Every stat is calculated from two counts: how often the player had an opportunity to act, and how often they took that action. The percentage is simply action ÷ opportunity.
426 ÷ 600 = 71.0% — all stats work the same way. Filtering by stake, format, or other criteria simply reduces the hands included in both counts.
Illustrative example
Illustrative example
A small amber number beside a stat, such as 48.8₁₇, means the stat is based on only 17 opportunities. This indicator appears below 30 opportunities.
Small samples can produce misleading percentages. For example, folding 2 out of 3 times gives a 67% fold rate, but three opportunities are not enough to say much about the player's usual behavior.
A dot means the situation never came up under the current filters, so there is nothing to show. That is different from 0%, which means the spot did occur and the player never took the action.
Poker stats settle slowly, and much more slowly than most people assume. We measured it rather than guessing: Limp is trustworthy after about 110 opportunities, but PFR needs 2,600 and VPIP 5,400… At 100 opportunities VPIP is still typically 9 points out.
If a number looks extreme, check its subscript before believing it.
Results — what they won: the winnings graph, plus breakdowns by stake and by month.
Preflop — everything before the flop. The starting-hand grid sits at its foot, since it is this tab's own stats cut by holding.
Postflop — flop, turn and river, following the hand's own timeline.
All-in — how often the stacks go in, on which street, and what with. It closes the timeline the two tabs before it walk through.
Activity — when they play, and their recent showdowns.
Everything above the tabs — the headline strip and the verdict — applies to all five.
Illustrative example
Every number on the page obeys the filter card. Narrowing to one game, stake, format, pot type or period gives you a truer read on how somebody plays in that specific spot — at the cost of sample size, so watch the subscripts appear as you filter.
Narrowing gives a truer read on one spot, at the cost of sample — watch the subscripts appear as you filter. A padlocked stake is one above your plan.
Illustrative example
No-Limit Hold'em, Pot-Limit Omaha and its variants. Omaha players see far more flops by the nature of the game, so never compare an Omaha VPIP against a Hold'em one.
The blind level, named as players name it: NL200 is a $2 big blind. Only stakes this player has actually played are offered.
A padlocked stake is one above your plan — its hand counts are shown but its stats need an upgrade.
How many players were dealt in. HU is heads-up (2 players), SH is short-handed (3–7), FR is full ring (8–9). People play very differently at each — a 45% VPIP is loose at full ring and tight heads-up.
Limped, single-raised, 3-bet or 4-bet+ pots. This re-cuts every postflop number, which makes it the deepest filter on the page — but it describes the HAND, not the player, and it makes some preflop stats circular: PFR reads 0% in limped pots for everyone alive, because a limped pot is defined as one where nobody raised. The page shows a warning saying exactly this whenever a pot type is selected.
All time, or a recent window. Each option is labelled with the hands it holds, and windows that would show the same data as a shorter one are dropped rather than offered twice.
The verdict panel keeps comparing against the full population pool and discloses the window.
Once you scroll, the tab bar pins to the top and repeats your active filters as small chips — so a table of numbers is never on screen without its scope. Clicking the chips scrolls back to the filter card.
The top of every player page: who they are, and sixteen tiles that characterise them fastest — money, the core preflop numbers, aggression, and the showdown trio.
Hovering any tile shows a one-line definition; this section is the longer version.
A tile printed in amber is one the verdict counts as a leak, and its tooltip carries the pool median behind that call.
Illustrative example. The amber tiles are the ones the verdict below counts as leaks — far enough from this player’s pool, on enough hands, to be worth playing against. Hovering a tile gives its definition; hovering an amber one also gives the pool median it is being judged against.
example
How many hands we have on record for this player under the current filters. This is your confidence meter for everything else on the page.
Under a few thousand hands, treat postflop numbers as impressions rather than reads.
example
Total money won or lost, in the currency of the games played. Green is profit, red is loss.
How it is worked out
sum of every pot won, minus every chip put in
example
Win rate, normalised so stakes can be compared directly. The single best measure of whether somebody is actually good. Greyed with a ± figure when the sample is too small to tell a winner from a loser.
How it is worked out
Rough guide above 0 is a winner · 2–8 is a solid regular · below −10 is a big loser
Safe to read across mixed stakes and currencies, because dividing by the big blind already removes them.
example
The same win rate with all-in luck averaged out: every all-in with cards to come is scored by its odds instead of the one card that fell. Read next to bb/100 — a player far above their EV has been running hot, not playing well.
How it is worked out
Only hands that actually went all-in before the river can differ, so on most players the two rates are close.
example
How often the player chooses to play a hand at all, rather than folding. The headline measure of loose versus tight.
How it is worked out
Rough guide under 18% tight · 20–28% typical regular · over 35% very loose
Posting the blinds does not count — that money is forced, not chosen. Nor do walks: when everyone folds to the big blind, the action never comes back and the hand asks no question, so it is left out of the denominator entirely. That is about one hand in fourteen at a full table, and counting them would measure how often somebody was dealt the big blind rather than how they play.
example
How often they raise before the flop. Compared against VPIP it tells you whether a player takes initiative or just calls along.
How it is worked out
Rough guide 16–24% for a regular. A gap to VPIP under 5 points is aggressive; a gap over 10 points means a lot of calling.
example
How often they re-raise somebody else's raise before the flop. A strong signal of aggression, and one of the first numbers experienced players look at.
How it is worked out
Rough guide 4–6% tight · 7–11% typical regular · over 13% very aggressive
Called '3-bet' because the blind counts as the first bet and the open raise as the second.
example
When they raise and somebody re-raises, how often they give up. High numbers mean you can profitably re-raise them; low numbers mean they fight back.
How it is worked out
Rough guide 45–60% is normal. Over 70% is exploitable — re-raise them more.
example
How often they call somebody else's raise having put no money in yet. 'Cold' because they are entering the pot cold, with money already committed by others. High cold-call rates usually mean a passive range full of hands that should have folded or re-raised.
