A Cheating Report: The Current Hand and 3,865 Historical Hands

Finding: the deals and settlements around this report checked out; we found no evidence supporting the cheating allegation. On October 31, 2026, a Hold’em player reported a tablemate without naming a hand or adding a description. We reviewed all three completed hands they shared immediately before the report, the hand still in progress when it was filed, and then all 3,865 saved hands involving the reported player through 14:37 Singapore time that day. This account is anonymized and does not publish account identifiers, room codes or unshown hole cards.

Start with the hand in progress when the report was filed

The hand ID’s pre-deal allocation time and the settlement time identify a hand spanning the report. In that hand, the reported player raised to 350 on the turn; the reporter folded after contributing 20. On the river, the reported player bet 880 and everyone else folded. The unmatched 880 was refunded, producing a net win of 1,130. There was no showdown, so no unshown hole cards are published here. The reconstructed deal, action contributions, refund and payout all matched. When a report names no hand, we inspect both the ongoing hand and the completed ones nearby; timing is not an explicit identification by the reporter.

What happened in the earlier ace-king all-in?

In the reported player’s first hand after joining, they held A♠K♠ and committed their full 10,000 stack before the flop. The caller was the player named in the separate collusion report, referred to below as B. Both ultimately made one pair, and ace-king won on the higher kicker, for a net gain of 10,220. The reporter had put in 20 before folding; most of the winnings came from the other caller. In the next two hands, the reported player lost 20 and then folded without committing chips.

The sequence matters: the reporter’s two earlier losses of 10,000 each happened before this player joined. They cannot establish that this person participated in cheating in those hands. Because the report did not specify a hand, these three hands are the complete nearby window identified by the filing time, not a disputed hand confirmed by the reporter.

What did the 3,865-hand review cover?

The records span October 21–31. Locally, we rebuilt each deal under its original protocol from the stored seed, player randomness and deal order, checking the commitment, all saved hole cards and the board. We then reevaluated showdown pot winners and reconciled contributions, refunds, split pots, each seat’s stack change and the player’s history index. All 3,865 hands passed these checks. Of these, 3,754 used the earlier commitment protocol and 111 used the newer zero-knowledge protocol. Passing the historical audit does not mean every older hand has a downloadable v2 proof.

We also examined starting cards and outcomes. The player received pocket aces 18 times against a random-dealing expectation of about 17.5; QQ, KK, AA or AK appeared 108 times against about 99 expected. These two measures showed no conspicuous excess of strong starting hands. There were 912 hands with a positive net result, 2,758 negative and 195 unchanged. Frequencies and losses are context, not standalone proof that a person never cheats. Our finding rests primarily on the deal and settlement checks; profitability itself is not a violation.

Did the decisions suggest knowledge of hidden cards?

We also estimated equity from the saved actual hole cards, using only board cards visible at each decision. The reported player folded on 27 rivers where their hand would have beaten every opponent still contesting the pot. Among 298 decisions that committed the remaining stack with an action of at least 20 big blinds, 88 had under 20% actual pot equity against the opponents who had not folded. Across 1,043 hands where the reported player was eligible for a contested showdown pot, the reported player received at least a share of a pot in 551, including ties. These records do not show perfect use of hidden information; they cannot exclude selective assistance, and this review did not examine players’ devices.

This is a hindsight diagnostic, not information the player had when acting. Turn and river runouts were enumerated; earlier streets used fixed-seed sampling. In the ace-king hand, B’s call had about 27.5% equity against the actual A♠K♠, compared with roughly 49% required by the pot odds. That risky call deserved inclusion in the collusion review; a correct deal alone does not explain it away.

Verify the zero-knowledge proof yourself

The hands around this report use the newer protocol, and verification is already live. Sign in with the account that played the hand, open History → the Hold’em hand → Verify fairness, or select the hand on the Fairness Verifier, then choose “Verify this hand.” The browser checks the shuffle proof and decryption proofs for cards you may see. Folded and undealt cards remain private. You can also download the proof JSON and follow the offline verification guide.

Use the device and browser you played on where possible. “Fully verified (witnessed on this device)” also confirms agreement with the commitment, randomness, seats and visible cards recorded locally during play. “Server record valid, not fully witnessed on this device” means the proof is valid but the device lacks a complete record. These results differ, as explained in our zero-knowledge proof guide. This audit found no dealing or settlement fault; it does not replace a player’s device record or verify off-table communication. A separate report about collusion at the same table prompted a review of shared-hand folds and chip contests.