatory body issued it. If it doesn't, that's worth noting. Our warning signs guide covers the specific things to look for in a site's licensing footer.

How this filters through to our reviews

Of the eight operators we cover (Stake, Roobet, BC.Game, Razed, Betplay, Gamdom, Duel, and Rainbet, several of which have historically operated under Curaçao sub-licences. The licensing status of each is noted in the individual reviews and updated when the operator's status changes. Razed, which launched in 2024, applied directly under the new CGA framework and has never held a NOOGH sub-licence, which gives it a cleaner post-LOK compliance record than operators that carried over from the old system.

The broader point for players is that the LOK reform has raised the baseline for what a Curaçao licence means, but it hasn't resolved the information asymmetry between operators and players overnight. A casino can claim CGA licensing in its footer while the certification is still being processed. The only reliable check is the CGA's own public licence register, which went live in 2025. We cross-reference it in our licence comparison guide. For a ranked view of the sites we cover and their current regulatory status, the best crypto casinos page is updated quarterly.

The LOK reform is a genuine improvement on the previous system, and it was overdue. Whether it becomes substantively meaningful for players depends on how aggressively the CGA enforces against non-compliant operators over the next two years. The framework is there. The track record is not, yet.

Standard on-chain Bitcoin deposits at sportsbooks carry two problems for live betting. First, blockchain confirmations. Even with one confirmation required, on-chain BTC takes a minimum of ten minutes in normal conditions and substantially longer during network congestion. Second, mempool dynamics. When block space is contested during major sporting events, fee estimates spike and lower-fee transactions sit unconfirmed for extended periods. A player trying to reload during the third quarter of an NBA game, precisely when live markets are most volatile, can find their deposit stuck for twenty minutes after the window they wanted has closed.

What Lightning actually fixes

Lightning Network routes BTC payments off-chain through pre-funded payment channels. Settlement is near-instant (typically one to three seconds) and fees are denominated in satoshis, usually under $0.01 regardless of network congestion. For live betting, this means a player can top up their Betplay balance between overs in a cricket match or between rounds of a UFC card and have funds available before the next market opens. The Bitcoin network's state at that moment is irrelevant.

Betplay's own data from 2021 showed that the average Lightning deposit was 40–50% smaller than the average on-chain deposit, consistent with Lightning being used primarily for quick reloads rather than large initial deposits. That pattern reflects how the technology fits the use case: large capital movements still default to on-chain for routing reliability, while frequent small top-ups shift to Lightning. The Betplay review has current deposit minimums and currency options.

A player trying to reload during the third quarter of an NBA game can find an on-chain BTC deposit stuck for twenty minutes after the window they wanted has closed. Lightning removes that problem entirely.

There's also a micro-bet case. On-chain fees make deposits below $20 economically inefficient: the fee might represent 5–10% of the deposit value. Lightning's fee structure, effectively zero, makes $5 reloads viable. That changes the math for players who want to test live markets cautiously without committing a large bankroll in one transaction. It's also useful for players trying Betplay for the first time: a $10 Lightning deposit is a low-stakes way to verify that the withdrawal flow works before moving a meaningful amount.

The latency argument for live markets

The latency issue extends beyond the deposit itself. Live sportsbooks update odds on short cycles; some in-play markets reprice every few seconds. A player who wants to bet a specific line has a narrow window before the market moves or closes. If their deposit is sitting in mempool, that window is gone. Lightning removes the deposit lag from the equation, leaving only the time to navigate the cashier and place the bet. That's not a marginal improvement; it's the difference between a live betting strategy being executable or not.

Which other operators will add Lightning? Among the eight we cover, the short answer is: probably Stake eventually, and not the rest any time soon. Stake has the technical infrastructure and the user base to justify the integration cost, and its game-library breadth means the sportsbook share of its revenue is large enough to make live-betting latency improvements commercially relevant. Roobet's product is more casino-skewed; the sportsbook is secondary. BC.Game has a large game library but its crypto deposit support (over 165 currencies) is already a complexity management challenge; Lightning adds another routing layer. Gamdom and Razed are focused on casino originals and rakeback rather than live sports volume.

The limits of the Lightning argument

Lightning is not universally better than on-chain. Large deposits (anything above roughly 0.01 BTC, depending on channel liquidity, may fail to route through Lightning channels. A player depositing $5,000 before a major event is better served by on-chain BTC or a stablecoin on a high-throughput chain (TRC-20 USDT settles in seconds and has near-zero fees). Lightning's advantages concentrate in the $5–$500 reload range. Above that, on-chain or alternative networks are more reliable.