How it is worked out
The blinds are excluded by definition — they already have money in, so calling from there is a different decision entirely.
example
Re-raising after one player has raised and at least one other has called. The name comes from squeezing the caller, who is now trapped between two aggressors.
How it is worked out
example
How often, across everything they are dealt, they open a pot and then 4-bet it. Not the same as the 4Bet stat on the Preflop tab — that one asks how often they 4-bet when the chance arose, this one folds in how often the chance arises at all, which is what tells you how strong a 4-bet from them is.
How it is worked out
Rough guide 1–3% of all hands. Under 1% means a 4-bet from them is close to aces or kings.
Not a fraction of anything, despite the name — it is RFI% × 4Bet%, the same formula Hand2Note uses. Read it as 'of all the hands they are dealt, how often do they open AND then 4-bet', which is why it lands near 1–3% while the Preflop tab's 4Bet column, asking 'when they could 4-bet, how often did they', reads in the teens. It is NOT the width of their 4-betting range: a true width would need their shown cards, which exist for only a handful of 4-bets even on a large sample.
example
Postflop bets and raises per call — a ratio, not a percentage. Reads together with AFq below: a player who folds relentlessly but raises when they do continue shows a high AF and a low AFq, so neither replaces the other.
How it is worked out
Rough guide under 1.5 passive · 2–3 typical · over 4 very aggressive
COUNTS ACTIONS, where AFq counts streets — and that is deliberate on both sides, not an inconsistency. AF asks how much aggression there was per call, so a flop bet followed by a re-raise is genuinely two aggressive acts. AFq asks how often the player chose aggression when it was their turn, which is a question about streets. Both match Hand2Note under their own rule.
example
Of the streets where they did anything at all, what share did they play aggressively — betting or raising rather than checking, calling or folding. A single summary of postflop style, and unlike AF it is a true percentage.
How it is worked out
Rough guide 35–50% typical. Under 30% is passive, over 60% is hyper-aggressive.
COUNTED IN STREETS, NOT ACTIONS, and that is the whole difference from AF above. A player who bets the flop, gets raised and re-raises has been aggressive on one street, not twice — counting the actions would let that single flop vote twice over. Checking counts as acting; so does folding. Measured against Hand2Note, counting actions read about nine points low.
example
Of the pots where they saw a flop, how often they ended up winning — whether by showdown or by everyone folding. A good single measure of postflop effectiveness.
How it is worked out
Rough guide 43–48%. Over 50% usually means an aggressive player taking pots away.
example
Once they have seen a flop, how often they stay in all the way to a showdown. High means curious or stubborn; low means they give up easily after the flop.
How it is worked out
Rough guide 24–30% typical. Over 35% is a calling station; under 20% folds too much after the flop.
example
Of the showdowns they reach, how many they win. Reads together with WTSD: a low WTSD with a high W$SD means they only continue with real hands.
How it is worked out
Rough guide 48–55%. Below 45% alongside a high WTSD is the classic calling-station pattern.
Illustrative example
A small arrow beside a tile means the last 30 days measurably differ from the lifetime figure — measurably, not just visibly: the arrow only appears when the gap clears two standard errors and at least a full point, on enough recent sample to mean something. Hover it for both numbers. No arrow means no proven drift, not 'no change'.
Effective stack 40bb — the deepest opponent, not this player’s 200.
Illustrative example
Not what somebody bought in for. The effective stack is the smaller of their own chips and the deepest opponent's, as the hand is dealt, in big blinds — the most money that can actually change hands.
Sitting with 200bb at a table where nobody else has more than 40 is a 40bb game, and every decision in it is a 40bb decision.
So the figure describes the games a player sits in as much as it describes the player: someone who always buys in full but plays where everyone is short records low effective stacks all session.
Illustrative example
The badge beside the name is the player's median effective stack over the hands currently in view, and it prints the figure itself. Three bands: short stack under 50bb, mid stack 50–80bb, deep above 130bb. About seven players in ten sit between 80 and 130 and carry no badge.
Only 'short stack' is tinted, because it is the only one with measured behaviour behind it — under half a buy-in this pool plays looser and more passively, the profile of somebody who never rebuys rather than of a short-stack specialist. 'Mid stack' and 'deep' are descriptions rather than findings; deep tables simply mean implied odds and postflop skill count for more.
Because it is a median over the hands in view, it obeys the filter card — a player can be short at one stake and not at another.
Signed in, you can add any player to your watchlist and keep private notes on them — reads, tendencies, table talk. Notes are visible only to your account and are never part of the stats.
The interpretation layer: this player's numbers compared against every other player in the same games, and the differences that are large AND proven enough to act on.
It sits under the headline strip because it is a reading of those numbers, not a replacement for them.
Compared against 3,214 players in NL25 6-max — matched per room, all pots.
Fold vs 3Bet 71.0% (+13.2 vs pool) — folds to re-raises far too often, so 3-bet them relentlessly.
Pool: NL25 6-max · 3,214 players · most unusual first
Illustrative example — the verdict panel: headline verdict, distance-from-pool table, named leaks.
Illustrative example
Winning, losing, break-even — or 'not enough hands', which is the honest answer more often than people like. A win rate only earns a colour when it sits at least two standard errors from zero; inside that, luck and skill are indistinguishable and the page says so.
Compared against 3,214 players in NL25 6-max — matched per room, all pots.
Illustrative example
The comparison group: players in the same game, format, stake — and, where the sample allows, the same room and pot type, because pools genuinely differ across both. The panel names the pool it used and how many players are in it.
Illustrative example
Each stat shown as player vs pool median, ordered by how unusual the player is — measured in interquartile ranges (IQR), the spread of the middle half of the pool. Distance is a description, not an accusation: being unusual is only a problem when it loses money.
