AI

The Bot You Think You Spotted Needs a Dossier, Not One Hand

The player three seats to your left just called your river bluff with third pair. Your first instinct screams bot. Your second instinct should be to open a spreadsheet.

Poker rooms do not ban accounts because someone made a sick call. Security teams at PokerStars and GGPoker build cases on behavioral dossiers thick enough to survive legal review. The gap between your tilted accusation and their standard of proof is where most bot reports die. Understanding that gap helps you clear the table instead of crying wolf.

What the Rooms Actually Prohibit

PokerStars draws a line between two categories of automation. A fully automated bot clicks its own buttons. A “brain” system calculates the optimal action and feeds it to a human clicker. Both violate policy because the human executes, but does not decide. GGPoker casts a wider net, banning partially automated tools that reduce decision load even if a warm body sits in the chair. Auto-folding trash pre-flop, auto-raising to fixed sizes, and any software that trims the thinking all trigger enforcement.

The common thread is decision ownership. The room does not care how sophisticated your setup is; it cares who determined the action.

The Patterns That Matter

Bot detection lives in repetition at scale, not brilliance in isolation. Security algorithms hunt for four specific signatures.

Identical timing regardless of complexity. A human takes longer on a river decision with a bluff-catcher than on a pre-flop fold. A bot often does not. Record the delay between street and action. If the same account shows 2.8 seconds for a snap-fold and 2.8 seconds for a tanked river call, that is data.

Session architecture that ignores biology. Humans fatigue, eat, tilt, and quit. An account logging 18-hour sessions seven days a week, or starting at exactly 08:00 UTC and vanishing at exactly 16:00 UTC, broadcasts automation. Track start times, end times, and breaks across weeks.

Mechanical betting rhythms. Population tendencies vary bet sizing by board texture, stack depth, and opponent type. A bot often locks to fixed percentages, always 67% pot on the flop, always 33% on the turn, because its script lacks conditional nuance. Note sizing patterns across dozens of hands in comparable spots.

Synchronized account clusters. Multiple accounts joining the same table simultaneously, targeting identical weak players, and using indistinguishable lines against each other suggest coordinated operation rather than coincidence. Track player IDs, table movements, and overlapping sessions.

Why Your Hero Call Proves Nothing

Variance swallows single-hand reads. That “impossible” call might be a misclick, a drunk read, or a solver-trained regular applying GTO frequencies that look alien to your population-adjusted intuition. Multi-tablers using hotkeys produce mechanical timing. Shift workers play graveyard hours that resemble bot schedules. Accessibility software generates consistent action patterns. Disciplined nits execute identical lines for thousands of hands because deviation costs them money, not because they lack humanity.

The absence of chat or emotional response is equally weak evidence. Some of the most profitable regulars are stoics who treat bad beats like cloudy weather. Silence is not a smoking gun.

Building the File

Responsible observation means systematic recording without public accusation. Create a private log with these columns: Hand ID, date, time, decision duration, action taken, bet size as percentage of pot, board texture, and any adjacent accounts at the table. After fifty hands against the same opponent, calculate standard deviation in timing. After two hundred, map session schedules against a calendar. Look for variance collapse, the statistical flatline where human behavior should show jagged edges.

If patterns crystallize, compile before contacting security. PokerStars and GGPoker both route bot reports to dedicated security addresses, not general support. Your submission should contain exact player aliases, specific Hand IDs with timestamps, and objective descriptions of observed repetition. “Player X took 2.7 seconds on every decision across 47 hands” is actionable. “This guy is obviously a bot” is noise. Avoid speculation, emotion, or demands. The security team has algorithms and data scientists you do not. Your job is to hand them a door worth opening.

The Discipline of Doubt

Spotting automation is a skill that punishes certainty. Every false positive erodes your reads, damages a potentially innocent player’s reputation, and trains you to see bots where you see skill you do not understand. The players who clear tables are not the ones with the sharpest single-hand hunches. They are the ones with files thick enough that security has to pay attention. Bring that dossier. One hand is just a hand.