Betplay's cashier handles both Lightning and on-chain BTC, so players can use whichever fits their deposit size. The practical workflow for live sports betting is Lightning for reloads, on-chain for initial funding. Our crypto deposit guide covers this split in more detail, and the live betting latency guide tests deposit-to-bet times across several operators. Betplay's Lightning integration, as of Q1 2026, remains the fastest on-ramp to a live market among the platforms we cover. The operators that haven't added it aren't making a mistake on product; they're making a calculation about which part of their user base the investment would serve.

BCD (BC Dollar) is BC.Game's internal token, pegged 1:1 to the USD. Every deposit triggers a rakeback bonus denominated in BCD, but that BCD sits locked until you wager it out. The release formula is: wager amount × 1% × your current release rate. The release rate starts at 20% for VIP levels 0 through 7 and climbs incrementally to 24% at SVIP 16 and above. So for every $100 you wager, you get $0.20 of BCD at the base rate, not $1.00, and certainly not whatever percentage the deposit bonus headline advertised.

The monthly deposit bonus layer

On top of the ongoing rakeback, BC.Game runs a monthly deposit bonus that resets on the first of each month. The first deposit of the month earns 180% up to $20,000 in BCD. The second earns 240% up to $40,000. The third earns 300% up to $60,000. The fourth earns 360% up to $100,000. These percentages are also delivered as locked BCD, subject to the same release formula. The upshot is that a high-volume player who deposits $10,000 in a single first deposit of the month receives $18,000 in locked BCD, which then requires $18,000 ÷ (1% × 20%) = $9,000,000 in wagering to fully release at base rate. In practice, players earn BCD incrementally as they play, so the figure isn't a wall, but it illustrates how the headline number diverges from the cash-equivalent return.

At $1K monthly turnover, Stake's welcome bonus is worth more than a full month of BCD rakeback; the wagering requirement alone takes three months at that volume.

Three turnover scenarios

Let's run approximate numbers at three monthly casino wagering volumes, assuming VIP level 0 (20% release rate) and a house edge of around 1% (consistent with BC.Game's own Crash game and similar originals with 99% RTP).

• $1,000 monthly turnover: At 1% house edge, expected loss is roughly $10. The BCD rakeback from ongoing play releases at $1,000 × 1% × 20% = $2.00 per month. If a first-month deposit bonus of $200 (20% of turnover, notional) is applied and the BCD is locked, you'd need $100,000 in wagering to fully release it, not achievable at this volume within the month. Net rakeback realised: approximately $2 in BCD. Stake's base welcome bonus, by comparison, is worth roughly $5–10 in playable value at this volume once rollover is met. Stake wins at low stakes.
• $10,000 monthly turnover: Expected loss at 1% house edge: $100. BCD rakeback released from ongoing play: $10,000 × 1% × 20% = $20 per month. At this volume, a player can meaningfully chip away at a monthly deposit BCD bonus too: a $1,000 deposit at 180% generates $1,800 BCD, and $10,000 in wagering releases $20 of it. Total BCD realised: roughly $40. Gamdom at this level offers around 15% instant rakeback (up to 60% long-term) on a welcome-week basis; post-week, the standard rakeback tier for $10,000/month volume lands in the 20–25% range of net losses returned. The comparison is close, with Gamdom's simpler instant-rakeback mechanism slightly more transparent. BC.Game is competitive.
• $100,000 monthly turnover: Expected loss at 1% house edge: $1,000. BCD released from wagering: $100,000 × 1% × 20% = $200 per month. At higher VIP tiers (SVIP 16+, 24% release rate), that rises to $240. Add deposit BCD from a $10,000 first deposit (180% = $18,000 locked BCD, $200 released at $100,000 wagering). Total BCD realised: approximately $400. Gamdom's long-term rakeback at this volume can approach 40–60% of rake, but requires sustained play to hit the upper tiers. Stake's VIP programme at $100,000/month volume offers comparable rake-equivalent returns through its tiered cashback, though exact figures are negotiated rather than published. BC.Game's BCD model wins clearly at this volume if the player is depositing to capture the monthly deposit bonuses.

The catch

The BCD model rewards players who understand it and penalises those who don't. The locked-BCD mechanic means that players who deposit large sums, collect the bonus headline number, and then don't wager enough to release it are leaving value on the table; more accurately, BC.Game is holding that value until the release threshold is met. For casual players making one or two deposits a month, the deposit bonus BCD is mostly cosmetic at low volumes.

The comparison to Stake's welcome bonus is particularly sharp at low stakes. Stake's welcome package, detailed in our welcome bonus comparison, is straightforward: a deposit match with a defined rollover requirement, met within a reasonable timeframe at $1,000/month play. BC.Game's BCD equivalent requires more active management to extract equivalent value.

For high-volume players, the calculus flips. The monthly deposit bonus percentages at BC.Game are genuinely large, and the incremental BCD from $100,000+ monthly wagering adds up to a meaningful rake return. For those players, the full BC.Game review covers VIP tier progression and the SVIP release rate schedule in more detail. If you're comparing BC.Game to Stake directly on this metric, the Stake vs BC.Game comparison puts both structures side by side.