A named leak must clear two bars at once: at least one IQR from the pool in a direction that costs money, AND at least two standard errors on the player's own sample with a minimum of 30 opportunities. Every leak comes with what to do about it.
The panel states which hands it compared (the count can be smaller than the strip's, if part of your filter cannot be matched to a pool), which filters it had to ignore, and how many years of history the pool spans. If those lines are missing nothing was ignored.
A second comparison, sitting under the verdict: not what everyone at this stake does, but what the players who actually make money do differently — and which side of that difference this player is on.
It is the most useful panel on the page and the easiest to over-read, so the last two entries here are a warning worth reading once.
| Stat | The field | Winners | Difference | This player | |
|---|---|---|---|---|---|
| Fold vs 3Bet | 58.1 | 44.2 | −13.9 | 63.2 | with the field |
| AFq | 32.6 | 37.9 | +5.3 | 26.4 | with the field |
| CBet Flop | 57.8 | 62.4 | +4.6 | 66.1 | with the winners |
| 3Bet | 6.8 | 8.9 | +2.1 | 8.6 | with the winners |
Illustrative example — the winner panel: the field, the winners, the difference, and this player.
The column that matters is the difference between them, because that is the only thing on the page that points at what pays rather than at what is merely common.
A stat where winners and the field agree tells you nothing about making money, however unusual this player looks on it — which is why such rows are left out entirely.
A winner is a player whose EV-adjusted win rate is positive over HALF their hands at this stake, and every number shown is measured on the OTHER half. That sounds fussy and it is load-bearing.
If you pick winners and measure them on the same hands, the luck that made them look like winners leaks into the stats meant to explain it.
How winners are proven, and what the split-half test measured →
They are different groups and the difference is large. Winners play about twice as many hands as the typical player, and high-volume players behave differently from everyone else regardless of results.
So the field shown here is re-weighted to match the winners' own volume profile — a like-for-like comparison.
The shape above the table: this player, the winners and the field on the same axes, so the profile is one read rather than eight comparisons.
The rings are PERCENTAGE POINTS AWAY FROM THE FIELD, one shared scale for every spoke, which is why the field is a perfect hexagon sitting mid-radius — outward is more than the field does, inward is less, and both directions have to exist.
A stat that swings 20 points therefore dwarfs one that swings 2, and that is the finding rather than a distortion: giving each axis its own scale makes every stat look equally dramatic, which is the lie radar charts are usually accused of. It also removes the numbers, since with a different unit per spoke there is nothing you could print on a ring.
Each spoke carries a shaded band: the range a player who genuinely matched the field could land in by luck alone, on this player's sample. Inside the band means no difference has been shown, however far from centre the line looks.
The band is not the textbook formula — hands from one session are correlated, so the real error is larger, and by how much was measured per stat rather than assumed.
Axes are yours to choose: preflop only, postflop only, everything, or any set of three or more. Dropping an outlier rebinds the scale to what is left, which is the escape hatch for the shared-scale trade-off above.
Biggest gap between winners and the field first. That is deliberately not the same ranking as the verdict panel above, which orders by how unusual THIS player is. The two panels answer different questions — 'is this player odd?' and 'is this player doing what pays?' — and sorting both the same way would quietly turn one into a worse copy of the other.
These numbers are necessary, not sufficient. Tested on a large sample at one stake, players who were provably losing sat about twice as far from the winner profile as players who were provably winning — so the comparison genuinely carries signal.
But the exceptions are real and they are not rare curiosities: one player with 57,739 hands, losing nine big blinds per hundred at a margin far too large to be bad luck, matched the winner profile MORE closely than the average winner did. VPIP, PFR, 3-bet, won-when-saw-flop — all of them sat where a winner's sit. He was still losing steadily.
A frequency says how OFTEN, never WHICH. Two players can both put money in preflop 22% of the time — a winner's number exactly — while one of them is playing the best 22% of hands and the other the worst.
Treat the panel as a direction to look in, never as a scorecard you can pass.
Every stat on this site is a FREQUENCY: how often someone does a thing.
Frequencies describe the shape of a strategy, never the quality of the decisions filling it.
Two players both bet the flop 60% of the time; one picks the right 60% and the other does not, and no frequency will ever separate them.
What actually decides that — the size of the bets, which specific hands were chosen for each action, which tables and seats were sat in, and whether they kept playing after losing three buy-ins — is either not recorded in hand histories at all or is only visible on the small share of hands that reach a showdown.
So the panel can tell you where a player's habits differ from the people who win. It cannot tell you whether they are making good decisions inside those habits.
When judging a player against these numbers, look at their WORST stat rather than their average. Measured on the same sample, the average distance from the winner profile barely separated losers from winners, while the single largest deviation separated them nearly three times as sharply.
It makes sense once said aloud: averaging one genuine leak against eleven stats played perfectly well hides exactly the thing you were looking for.
This is why the verdict panel names individual leaks and never reports an overall score.
The stats that discriminated most in that test were AFq — how often a player bets or raises after the flop instead of checking and calling — followed by WTSD, how often they keep paying to see a showdown.
What this player has won: the winnings graph over time, then the same record broken down by stake and by month. The landing tab — but the graph is one of three things here, which is why the tab is not called 'Graph'.
| Stake | Hands | Won | bb/100 |
|---|---|---|---|
| NL25 | 121,406 | −$2,046.18 | −6.7 |
| NL10 | 21,884 | −$262.61 | −12.0 |
| NL50 | 5,630 | −$73.93 | −2.6 |
Illustrative example. The EV line above the actual one means mildly unlucky — nowhere near enough to explain a loss this size, which is what makes the verdict a verdict.
Actual money won or lost, accumulating hand by hand. This is what really happened.
How it is worked out
running total of money won
What the player would have won on average if every all-in had been resolved fairly instead of by one specific card. If this line sits above Net Won, they have been unlucky in all-in pots; below, lucky.
