PPPlatform Software helps poker operators detect automated play, multiaccounting, and coordinated abuse using statistical and machine-learning analysis of hand histories—supported by expert review to build a clear, investigation-ready evidence base.
Designed for real operations: actionable flags, explainable reasoning, and workflows your security team can use.
Hierarchical clustering on normalized multi-stat distance, plus confidence interval and “range” analysis to reveal algorithmic consistency that humans rarely match.
Visual pattern dynamics across players and groups to detect abnormal bet sizing behavior—often a strong signal when combined with stats and schedule.
Advanced signatures focused on automated state reading and action input. Known signatures can be identified quickly; new ones may be detected after ~1000 hands in some environments.
Multiaccounters become obvious with enough hands: near-identical statistics + overlapping schedules. We use the same clustering foundation enhanced with schedule analysis to surface related accounts.
Automated penalty scoring (e.g., strong hand folds, suspicious 3-bets), threshold-based flagging of pairs, and money-stream summaries—built to accelerate manual review.
Parse hand histories, compute a broad set of poker statistics, plus bet sizing and schedule features.
Find suspicious outliers, synchronized changes, and tight-value clusters not typical of human play.
Review signals and build an evidence bundle: why it’s suspicious, what patterns recur, and where.
Flagged accounts/pairs, summary tables, charts, and investigation notes—ready for enforcement workflows.
Demo scope example: 113,068,657 processed hands across 32,446 players, with 63 identified as suspicious in the sample.
Reports highlight groups with many synchronized statistics—sometimes nearly identical values across many dimensions, which is highly improbable for independent human play.
Expert validation emphasizes combinations of signals and recurring signatures—not single metrics in isolation.
The case-study sample describes an operator engagement where a suspicious group was identified by algorithms, validated by experts, then accounts were closed and funds confiscated for AI use (as described in the sample).
Collusion/chip dumping modules use penalty points, thresholds, date grouping (moving toward session-based grouping), and can produce money-stream visuals and investigator reports.
index.html. The viewer below will load them locally.
Use the buttons to switch between documents.
Message us on Telegram and share: game types (NL/PLO), table sizes, timeframe of hand histories, and your enforcement goals. We’ll reply with an engagement plan and expected deliverables.