The short version: BCD rakeback is a good deal if you wager enough to release it and bad deal if you don't. That's not a flaw in the design so much as a statement about incentive alignment: BC.Game benefits when players chase release thresholds.

The SPA issued 15 full licences and 52 provisional approvals in January 2025, and Stake was in the first batch. The headline benefits were predictable from day one: a .bet.br domain obligation, mandatory BRL deposit rails, local dispute resolution via SENACON, and KYC aligned to Brazilian federal standards rather than the looser Curaçao regime Stake had operated under previously. Whether those things matter to any given player depends almost entirely on how they used the site before.

What the licence actually delivers

BRL deposits via Pix now work natively on Stake's Brazilian-facing platform. That's not nothing. Before the licence, Brazilian players were routing funds through crypto or international card processors, which added friction and, occasionally, conversion fees. Pix settlement is near-instant, which matters for a sportsbook where live-market reloads are time-sensitive. If you want to compare deposit flows across the sites we cover, the deposit guide has the current picture.

Local KYC is the other meaningful change. Stake now verifies Brazilian players against CPF (Cadastro de Pessoas Físicas) numbers, which pulls from the same identity infrastructure that Brazilian banks use. In practice this means verified players face fewer ad-hoc document requests mid-withdrawal, a persistent annoyance under the old Curaçao sub-licence model. KYC disputes that previously went nowhere beyond a support ticket now have a route to SENACON, Brazil's consumer protection body, which has actual enforcement powers and has already levied fines against non-compliant operators in other sectors.

Dispute resolution is the underrated benefit. Curaçao's Gaming Control Board has historically been responsive primarily to operators, not players. A Brazilian player with a disputed withdrawal can now file with SENACON and expect the complaint to be taken seriously, provided the operator is operating on the licensed platform. That is a material improvement, even if the average Stake user never needs to invoke it.

What still annoys players

The 15% CIDE-Bets levy, approved by Brazil's Senate in December 2025, is the outstanding irritant. Under this framework (technically a "Contribution for Intervention in the Economic Domain" applied to player deposits: a R$100 deposit arrives in a player's account as R$85 of playable balance. The deduction happens before a single bet is placed. Stake's FAQ on this point, as of March 2026, is best described as evasive. Whether the 15% is deducted at the platform level, withheld at the payment processor, or somehow passed through differently for crypto deposits remains genuinely unclear to most Brazilian users.

A Brazilian player with a disputed withdrawal can now file with SENACON and expect the complaint to be taken seriously. That is a material improvement, even if the average Stake user never needs to invoke it.

The winnings tax situation is more nuanced. Brazil's income tax schedule taxes gambling winnings on a bracketed basis: zero up to BRL 2,428.80, then 7.5% to 27.5% depending on the amount. None of that is unique to Stake, but the operator's in-platform guidance on withholding obligations is minimal. Players who win significant amounts in BRL and don't self-report face enforcement risk. Stake has not, at least publicly, built tax-reporting tooling into the Brazilian platform the way some domestic operators have.

The .bet.br domain requirement has also created minor confusion. Stake's global site remains accessible from Brazil without a VPN; the SPA relies on payment blocking rather than DNS-level enforcement; which means some Brazilian players are still using the unlicensed global version and missing both the Pix rails and the SENACON protections. Stake's onboarding flow does not redirect .com users to .bet.br automatically.

The wider picture

Stake's Brazil move matters beyond Brazil. It's the clearest signal that the operator is serious about regulated-market expansion after years of operating primarily under Curaçao sub-licences. The January 2025 appointment of Thomas Carvalhaes as Brazil country manager, someone with local regulatory experience, which suggests the compliance infrastructure is being built properly rather than bolted on.

For players deciding between Stake and unlicensed alternatives, the licence is a genuine differentiator on dispute resolution and KYC clarity. For players who care primarily about product depth, Stake's licensed Brazilian platform and its unlicensed global platform are nearly identical: the game library, live odds, and VIP structure are unchanged. The full breakdown of what Stake offers is in the Stake review. If you're comparing licensed operators head to head, our licence comparison guide covers how Brazilian, Curaçao, MGA, and Gibraltar credentials stack up in practice. And for those weighing several operators at once, the comparison tool filters by licence jurisdiction.

The honest summary: the Brazil licence has made Stake meaningfully better for Brazilian players on the specific issues of Pix deposits and dispute resolution. On tax clarity, it has changed almost nothing. That's not a criticism of Stake in isolation; it's a criticism of a regulatory framework that approved licensing without building out the player-education infrastructure that makes compliance legible to ordinary users.