How it is worked out
running total of expected winnings
Only hands that actually went all-in with cards to come can differ, so the label states how many hands were adjusted. If that count is small, the two lines mean nearly the same thing.
Money won only in hands that reached a showdown — where cards were turned over and the best hand took it.
How it is worked out
running total of money won in hands that reached showdown
Money won in hands that ended before a showdown, because everybody else folded. Together with the line above, this splits winnings into 'made the best hand' and 'made them fold'.
How it is worked out
running total of money won in hands with no showdown
A classic aggressive winner shows this line rising and the showdown line falling — they win by betting, and pay to see showdowns. The opposite pattern usually means a passive player.
Switches the vertical axis between money and big blinds. Big blinds is the honest view when the filter spans several stakes, since a hand at $5/$10 would otherwise dwarf a hundred hands at $0.05/$0.10.
Under the graph, the win rate is split into bb/100 won at showdown and bb/100 won without one — the same story as the two graph lines, as numbers you can read off directly.
The record per stake: hands, win, bb/100, VPIP, PFR and when they last played it. A player can be a clear winner overall and a clear loser at one stake — check the game you actually face them in. Padlocked rows are stakes above your plan: their hand counts are real, their results need an upgrade.
The last months of hands, winnings and bb/100. Steady volume suggests a regular; sporadic bursts usually mean a recreational player.
The most recent hand we hold for that row. A regular who stopped appearing months ago may have moved sites, moved stakes, or quit.
Everything before the flop, in three sections: the seat-by-seat table (position matters more than almost anything else in poker — nearly every number should climb toward the button), the fold/call/raise meters for the spots where they faced aggression, and the two position grids. Columns already defined in the header section — VPIP, PFR, 3Bet, Fold vs 3Bet, Cold Call, Squeeze — are not repeated here.
| Stat | UTG | MP | CO | BTN | SB | BB | All |
|---|---|---|---|---|---|---|---|
| Hands | 1,204 | 1,187 | 1,232 | 1,241 | 1,198 | 1,220 | 7,282 |
| Rangeshare of all hands dealt | |||||||
| VPIP | 18.2 | 21.4 | 26.8 | 38.6 | 41.2 | 32.9 | 29.8 |
| PFR | 14.1 | 16.8 | 19.9 | 26.4 | 17.2 | 8.6 | 17.0 |
| Limp | 0.8 | 1.2 | 2.1 | 4.0 | 12.6 | · | 3.4 |
| Facing an open | |||||||
| Cold Call | 6.1 | 8.4 | 11.2 | 16.8 | 9.4 | 18.2 | 12.4 |
| 3Bet | 4.2 | 4.8 | 5.9 | 7.4 | 6.1 | 5.2 | 5.6 |
| Their raise got re-raised | |||||||
| Fold vs 3Bet | 78.4 | 74.2 | 71.8 | 66.4 | 69.1 | 62.8 | 71.0 |
Illustrative example — preflop by position, the facing-aggression meters, and the two position grids.
When everybody has folded to them, how often they come in with a raise rather than folding. The purest measure of positional awareness, because the situation is identical every time apart from the seat.
How it is worked out
Rough guide Rough guide, 6-max: around 15% under the gun rising to 45%+ on the button. A flat line across seats means they are ignoring position.
How often they enter a pot by just calling the big blind instead of raising. Widely considered weak play, so a meaningful limp rate marks an inexperienced opponent.
How it is worked out
Rough guide regulars are under 1%. Anything above 5% is a strong sign of a recreational player.
Having limped and then faced a raise, what they do next. Limp/Call is the default recreational line; Limp/Raise from a non-trivial sample is almost always a monster — the old 'limp-reraise with aces'.
How it is worked out
Only meaningful for players who limp at all — for a 0.5% limper these three rest on almost nothing.
When they re-raise and get re-raised again, how often they raise once more. See also the header's 4Bet Range, which measures the same action against a different denominator.
How it is worked out
This is the OPENER's 4-bet, which is what the term means everywhere else. Re-raising somebody else's 3-bet without having opened is a cold 4-bet — a different decision, and not counted here. The header's 4Bet Range does count both, because a range is about which hands they 4-bet rather than how often they do it from one seat.
Having re-raised, how often they back down to a 4-bet. High numbers mean their 3-bets can be attacked cheaply.
How it is worked out
Raising yet again over a 4-bet — usually an all-in at normal stack depths. So rare that the number is mostly interesting for existing at all.
How it is worked out
Expect a subscript almost forever — this is one of the rarest spots in the game.
Raising from the cut-off, button or small blind when it is folded to them — seats where a raise is often aimed at simply taking the blinds rather than having a strong hand.
How it is worked out
Rough guide 30–45%. Above 50% means they are attacking the blinds relentlessly.
Sitting in a blind facing one of those steal raises, how often they surrender. This is the other side of the Steal stat. Read it from the SB vs Steal / BB vs Steal meters below the table, not from the table itself — see 'Where blind defence lives'.
How it is worked out
Rough guide 65–80%. Over 85% and stealing from them is close to free money.
Instead of folding a blind to a steal, re-raising back at them. Players who never do this can have their blinds attacked without risk. This is the Raise segment of the vs-Steal meters.
How it is worked out
The third answer to a steal: not folding, not re-raising, but calling to see a flop from the blind. The Call segment of the vs-Steal meters, derived as the remainder — the three segments always add to 100%.
How it is worked out
The three bars under the table, one per blind plus the two added together. Where Fold vs Steal, Call vs Steal and 3Bet vs Steal above are the three SEGMENTS, this is the spot they divide — every time a steal seat opened and this player was sitting in that blind. Reading the whole bar rather than one segment is the point: folding 70% means something different when the other 30% is all raises than when it is all calls.
How it is worked out
Total is the two blinds ADDED, not averaged: the counters are additive, so averaging two percentages would weight a thin SB sample like a busy BB one.
