coinpokerbot Get a free trial
Home / Detection Signals

Behavioural Signals for the Detection of Automated Poker Play

Summary: Because cryptography can't identify a bot, detection is a behavioral problem. The strongest signals are statistical, not visual: the shape of a player's action-timing distribution, the entropy of their bet sizing, and multi-tabling patterns that exceed human capacity. No single signal is proof; detection systems combine many weak signals into a confidence score, then route the high-confidence cases to human review.

Empirical detection of automated play uses distributions, not individual hands: a population of seats is classified against a reference population of human play, and divergence is accumulated over many observations. Three signal families account for most of the classifier weight in operator-side detection pipelines.

Signal 1: Action-timing distributions

The single most reliable behavioral signal is time-to-act: how long a player takes to make each decision. Humans produce a broad, right-skewed distribution: fast on trivial spots, slow on hard ones, with natural noise from attention, fatigue, and tabbing between tables. A naive bot acts on a tight internal clock and produces a sharp spike.

The giveaway is not the average, since a bot can be tuned to a human-looking mean. It's the shape. A human histogram is wide and lumpy; a scripted one is narrow, or shows artificial regularity like quantization to fixed intervals. Plot enough decisions and the two separate cleanly:

Histogram comparing a broad human time-to-act distribution against a sharp scripted spike
Same decision spot, 200 samples. The human distribution is wide and right-skewed; the scripted one collapses into a spike.

Sophisticated bots add randomized delays drawn from a human-like distribution. Randomized delays rarely eliminate timing differences, because human timing is context-dependent , correlated with hand difficulty, stack depth, and prior action. Reproducing the marginal distribution is easy; reproducing the conditional structure (slow exactly when the spot is hard) is much harder, and mismatches there are themselves a signal.

Timing-analysis features:

Signal 2: Bet-sizing entropy

Solvers output precise, repeatable sizings (33% pot, 67% pot, 125% pot) and they hit them to the chip across thousands of hands. Humans round to convenient numbers. Their sizing drifts with mood and stack size, and they rarely reproduce the same exact sizing in the same spot every time.

Measuring the entropy of a player's sizing choices captures this. Two failure modes both look suspicious:

On a provably-fair site the recorded hand histories are tamper-evident, which makes this kind of long-run statistical aggregation more trustworthy: the data you're measuring is verifiably real, as covered in the note on provable fairness.

Signal 3: Multi-tabling and session patterns

Automation's economic point is scale, and scale leaves prints:

PatternWhy it flagsStrength
Table count beyond human capacityConsistent A-game across 16+ tables exceeds attainable human attentionHigh
Marathon sessions, no fatigue driftDecision quality and timing don't degrade over many hoursMedium
Hardware / client fingerprint reuseMany “players” share one machine signatureHigh
Headless or instrumented client artifactsAutomation hooks visible at the client layerHigh
Synchronized action across seatsSeveral accounts act on a shared clockMedium

None of these is conclusive alone. A strong human grinder can multi-table heavily, and shared hardware can mean a household, so these metrics feed a combined score rather than triggering action directly.

Combining weak signals

Detection teams use this architecture:

per-player features
    timing-shape divergence
    sizing entropy
    multi-table / fatigue metrics
    client + network fingerprints
      → weighted anomaly score
      → threshold → queue for human review
      → manual confirmation → action

Any single feature is noisy and beatable, but they are hard to beat simultaneously. A bot tuned to pass the timing test may give itself away on sizing entropy; one that randomizes both may still trip the multi-tabling or fingerprint checks. The final step is human: automated scoring decides where to look, not who gets penalized, which keeps false positives from hitting legitimate grinders.

The false-positive problem

The hard constraint on any detection system is not catching bots; it is not punishing humans. A skilled, study-heavy regular who multi-tables and plays close to GTO will look somewhat bot-like: tight sizing, fast accurate decisions, long sessions. Treating “plays well” as “plays automated” would gut the player base of exactly the customers a room wants. This is why mature systems weight mechanical impossibility signals (sub-human reaction times, shared client fingerprints, perfect cross-table synchronization) far above skill signals like good sizing. You can be a brilliant human; you cannot be in two places acting on the same millisecond.

The arms race

Detection and evasion change in response to each other. Countermeasures against detection signals produce further signals:

EvasionCounter-signal
Randomized human-like delaysCheck conditional timing: slow on hard spots, not just on average
Deliberate sizing “mistakes”Errors that never cost EV are themselves a tell
Rotating accounts and IPsHardware fingerprint and play-style clustering across accounts
Capping table count to look humanLower volume per account, but identical style across the cluster

The asymmetry favors the defender in the long run: the bot must look human on every axis at once and over thousands of hands, while the detector only needs one durable inconsistency. The more an operator tunes a bot to defeat one signal, the more it tends to distort another. A bot must mimic human play perfectly across the full feature set over a large sample, a difficult requirement.

Feature engineering, in practice

For anyone building one of these systems, the work is less about exotic models and more about engineering features that are robust to mimicry. A few principles that tend to hold:

The output is rarely a clean “bot / not bot” label. It is a ranked queue of accounts whose behavior is statistically improbable for a human, ordered by confidence, with the impossible-for-a-human signals (sub-perceptual reaction times, shared fingerprints, synchronized action) weighted heaviest because they are the hardest to explain away innocently.

What detection cannot do

Behavioral detection is statistical and probabilistic. It produces confidence, not certainty, and a sufficiently restrained bot (low volume, heavily randomized, single account, clean hardware) can stay under the threshold for a long time. Every step the bot operator takes to evade detection (fewer tables, more randomness, deliberate suboptimality) directly erodes automation's economic advantage. Detection doesn't need to be perfect; it needs to make undetectable botting unprofitable. Systems optimize for this more achievable goal.

What this means for crypto poker specifically

The detection problem on a provably-fair site is, mechanically, the same behavioral problem as on any poker room. The on-chain layer changes the trust in the data, not the method of the verdict. Verifiable hand histories provide cleaner, tamper-evident data, making detection statistics more reliable. Cryptography proves the deal; only behavioral analysis assesses the players.

Raul Moriarty
Raul Moriarty
Poker Software Expert & Communications Lead at Poker Bot AI

Writes about poker automation, game-integrity systems, and how detection actually works in practice, separating the cryptography from the behavioral side.