Find the Leaks Hidden Behind
Your PokerTracker Stats
A PokerTracker stat can look healthy while the hands that built it are still leaking EV.
A statistical deviation is not automatically a decision leak.
REQUEST EARLY ACCESS · ALPHA IN PREPARATION
REQUEST EARLY ACCESS · ALPHA IN PREPARATION
POST-SESSION ONLY · NO REAL-TIME ASSISTANCE
Why normal-looking PokerTracker stats can still hide leaks
A headline frequency can sit inside an ordinary-looking range. The aggregate number alone is incomplete. It does not reveal the decisions that built it.
Separate example from the sample review below.
FOLD TO 3BET
55.0%
LOOKS ORDINARY
NOT VISIBLE IN THE AGGREGATE
- Position
- Opponent action
- Hand class
- Decision composition
- Stack / context
- Which hands folded
- Which hands continued
- Whether the mistakes are expensive
The frequency is a triage signal. It is not proof that the play is sound, and it is not proof that a leak exists.
Statistical leak vs decision leak
A statistical signal is useful because it shows where to look. A decision leak is a repeating choice that can be shown with hands and confidence.
STATISTICAL SIGNAL
A number differs from a benchmark or expected range.
- Headline frequency
- Gap versus a reference
- Aggregate tracker stat
WHAT IT TELLS YOU
Where to investigate.
WHAT IT DOES NOT PROVE
Which specific decision is wrong.
DECISION LEAK
A repeating choice in a defined spot.
- Defined spot
- Repeating decisions
- Enough opportunities
- Evidence and confidence
- Enough expected impact to act
WHAT IT TELLS YOU
Whether a pattern is evidenced and material.
WHAT IT DOES NOT PROVE
That every unusual stat is a leak.
A statistical deviation is a candidate. It becomes a decision leak only after the hands are inspected.
Example: two players, same Fold to 3Bet, different mistakes
Two players can arrive at the same headline frequency through different underlying decisions. The aggregate stat cannot tell us which player is over-folding or over-continuing in a specific branch. The hands have to show that.
PLAYER A
FOLD TO 3BET
55.0%
UNKNOWN WITHOUT HANDS
- Position mix
- Hand classes
- Which continues were folds
- Which folds were continues
- Stack / action context
PLAYER B
FOLD TO 3BET
55.0%
UNKNOWN WITHOUT HANDS
- BB vs BTN?
- SB vs CO?
- 3-bet pot?
- Single-raised pot?
The filter questions are places to look — not known leaks for either player.
How candidate stats lead to filtered hands
A PokerTracker frequency starts an investigation. The next useful object is a filtered hand set in a defined spot — not a verdict from the headline number.
- PT4 STAT SIGNAL
- DEFINE THE SPOT
- FILTER RELEVANT HANDS
- COUNT OPPORTUNITIES
- INSPECT REPRESENTATIVE DECISIONS
- CANDIDATE PATTERN
FILTER DIMENSIONS
- POSITION
- STREET
- POT TYPE
- ACTION CONTEXT
- DECISION FAMILY
A candidate pattern is a place to inspect. It is not yet a verified leak.
How decisions are verified
A candidate becomes useful only when representative decisions repeat and the sample in that family can support a claim. Structured evidence comes first. AI may explain that evidence; it does not invent poker truth.
- OPPORTUNITIES
How often the defined spot actually occurred.
- RELEVANT DECISIONS
The choices inside that family.
- REPRESENTATIVE HANDS
Compact examples of the repeating branch.
- CONFIDENCE
Assigned to the finding, not the whole database.
- MATERIALITY
Whether the estimated impact is worth study time.
- REFERENCE
A benchmark or strategy reference only where one is supported.
Not every spot has a solver line. A missing reference is not permission to invent one.