They opened, someone re-raised: this is what they did about it, as one bar. Fold vs 3Bet in the table above is this bar's fold segment; the rest of it is the half a single percentage cannot show — whether the hands they keep are calls or 4-bets.
How it is worked out
They 3-bet, it came back at them. The same three-way split one level deeper, and the rarest spot on the preflop tab — expect a thin sample for a long time.
How it is worked out
Having squeezed, how often they then fold to further aggression — a measure of whether their squeezes are real hands or pure pressure.
How it is worked out
A rare spot inside an already rare spot, so this stays low-sample almost indefinitely.
Raising over the top of a limper, usually to get heads-up with a weak player rather than let the pot go multiway. Strong players do this a lot when limpers are present.
How it is worked out
A grid: rows are the seat the player was sitting in, columns the seat of whoever raised. The cell at row BB, column BU tells you how often they called from the big blind against a button raise.
How it is worked out
Mostly triangular, because the raiser normally acts before the caller. Cells above the diagonal are not errors — a limper who then faces a raise from a blind is a genuine spot. Blinds are excluded from the rows by cold call's own definition.
The same grid applied to re-raising: not just how often they re-raise, but from where and against whom. A player who only re-raises late seats against late raisers is playing a standard, position-aware game; one who re-raises from anywhere against anyone is either very good or very reckless.
How it is worked out
The Tot column is summed from the raw counts, never averaged across the row — averaging percentages would let a 3-hand cell count as heavily as a 3,000-hand one.
Stats run down the side and seats across the top, which is the opposite of most trackers. With eighteen stats against at most eight seats, the usual orientation scrolls sideways forever and you lose the row label; this way it scrolls vertically like every other table here, and comparing one stat across seats — the thing people actually do — reads along a row. Total is the last column, separated from the seats.
Rows are grouped by the SITUATION they measure, because that is what makes the denominator legible: Range (share of all hands dealt), Taking the initiative (nobody has raised yet), Facing an open, Their raise got re-raised, and After open-limping. Only stats inside the same band are comparable with each other, and two of the bands are exhaustive fold / call / raise splits that add to 100%.
Five bars, each one whole decision split into fold / call / raise, so every bar adds to 100% and each cell of the meter is 2% of the spots: SB vs Steal, BB vs Steal, their Total, then vs 3Bet and vs 4Bet. The blind-defence rows and the re-raise rows are separated by a rule because they are different decisions — defending cheap money versus answering a re-raise.
Fold vs Steal, Call vs Steal and 3Bet vs Steal are deliberately kept OUT of the position table and shown only as meters. A three-way decision drawn as one bar beats three separate percentages a reader has to add up — and the table's versions were the weaker pair, because 3Bet vs Steal is measured over a wider set of spots than the other two, so the three did not even sum to 100.
At the foot of the Preflop tab: the same six stats one more time, cut by HOLDING instead of by seat. VPIP, PFR, 3Bet, 4Bet, Cold Call and Limp are columns of the table at the top of that tab and squares of this grid, off the identical counters. Pairs run down the diagonal, suited hands above it, offsuit below. Read the caveat below before drawing conclusions — this is the one panel on the tab built on shown cards rather than on every hand.
| AA | AKs | AQs | AJs | ATs | A9s | A8s | A7s | A6s | A5s | A4s | A3s | A2s |
| AKo | KK | KQs | KJs | KTs | K9s | K8s | K7s | K6s | K5s | K4s | K3s | K2s |
| AQo | KQo | QJs | QTs | Q9s | Q8s | Q7s | Q6s | Q5s | Q4s | Q3s | Q2s | |
| AJo | KJo | QJo | JJ | JTs | J9s | J8s | J7s | J6s | J5s | J4s | J3s | J2s |
| ATo | KTo | QTo | JTo | TT | T9s | T8s | T7s | T6s | T5s | T4s | T3s | T2s |
| A9o | K9o | Q9o | J9o | T9o | 99 | 98s | 97s | 96s | 95s | 94s | 93s | 92s |
| A8o | K8o | Q8o | J8o | T8o | 98o | 88 | 87s | 86s | 85s | 84s | 83s | 82s |
| A7o | K7o | Q7o | J7o | T7o | 97o | 87o | 77 | 76s | 75s | 74s | 73s | 72s |
| A6o | K6o | Q6o | J6o | T6o | 96o | 86o | 76o | 66 | 65s | 64s | 63s | 62s |
| A5o | K5o | Q5o | J5o | T5o | 95o | 85o | 75o | 65o | 55 | 54s | 53s | 52s |
| A4o | K4o | Q4o | J4o | T4o | 94o | 84o | 74o | 64o | 54o | 44 | 43s | 42s |
| A3o | K3o | Q3o | J3o | T3o | 93o | 83o | 73o | 63o | 53o | 43o | 33 | 32s |
| A2o | K2o | Q2o | J2o | T2o | 92o | 82o | 72o | 62o | 52o | 42o | 32o | 22 |
Illustrative example. Suited hands above the diagonal, offsuit below, pairs on it. This player reaches showdown with a visibly wide range — small suited gappers and weak offsuit aces that most winning players fold before the flop.
Each square is one starting hand — AA in the top-left corner, 22 in the bottom-right, AKs above the diagonal and AKo below it. The shading shows how often they took the selected action holding that hand.
How it is worked out
Switches which action the grid is showing — VPIP (played the hand at all), PFR, 3Bet, 4Bet, Cold Call or Limp. Switching is instant; all six are already loaded.
Opacity carries confidence. A square built on eight or more observations is solid; one built on a single hand is nearly transparent. This stops a 100% resting on two hands from looking like a 100% resting on two hundred.
