Stake Crash

About Stake Crash

Stake Crash is the in-house crash game offered by Stake.com, one of the largest crypto-native online casinos by reported wager volume. The game launched in 2017 alongside Stake's other "Originals" suite (Dice, Plinko, Mines).

The operator declares a return-to-player (RTP) of 99%, corresponding to a 1% house edge — the lowest publicly stated house edge among major commercial crash titles. Each round's multiplier is derived from an HMAC-SHA-256 computation over a server seed, a player-controllable client seed, and a per-round nonce. The server seed is committed via SHA-256 hash before any round under it is played, and is revealed when the player rotates their seed.

Why this game matters

Stake.com is a high-volume reference point for crash gameplay across the crypto-gambling segment. A confirmed deviation in either column A (RTP) or column C (provably-fair integrity) at this scale would have outsized implications for player trust in the broader category. For Clash Watchdog AI, Stake Crash is the first benchmark target — both because of its market position and because its provably-fair surface is well-documented and amenable to full hash-chain verification under our methodology §4.2.

What we are watching for

• Column A (RTP): the 99% claim is testable under our minimum sample threshold of 5,000 rounds. We will publish observed RTP with a 95% bootstrap confidence interval and report any deviation in absolute percentage points.
• Column B (Distribution): truncated geometric distribution of the form M ~ 1/(1−U) with U ~ Uniform[0, 1−0.01] is the theoretical expectation. We test goodness-of-fit via chi-square and Kolmogorov–Smirnov.
• Column C (Provably-fair): every round in our verified window will be re-derived using HMAC-SHA-256 from the revealed server seed and observed client seed/nonce. Hash-chain commit-to-reveal continuity is the central check.
• Rotation analysis (Phase 2): seed lifetime distribution and rotation-event correlation with proxy-account events. See Rotation Analysis.

Public data sources we are using

• Source 1 (Official): Stake's per-round provably-fair history, accessible via the in-game "Fairness" tab on each completed round.
• Source 3 (Self-operated proxy): a Clash Watchdog AI proxy account, funded with a small standing balance, placing minimum stakes purely for observation. No proxy account is used to attempt to win — the goal is independent observation of the same rounds we read from Source 1.
• Source 2 (Community): opens in Phase 2 when the browser-extension ingest path is live.

Cross-validation rules are described in Three Data Sources and our methodology §2.

Audit status

A first Tier 1 (Provisional) audit is in preparation. No conclusion will be published until the minimum sample sizes in our methodology §2.3 are met. If you have observed something unusual on Stake Crash that you would like us to look at, tell us.