Canonical examples of the repeating spot — not a full replay of every hand, and not from the aggregate Fold to 3Bet profile above.
hh-river-01
BB · Single-raised pot
Called a large river bet after missing
preflop / flop / turn / river
Similar weak bluff-catchers are being continued too frequently in this branch.
hh-river-02
CO · 3-bet pot
Called river after a missed draw
preflop / flop / turn / river
Ace-high with no improvement after a passive turn — the same branch, a different runout.
How leaks are ranked
Priority considers recurrence and opportunity count, estimated impact, confidence, and whether the pattern is trainable. More frequent is not automatically higher priority.
- Recurrence / opportunity count
- Estimated impact
- Confidence
- Trainability / actionability
This comparison is from a separate illustrative PokerTracker review — not the aggregate Fold to 3Bet example above.
PRIORITY #1
River Bluff Catching
−0.47 bb/100
RELEVANT DECISIONS
186
412 opportunities
HIGH CONFIDENCE
PRIORITY #2
BB vs BTN Defence
−0.31 bb/100
RELEVANT DECISIONS
940
2,104 opportunities
HIGH CONFIDENCE
This is not a single frozen equation. Frequency is one input, not the ranking.
Why you shouldn't fix every deviation
Most deviations are not worth study time. A signal can resolve to a priority, to no action this cycle, or to insufficient evidence.
A deviation may be real and still not a priority now.
Too few opportunities to make a strong claim.
Sample verified PokerTracker leak
This is an illustrative PokerTracker 4 finding, not customer data and not a live result from this page. It is a separate example from the aggregate profile above.
PRIORITY
River Bluff Catching
REPEATING PATTERNCalling frequency stays high versus large river bets after the hero has missed.
- ESTIMATED IMPACT
- −0.47 bb/100
- RELEVANT DECISIONS
- 186
- OPPORTUNITIES
- 412
- CONFIDENCE
- HIGH
A signal can resolve to priority, no action, or insufficient evidence
The useful result is a short list of what to train, what to leave, and what the sample cannot support yet.
River Bluff Catching
WORK ON THIS
UTG open frequency
A deviation may be real and still not a priority now.
NO ACTION
Turn probes in position
37 opportunities
Too few opportunities to make a strong claim.
INSUFFICIENT EVIDENCE
From PokerTracker leak to training goal
Once a priority is verified, train that one spot, play new hands, then re-measure the same filter. Training is post-session. There is no live table prompt and no HUD coach.
- VERIFIED PRIORITY
- TRAIN ONE SPOT
- PLAY NEW HANDS
- RE-MEASURE SAME FILTER
River Bluff Catching
Train fold/call decisions versus large river bets on missed-draw textures, then re-measure the same filter on new hands.
PokerTracker stays the tracker. Export the data you already have.
Use PT4 exports and hand histories. Open Analyze. You do not need to move or replace the database.
The current Analyze preview only checks whether the selected file type looks supported. It does not parse or analyze the file yet. The selected file stays on this device during the current preview. Full automated leak analysis is not live.
Full database analysis is not live yet. We're opening early access to players with real PT4, HM3 and hand-history databases.
Early users may be asked to share anonymized exports manually while the automated analyzer is being built.
Questions
What is the difference between a statistical leak and a decision leak?
A statistical leak is a number that differs from a benchmark. A decision leak is a repeating choice in a defined spot, shown with hands and confidence.
Should I fix every deviation?
No. Most deviations are not worth study time. The point is to rank a few issues by impact and evidence.
See what PPC adds to PokerTracker
Keep the tracker. Add a performance layer on the database you already built.
POKERTRACKER DATABASE REVIEWSee what a PT4 review returns
Scope, deliverables, evidence, priority, and follow-up.
ARE MY POKER STATS GOOD?See how aggregate stats are interpreted
Editorial triage for common tracker frequencies — not a leak verdict.
METHODOLOGYSee how evidence is required
Confidence, sources of truth, and why insufficient evidence is valid.
Don't fix the stat.
Verify the decision.
Your PokerTracker stats tell you where to look. The hands tell you whether there is a leak worth study time.
Post-session analysis only. No real-time assistance.