Hole cards only appear in a hand history when they were actually shown, which is roughly 4–6% of hands, and those are overwhelmingly hands that went to a showdown. So the frequency maps over-represent hands people play to the river and under-represent hands they fold early or win without showing. They are a genuine hint about which hands turn up in a range, but they are not the range, and no amount of extra data fixes the bias — only reduces the noise around it.
One situation escapes it entirely: an all-in. The pot went in, so the hand got shown — 95.5–96.2% of all-in hands have known hole cards, against 5.8% of hands overall. So the All-in tab's holding maps are built on nearly their whole population where these are built on a twentieth of theirs. If you want a range you can lean on, that is the one.
'AKs' means ace-king of the same suit, 'AKo' means different suits. Suited hands are meaningfully stronger because they can make flushes, which is why the grid separates them.
Everything after the flop, ordered by who is doing the betting: Betting (their aggression when they have the lead), Leading & stabbing (their aggression when somebody else has it), Facing a c-bet (what they do when a bet comes at them), and Showdown (how hands end). Each table has a Flop / Turn / River row — reading DOWN a column shows how behaviour changes as the pot grows and the bluffs run out. The small counts beside the street names are how many flops, turns and rivers the numbers rest on.
| vs c-bet | Fold | Fold IP | Fold OOP | Call | Raise | Fold 3BP |
|---|---|---|---|---|---|---|
| Flop6,140 | 58.4 | 54.1 | 62.7 | 33.8 | 7.8 | 61.2 |
4,912 spots | ||||||
| Turn3,880 | 62.1 | 57.6 | 66.3 | 31.4 | 6.5 | 64.8 |
2,148 spots | ||||||
| River2,410 | 66.9 | 62.2 | 71.4 | 27.4 | 5.7 | 69.5 |
1,036 spots | ||||||
Illustrative example — the Betting table above Leading & stabbing, Facing a c-bet and the Showdown box.
When they raised before the flop and then bet the flop as well, continuing their story. Nearly every strong player does this often; the interesting question is what happens on later streets.
How it is worked out
Rough guide flop 55–70% · turn 40–55% · river 35–50%
THE THREE STREETS ARE A CHAIN. There is a turn c-bet spot only for the player who c-bet the flop, and a river one only for the player who c-bet the turn — and only while they still hold the lead.
A c-bet that gets RAISED hands the initiative to the raiser, so a player who calls that raise and then bets the next street is leading, not continuing, and it does not count here; re-raising keeps the chain, because they are the aggressor again.
That is also why the flop figure is the steadiest of the three and the river one rests on very few hands.
The same stat split by whether they act last (in position) or first (out of position). Almost everyone c-bets more in position, because acting last is easier. A player who c-bets the same either way is usually not thinking about position.
How it is worked out
How often they bet into a pot nobody has bet into yet, when they were not the preflop raiser. This is initiative taken rather than inherited.
How it is worked out
Having bet and then been raised, how often they fold. Very high numbers mean raising their bets works as a bluff. It sits next to Bet because it is what happens to a bet — the follow-up, not a separate subject.
How it is worked out
The denominator is EVERY bet of theirs that got raised — c-bets and stabs included — so it is wider than the Bet column beside it, which counts only firing into an unopened pot.
Raising a bet they did not check to — somebody bet at them and they raised on the spot. One of the two halves of XR & R.
How it is worked out
Checking first and then raising when somebody bets — the other half of XR & R, and the more deceptive one: it invites a bet in order to punish it.
How it is worked out
The two raising lines combined — how often they raise a bet at all, whichever way they got there. Usually the strongest single signal of postflop aggression, and the figure most trackers quote.
How it is worked out
Rough guide under 8% is passive · over 15% is very aggressive
How often they fold when the preflop raiser bets at them — the Fold column of the 'Facing a c-bet' table, with IP / OOP splits beside it and Call and Raise completing the decision. One of the most directly exploitable numbers on the page: a high figure means betting at them works regardless of your cards.
How it is worked out
Rough guide 40–55% on the flop. Over 60% means bet at them relentlessly.
Turn and river hang off the same chain as CBet above: you can only face a turn c-bet from the player who c-bet the flop and still had the lead when that street closed. The same goes for Call and Raise beside it.
The bar under each street of the Facing a c-bet table: the same decision the Fold, Call and Raise columns show, drawn as one shape so the three read as one choice rather than three numbers to add up.
How it is worked out
Turn and river need parser_version 11, which added their raise counters. A street with no re-parsed rows shows 0 opportunities — 'never faced this spot' — rather than a bar claiming they never raise.
The same fold-to-c-bet number, but only in pots that were re-raised before the flop. These pots are bigger and both ranges are stronger, so behaviour often changes sharply — many players fold far too much here.
How it is worked out
Naturally a small sample — 3-bet pots are uncommon. Expect subscripts for a long time.
Betting into the player who was aggressive on the previous street, instead of the usual check to them. The name is unkind and dated — it is often a perfectly good play — but it has stuck.
How it is worked out
Rough guide under 5% is normal. Consistently high donk rates usually mark an inexperienced player.
Having led out and been raised, whether they stand their ground or give up. This is what separates a donk bet with a real hand from one thrown out to see what happens.
How it is worked out
Being led into and then raised is rare, so these two are among the slowest numbers on the whole page to become trustworthy.
When the preflop raiser checks instead of betting — showing weakness — how often this player attacks that weakness with a bet. Also called a stab. Shown with IP and OOP splits.
How it is worked out
Rough guide 40–60% in position. Attentive players stab far more when they act last.
Betting the turn after the preflop raiser declined to c-bet the flop — the turn-street version of a stab, asking whether their flop check meant what it looked like.
How it is worked out
Calling a flop c-bet in position, then betting the turn when the raiser checks — a two-street move that turns a call into a bluff. A meaningful float rate marks a player who fights for pots after the flop.
How it is worked out
WWSF's later-street siblings: of the turns (rivers) they saw, how often they took the pot. The three are NOT a funnel — each stands on its own denominator, and they should climb, because fewer players survive each street. A WWSR barely above WWST says the rivers they reach are ones they lose: calling down too light, or giving up on bluffs one street too late.
