Duel.com holds PF-2026-DUEL certification under Framework v1.1
- Certified
- June 8, 2026
- Updated
- October 2026
- Next review
- December 2026 · Unannounced
- Scope
- 11 in-house games
You don't have to trust us. You can check our work.
All 11 game repositories publish the code, data and checks. Re-run the suite locally.
Browse all 11 repositoriesGame audit matrix
Open a game to inspect its full audit, test results and source repository.
| Game | RNG model | Live bets | Sims | Simulated RTP | Proven RTP | Result | |
|---|---|---|---|---|---|---|---|
| Plinko | HMAC-SHA256 | 8,100 | 27M | 99.917% | 99.9% | Pass | |
| Dice | HMAC-SHA256 | 6,700 | 98M | 99.922% | 99.9% | Pass | |
| Mines | HMAC-SHA256 | 7,050 | 24M | 99.904% | 99.9% | Pass | |
| Keno | HMAC-SHA256 | 5,100 | 40M | 99.970% | 99.9% | Pass | |
| Beef | HMAC-SHA256 | 6,450 | 4M | 99.147% | 99.2% | Pass | |
| Crash1 finding | HMAC-SHA256 + drand | 1,100 | 5M | 99.872% | 99.9% | Pass | |
| Castle Roulette | HMAC-SHA256 + drand | 1,350 | 5M | 100.007% | 100.0% | Pass | |
| Video Poker1 finding | HMAC-SHA256 | 5,400 | 10M | — † | 99.9% | Pass | |
| Blackjack | HMAC-SHA256 | 6,000 | 10M | 99.4670% | 99.4296% | Pass | |
| Groomer's Van2 findings | HMAC-SHA256 → PCG32 | 6,225 | 30M | 100.06% | 99.9720% | Pass | |
| Limbo | HMAC-SHA256 | 7,700 | 13M | 99.911% | 99.9% | Pass |
| Game / result | Live bets | Sims | Sim. RTP | Proven RTP | RNG model |
|---|---|---|---|---|---|
| PlinkoPass | 8,100 | 27M | 99.917% | 99.9% | HMAC-SHA256 |
| DicePass | 6,700 | 98M | 99.922% | 99.9% | HMAC-SHA256 |
| MinesPass | 7,050 | 24M | 99.904% | 99.9% | HMAC-SHA256 |
| KenoPass | 5,100 | 40M | 99.970% | 99.9% | HMAC-SHA256 |
| BeefPass | 6,450 | 4M | 99.147% | 99.2% | HMAC-SHA256 |
| CrashPass1 finding | 1,100 | 5M | 99.872% | 99.9% | HMAC-SHA256 + drand |
| Castle RoulettePass | 1,350 | 5M | 100.007% | 100.0% | HMAC-SHA256 + drand |
| Video PokerPass1 finding | 5,400 | 10M | — | 99.9% | HMAC-SHA256 |
| BlackjackPass | 6,000 | 10M | 99.4670% | 99.4296% | HMAC-SHA256 |
| Groomer's VanPass2 findings | 6,225 | 30M | 100.06% | 99.9720% | HMAC-SHA256 → PCG32 |
| LimboPass | 7,700 | 13M | 99.911% | 99.9% | HMAC-SHA256 |
RTP values apply to each report’s tested configuration and strategy, at the base stake tier. Simulated RTP is a Monte Carlo estimate; proven RTP is our own derivation from the published rules. † Video Poker’s return depends on strategy; only its optimal-play proof is shown.
Audit verdict
Five independently tested control areas underpin the certification.
All 11 games fully reproducible — every outcome recomputable from (serverSeed, clientSeed/drand, nonce, state)
HMAC-SHA256 across all games, with bias-free rejection sampling wherever a value is reduced to a range, except Plinko, whose u mod 2 is unbiased without it, and Castle Roulette, whose plain uint32 % 48 carries a disclosed bias of 1.55 × 10⁻¹⁰ per position (Crash and Limbo map their 32-bit value directly, with no reduction step); 2 games (Crash, Castle Roulette) additionally bind drand public beacon
100% match — 61,155 / 61,155 recomputable live bets reproduced exactly against an independent verifier
266M simulated rounds across all games; every theoretical RTP independently re-derived from first principles and matched within statistical tolerance
160 fairness integrity checks · 156 pass · 4 findings — all at the API input-validation layer, none affecting outcome determinism, RTP, or payout integrity · no fairness failures
Findings & disclosures
Four input-validation findings were disclosed to Duel. No fairness failure was recorded.
Malformed payloads receive an HTTP 200 response, including game-specific request probes.
View probesAuto-cashout values above the theoretical maximum crash point are accepted.
View probesThese findings concern API request validation. The audited cryptographic core, RNG, RTP derivation and payout logic operated correctly. Reproducible probes are published in Section 5 of each affected game report.
Audit process
Each game followed the same framework, with an independent implementation and captured live bets.
- Mar 2026Audit openedTesting period: March–May 2026.
- Initial auditFindings issued4 findings disclosed across 3 games.
- Jun 8CertifiedPF-2026-DUEL
- Oct 2026Re-audit under Framework v1.1The ten certified games retested under the updated framework.
- Oct 2026Limbo11th game added to the audit scope.
- Dec 2026Next check windowUnannounced
RNG, reproducibility, RTP and payout integrity within game logic. Platform finances, licensing, infrastructure, third-party titles and promotional programs are outside this audit.
Transparency & verification
Inspect the code, reproduce the audit or check a result from its disclosed inputs.
Reproduce the audit
Every game has a public repository with its verifier, captured dataset and simulation harness.
- Choose a game repository
- Clone the audited commit & install
- Run npm test
Verify a single bet
Recompute a game result from its revealed seeds and public inputs, independently of the casino.
Open ProvablyFair verifierDuel’s own verifier ↗Operator-run tool. Listed separately from the independent ProvablyFair verifier.
Read the methodology, scope and conclusions in the executive summary.