Five Wins in Six Hands: A 4,464-Hand Cheating Review
A report reviewed through six consecutive hands, historical losses, 4,464 starting hands and a 258-hand all-in benchmark, including the unusually lucky day.
Five wins in six hands and a net gain of 80,280 chips deserve review, but do not establish cheating.On November 1, 2026, we received a Hold’em cheating report. This article adapts the investigation response without publishing names, accounts, emails, room codes or hand IDs. Specific hole cards below were publicly shown in the original hands.
We examined all 4,464 distinct hands retained for the reported player through that evening, spanning October 21 to November 1 and 72 room sessions. The review covered hole cards, boards, actions on each street, contributions, refunds and pot allocation. All amounts are game chips, not cash profits.
Read the six hands in sequence
The blinds were 10/20. The reported player won four consecutive hands, lost one, then won again. AJ and K7 denote hole-card ranks; T means ten.
| Hand and hole cards | Main action and outcome | Net chips |
|---|---|---|
| 1: A♠J♦ | Shoved preflop for a total of 10,350; AQ and T3 called. Flopped a jack and won with jacks and eights. | +20,720 |
| 2: A♣9♦ | Raised to 8,940, then called another 1,610 after 53 shoved. The board brought Q, K, J and T for an ace-high straight. | +10,600 |
| 3: K♣7♦ | Called 9,450 preflop, flopped a pair of sevens and checked the turn. A river seven made trips; a 15,670 bet folded the other player still able to act. That uncalled bet was refunded. | +18,920 |
| 4: A♥6♣ | Raised to 10,060; two opponents committed their full 10,000 stacks. The excess 60 was refunded. Aces and sixes won. | +20,010 |
| 5: K♦T♦ | Called to a total contribution of 10,000 against AQ and A9. The flush missed and the hand lost. | −10,000 |
| 6: A♠4♥ | Called to a total contribution of 10,000 against two AJ hands. A flopped four became the winning pair. | +20,030 |
AJ beating AQ and A4 beating two AJ hands were favorable runouts. Calling 9,450 with K7 meant committing 472.5 big blinds, another decision worth examining. But this hot stretch was selected after the report: multiplying a few hand probabilities would not produce a “probability of cheating.”
The loss inside the streak matters too
In hand 5, AQ raised to 4,190 and A9 shoved to 10,000. The reported player held K♦T♦ and called another 9,980, bringing the total contribution including the big blind to 10,000. The board ran 3♦, 8♦, 6♥, 7♥, 3♥. The two players with chips checked it down; the flush missed and AQ won with an ace kicker.
That 10,000-chip loss came after the four wins and before the A4 win. The same sequence therefore includes a large commitment that lost. One losing hand cannot rule out cheating, so we also checked the history.
Were there historical counterexamples?
Yes, including a loss to the reporter. The following hole cards were publicly shown in their respective hands:
- Another hand on November 1:the reported player held A♦J♠ on 3♦, 7♠, 6♦, J♣, 9♠ and called the reporter’s 5,560 river shove. J♦9♥ won with two pair; the reported player lost 15,945 overall.
- A hand on October 31:the reported player held K♦K♠, put in 10,000 preflop and called another 1,080 on the flop. The board ran 2♥, 7♠, 4♠, 8♥, A♦; A6 overtook the kings on the river for a total loss of 11,080.
Across October 31, using Singapore dates, the player lost a net 30,251 chips. In the narrower sample of heads-up river decisions with one contested pot, 15 of 113 folds would have won outright against the actual opposing hole cards. Of 95 calls or all-in calls, 51 were behind. These are retrospective card comparisons, not a strategic-error rate: players cannot see opponents’ actual cards, and some folds may have been automatic.
The records do not show consistently perfect avoidance of losing hands. They also cannot, by themselves, rule out intermittent assistance.
Was the player consistently dealt better starting hands?
We compared all 4,464 starting hands with the theoretical distribution of two cards drawn without replacement from a standard 52-card deck. There are 1,326 equally likely combinations; 78 are pocket pairs, giving a theoretical rate of 5.88%. This is a random-dealing benchmark, not an observed average for other players. The categories overlap.
| Starting-hand category | Count | Observed | Theoretical |
|---|---|---|---|
| Pocket pair | 253 | 5.67% | 5.88% |
| Two cards of one suit | 1,070 | 23.97% | 23.53% |
| At least one ace | 700 | 15.68% | 14.93% |
| QQ, KK, AA or AK | 126 | 2.82% | 2.56% |
These measures are close to theory and do not show an unusually high long-run frequency of strong starting cards. Receiving good cards and winning a hand are different outcomes: the latter also depends on field size, decisions and the board.
Does the win rate match theory? Define the denominator first
The player finished ahead in 1,032 of 4,464 hands: 23.1% were net-winning hands. That includes hands folded along the way. Multiplayer poker has no universal “normal 50% win rate,” and this number does not establish parity with similarly skilled players.
To compare future board outcomes, we separately selected 258 hands with one contested pot and the contenders locked in by all-in action before the river: at most one contender had chips remaining, with no later-street fold, call or raise decisions. We calculated outcomes from the actual hole cards and the board already visible at the lock. “Win or tie” has the same definition on both sides of the comparison.
| Sample | Observed wins or ties | Theoretical rate |
|---|---|---|
| Historical subset: 258 hands | 147 hands, 57.0% | 56.8% (about 146.5 hands) |
| November 1 within that subset: 13 hands | 11 hands, 84.6% | 61.3% (about 8.0 hands) |
The historical subset was close to its theoretical benchmark; that day’s subset ran considerably better.Both results matter. Long-run proximity does not erase a lucky day, and 13 hands do not establish a long-run pattern. Nor is the 258-hand figure a theoretical win rate for all 4,464 hands.
The calculation removes all recorded dealt hole cards from the remaining deck, including folded cards, and uses only board cards visible at the lock—not the actual future board. We enumerate when one or two board cards remain and otherwise use 20,000 samples per hand with a fixed seed. Selection uses completed-hand structure, making this a retrospective descriptive comparison, not a pre-specified, outcome-independent cheating test.
What did the investigation establish?
We independently reconciled each seat’s chip movements across 4,464 hands and recomputed the winners of 2,884 contested pots. We found no duplicate cards or discrepancies in hand ranking, pot allocation or settlement. Starting-hand measures did not look unusually elevated, and the comparable historical all-in subset was close to theory; the six-hand stretch and that day’s small subset did benefit from favorable runouts.
The reviewed records did not provide sufficient evidence to substantiate cheating.Settlement reconciliation is not a security audit of hole-card access, player devices or off-table coordination; this review did not reverify the full shuffle commitment chain. Historical action records also lacked individual decision durations, so we cannot rule out real-time assistance or automation on that basis. For the scope of deal proofs, see Texas Hold’em: Zero-Knowledge Proofs; for analysis of player interaction, see our shared-hand collusion review.
The conclusion already uses the retained hand and action records; the reporter does not need to resubmit those same hands. A useful response to a winning streak should present the wins, the losses, the historical distributions and the scope of the calculations together.