How it is worked out
Measured from parser version 12, so on a partly re-imported archive these rest on fewer hands than WWSF — the counts in the box header say how many.
W$SD cut by what they DID on the river: of the showdowns where they bet (raised) the river, how often they won. Betting the river is a claim to be ahead — these should sit ABOVE their overall W$SD, and when they do not, their river bets are worth calling wider.
How it is worked out
One hand can count in both cuts: bet, get raised, re-raise. WWSF, WTSD and W$SD themselves are defined in the header section.
Most players c-bet the flop often and then slow down — a big drop from flop to turn means they give up on later streets, and floating them (calling once with the intention of taking the pot later) becomes profitable. The same read applies in reverse to their folding: someone who folds little on the flop and a lot on the turn is peeling once and giving up.
Under each street row sits a fold / call / raise bar — the same decision as the Fold column, drawn as one whole. The three segments always add to 100%; each cell is 2% of the spots.
In 'Facing a c-bet', the Fold, Call and Raise columns carry a faint wash of the same colours the meters use, so column and bar read as the same quantity.
Where the whole stack goes in. How often that happens on each street, how the all-ins split between the four streets, and — because an all-in nearly always shows its cards — what they actually held at the moment the money went in.
| Street | Rate | Hands | Share of all-ins |
|---|---|---|---|
| Preflop | 1.8 | 684 | 31% |
| Flop | 2.4 | 826 | 38% |
| Turn | 1.3 | 451 | 21% |
| River | 0.6 | 218 | 10% |
| Total | 6.1 | 2,179 | no conditional rate — see below |
Illustrative example. Cards are known on nearly every one of these hands, which is what makes this the one tab not built on a biased slice.
How often the stacks go in on this street, out of every hand dealt. The four streets add up to their total all-in rate, which makes this the column to use when comparing two players' overall willingness to stack off.
How it is worked out
The same all-ins, divided by the hands that got to that street WITH CHIPS STILL BEHIND. This is the number that means something: it usually climbs steeply street by street, because the further a hand travels the more likely it ends with everything in the middle.
How it is worked out
Not simply saw_flop. A player who is already all-in preflop still gets dealt the runout, so saw_flop is 1 on virtually every preflop all-in — dividing by it would include hands where a flop all-in was impossible and quietly understate the rate. Hands whose stacks went in on an earlier street are removed from the denominator.
Each street's slice of this player's all-ins. The four add to 100% by construction, which is what makes the donut honest — it is a split of one quantity, not four separate numbers drawn together.
How it is worked out
The hand they held at the moment the stacks went in — read on the board as it was THEN, so a flop all-in is classified on three cards. Bottom pair that went on to river a full house counts as bottom pair, because bottom pair is what the decision was made with. Preflop uses holding classes instead, since there is no board yet.
How it is worked out
Net is what that holding actually returned; EV is the same all-ins with the luck averaged out, each scored by its equity when the money went in rather than by the card that fell. Read them TOGETHER — that pair is the most actionable thing on the tab. 'Two pair on the turn, 15 all-ins, −226 Net with −219 EV' is a spot that loses money. 'Three of a kind on the flop, −117 Net with +181 EV' is the same player running badly with sets. A frequency can tell you neither.
How it is worked out
net big blinds across that bucket's all-ins
Both are results, not proven rates. One all-in swings a whole stack, so even a dozen is a small sample — which is why the hand count sits immediately beside them and every card ends in a Total row. River cards show no EV: a river all-in has no cards to come, so there is nothing to average and EV would just repeat Net.
How many of that bucket's all-ins were won and lost, as plain counts. The shaded bar behind each row draws the same thing: its LENGTH is the bucket's share of the street and it fills green for won, red for lost, so you can read the split down a card without reading numbers. The Total row's bar runs full width, because the street is all of itself — its split is the street's whole record.
Length carries the sample on purpose, the same rule the holding grid uses. A bucket holding one hand draws a stub whether it won or lost, so a single lucky or unlucky all-in cannot shout. Chops are drawn grey and are rare — about 0.15% of all-ins.
Sums each card's hands, Net and EV. It is summed from the rows above it rather than taken from the street, so the bottom line always adds up to exactly what is printed over it — and it lets you see at a glance whether a street made money overall before reading which holdings drove it.
The same all-ins on a 13×13 starting-hand grid, one map per street. Squares hold a hand COUNT and the tooltip gives that holding's share, so each map is a distribution that adds to 100%. Watch it widen: a stack-off range that fills 14 squares preflop typically covers two or three times as many by the river.
How it is worked out
This is a composition, not a frequency — which is exactly why it is here rather than on the Range tab. Asking 'how often do they stack off with AKs' would need every AKs they were ever dealt as a denominator, and only the shown few per cent of those exist. Asking 'what did their all-ins consist of' keeps both halves inside the same nearly-complete set.
Above the grid, for the street on screen: how many of those all-ins were won, the net result in big blinds, and the same result with the luck averaged out — each all-in scored by its equity when the money went in rather than by the card that fell. The gap between Net and EV is the read: winning more than you should have is not the same as playing well, and this is the only number that separates them.
How it is worked out
In big blinds rather than money, because a filter can span currencies. River all-ins are never EV-adjusted — there are no cards to come — so EV equals Net there by construction, and the panel says so instead of showing a gap of zero.
The bottom strip of every occupied square is split in two: hands WON on the left over a green ground, hands LOST on the right over a red one, with the counts printed in them. Add the two for the square's total. A grey middle segment appears only where a pot was chopped.
The two halves are equal in width rather than proportional — a square holding one win and twelve losses would squeeze the '1' into a few pixels. The numbers carry the ratio; the colour only says which half is which.
