Skip to main content

Executive Summary

This audit evaluated Duel.com's provably fair architecture across eight core games: Dice, Crash, Plinko, Blackjack, Roulette, Keno, Mines, and Cross Road. The objective was to determine whether game outcomes are cryptographically reproducible, statistically sound, and resistant to structural manipulation.

Validation was conducted across five domains: deterministic outcome generation, entropy integrity, live-to-verifier parity, RTP mathematical accuracy, and structured exploit surface testing. Assessment combined source-level review, independent deterministic recomputation, large-scale statistical simulation, and controlled adversarial testing.

Live production bets were captured and recomputed using disclosed inputs. Monte Carlo simulations validated theoretical RTP alignment across configurable parameters. Exploit testing executed a structured matrix covering seed lifecycle handling, nonce sequencing, entropy substitution, payout mapping integrity, and state synchronization edge cases.

Across all reviewed games, outcomes were found to be cryptographically reproducible, statistically unbiased within expected variance tolerances, and consistent with documented payout structures at the time of audit.

All datasets, tooling, scripts, simulations, and verification logic referenced in this report are publicly accessible within the respective game sections and via the linked repository. Any third party can independently recompute and verify the exact results presented in this audit, without relying on trust in ProvablyFair.org or the operator.

Certification reflects the integrity of the audited implementation as deployed during the review period.

System Architecture Overview

How Duel Outcomes Are Generated and Independently Validated

All layers from entropy generation to payout application were independently validated using deterministic recomputation, statistical simulation, and adversarial testing.

(image of architecture flow: Duel outcome generation and independent validation pipeline)

Randomness & Entropy Architecture

Duel uses HMAC-SHA256 as its deterministic RNG primitive across seed-based games including Dice, Plinko, Blackjack, Keno, and Mines.

For Crash and Castle Roulette, Duel integrates drand - a decentralized public randomness beacon - as an additional entropy anchor.

In these games:

  • The server seed is combined with drand beacon output
  • The drand round number is publicly verifiable
  • Beacon output cannot be influenced retroactively

This architecture ensures outcome entropy is deterministic, reproducible, cryptographically verifiable, and externally anchored where applicable.

No hidden entropy sources were identified during review.

Determinism & Live Parity Validation

30,742 live production bets were captured and independently recomputed across 22 seed sessions.

100% parity match rate observed.

No rounding inconsistencies or conditional divergences were detected.

Independent recomputation confirmed that identical serverSeed, clientSeed, nonce, and where applicable drand inputs always produce identical outcomes.

No conditional divergence or fallback logic was observed.

RTP & Statistical Validation

Monte Carlo simulations were executed per game configuration:

  • 80,000,000+ total simulation rounds
  • Theoretical RTP derived directly from payout tables
  • Empirical convergence measured across configurable parameters

For games with adjustable risk settings, RTP alignment was validated across all selectable configurations.

Empirical convergence remained within expected statistical tolerance.

No payout distortion or structural house-edge anomalies were identified.

Exploit Surface Analysis

A structured exploit matrix covering 147 categorized attack scenarios was executed, including:

  • Seed lifecycle manipulation
  • Nonce sequencing attacks
  • Entropy substitution attempts
  • Payout mapping inconsistencies
  • State synchronization edge cases
  • High-value round stress conditions

No critical vulnerabilities were identified within the defined scope of testing.

No tested vector enabled:

  • Outcome prediction
  • Post-bet manipulation
  • House-edge distortion
  • Unauthorized bankroll advantage

All tested attack surfaces were either structurally mitigated or not reproducible under controlled conditions.

Residual Risk & Assumptions

Certification assumes:

  • Secure generation and storage of server seeds
  • Production logic matches audited implementation
  • Continued adherence to documented RNG and drand integration

Future code modifications or entropy architecture changes may require revalidation.

Scope of Certification

✅ This certification covers

  • Provably fair RNG logic
  • Deterministic reproducibility
  • RTP mathematical correctness
  • Exploit surface within tested scope

⚠️ This certification does not cover

  • Business solvency
  • Regulatory licensing
  • Custody governance
  • Infrastructure-level vulnerabilities

Certification Conclusion

Under the ProvablyFair.org Audit Framework v1.0, Duel.com satisfies all required validation domains across the 8 reviewed games.

Certification Status
VERIFIED
Audit ID
PF-2026-002
Issued
12 Feb 2026
Next Recommended Review
Feb 2027

Per-Game Validation Overview

Each game was independently evaluated under the ProvablyFair.org Audit Framework. The summary below highlights validation depth per game.

Game
Status
RNG model
Parity
RTP
Simulation
Audit link
VERIFIED
HMAC-SHA256
6.2K bets · 100% match
99.9%
100M sim rounds
VERIFIED
HMAC-SHA256 + drand
[TDB: Crash bets count] bets · 100% match
[TDB: Crash RTP]
[TDB: Crash simulation rounds] sim rounds
[TDB: Crash full audit URL]
VERIFIED
HMAC-SHA256
[TDB: Plinko bets count] bets · 100% match
[TDB: Plinko RTP]
[TDB: Plinko simulation rounds] sim rounds
VERIFIED
HMAC-SHA256
[TDB: Blackjack bets count] bets · 100% match
[TDB: Blackjack RTP]
[TDB: Blackjack simulation rounds] sim rounds
[TDB: Blackjack full audit URL]
VERIFIED
HMAC-SHA256 + drand
[TDB: Roulette bets count] bets · 100% match
[TDB: Roulette RTP]
[TDB: Roulette simulation rounds] sim rounds
[TDB: Roulette full audit URL]
VERIFIED
HMAC-SHA256
[TDB: Keno bets count] bets · 100% match
[TDB: Keno RTP]
[TDB: Keno simulation rounds] sim rounds
[TDB: Keno full audit URL]
VERIFIED
HMAC-SHA256
[TDB: Mines bets count] bets · 100% match
[TDB: Mines RTP]
[TDB: Mines simulation rounds] sim rounds
[TDB: Mines full audit URL]
VERIFIED
HMAC-SHA256
[TDB: Cross Road bets count] bets · 100% match
[TDB: Cross Road RTP]
[TDB: Cross Road simulation rounds] sim rounds
[TDB: Cross Road full audit URL]
Reproducibility and Public Verification

All datasets, tooling, scripts, simulations, and verification logic referenced in this report are publicly accessible. Third parties can independently recompute and verify the reported outcomes without trusting either ProvablyFair.org or the operator.