And note what this deliberately is NOT: a win-rate tint over the whole square. Most squares hold one or two all-ins, so a rate there is nearly pure luck — printing two counts states a fact and shows you its size, where a colour ramp would assert a rate the sample cannot support.
Pairs / suited / offsuit for the map on screen. It pools 169 squares into three, which is the read that survives a thin sample — and under 30 hands it switches from percentages to raw counts rather than printing '67%' off two hands.
On the flop and turn only: how many of those all-ins also held a flush draw, an open-ended straight draw or a gutshot. Without it a card room full of semi-bluffs reads as 'high card', which is true and useless. Open-ended versus gutshot is decided by outs — two or more ranks that complete the straight is open-ended, so double-gutshots land in the right bucket.
Hand strength needs hole cards, and hole cards are usually missing — except here. Across the archive, 95.5–96.2% of all-in hands have known cards, against 5.8% of hands overall, because a pot that goes all-in gets shown. The all-in tab is therefore built on nearly its whole population, unlike the Range tab's frequency maps.
A hand counts on the street the POT went all-in, for every player still in it. Someone who called a shove is measured on the same footing as the one who pushed — this is 'what they had chips in with', not 'what they shoved'.
The four 'when reached' figures each have a different denominator, so adding them would be meaningless. Only the '% of hands' column sums, and it sums to the overall all-in rate.
When this player plays, and the individual hands worth looking at. The game-selection tab: filtered to your stake, the heatmap answers 'when is this player in MY game', and the hand list shows what they actually turned over.
| Mon | ||||||||||||||||||||||||
| Tue | ||||||||||||||||||||||||
| Wed | ||||||||||||||||||||||||
| Thu | ||||||||||||||||||||||||
| Fri | ||||||||||||||||||||||||
| Sat | ||||||||||||||||||||||||
| Sun | ||||||||||||||||||||||||
| 0 | 6 | 12 | 18 |
Illustrative example. Weekday evenings, heavier on Friday and Saturday, and a tail that runs past 3am at the weekend — the hours after midnight are where this player’s worst sessions live. Hours are in the player’s recorded table time, not your timezone.
A grid of the week — days down the side, hours across — where darker means more hands, shown in YOUR timezone (the label under it says which). The line above it puts the rhythm in words: total hands, active days, hands per active day, and their busiest hour. The shading is relative to their own busiest hour, not an absolute volume.
One table, three orderings. Recent showdowns is the last hands whose cards are known, newest first. Biggest wins and Biggest losses are their largest pots. Switching is instant because all three arrive with the page. Columns are date, stake, their cards, the board, pot type (SRP is a single-raised pot, 3BP a 3-bet pot), whether they won the showdown, and the hand's result.
Sorting by money would only ever rank a player's highest stake first — one hand at NL1000 beats every hand at NL10 before anyone plays a card. Big blinds is the same normalisation bb/100 uses, so a 100bb pot at NL10 and one at NL1000 rank as the achievement they each are.
Cards exist in a mined hand history only when they were shown. Recent showdowns therefore lists only hands that CAN be listed — not a sample of all their showdowns.
The two biggest lists deliberately do NOT filter on shown cards, because the fact on display is the size of the pot and hiding the unshown ones would quietly make it something other than 'their biggest pots'.
In practice little is missing: cards are known for roughly 90% of pots over 100bb, since big pots are how hands reach a showdown in the first place.
In Cards, the hand was never shown. In Showdown, the hand ended before one — which is not the same as losing it, so it is not drawn as a loss. In Board, either the hand ended before a flop or it was imported before the board was recorded; the tooltip says so rather than asserting one.
Seat names used throughout the page, and exactly which chair gets which name at each table size. Position is decided by when you act: acting last is a large, permanent advantage, which is why nearly every stat here is broken out by seat.
Small numbers under each seat are the stored seat index: 0 is the small blind, then clockwise, with the button always highest.
| Dealt | UTG | UTG+1 | UTG+2 | LJ | HJ | CO | BTN | SB | BB |
|---|---|---|---|---|---|---|---|---|---|
| 2HU | — | — | — | — | — | — | 0 | — | 1 |
| 3short | — | — | — | — | — | — | 2 | 0 | 1 |
| 4short | — | — | — | — | — | 2 | 3 | 0 | 1 |
| 5short | — | — | — | — | 2 | 3 | 4 | 0 | 1 |
| 6short | — | — | — | 2 | 3 | 4 | 5 | 0 | 1 |
| 7 | — | — | 2 | 3 | 4 | 5 | 6 | 0 | 1 |
| 8full | — | 2 | 3 | 4 | 5 | 6 | 7 | 0 | 1 |
| 9full | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 0 | 1 |
Two seats are fixed from the bottom — the small blind and big blind always sit at seat 0 and 1. The rest are counted back from the button: the button is the last seat, the cut-off is one before it, then middle position, then early position. Whatever is left over at the front of the table is under the gun. This means the names are relative to table size, not absolute chairs.
With only two players there is no room for the middle names, and the button also posts the small blind. So heads-up has just BU and BB. This reverses the usual order before the flop: the button acts FIRST preflop and last on every street afterwards. Never compare a heads-up VPIP against a full-ring one — 45% is tight heads-up and wildly loose at nine.
Names drop off the front as the table shrinks. At six-handed there is no early position at all — it needs seven dealt before there is a seat spare for it. At five-handed there is no under-the-gun either, at four no middle position, and at three you are down to just the blinds and the button.
At eight and nine dealt there are more early seats than there are names for them, so UTG becomes a bucket: it covers two seats at eight-handed and three at nine-handed. A 9-handed 'UTG' number therefore mixes the true under-the-gun seat with the two after it. That is a deliberate simplification, not an error, but it does mean full-ring UTG stats are slightly looser than the very first seat alone would be.
Several stats cannot exist in certain seats. RFI needs the pot folded to you, which cannot happen in the big blind. Cold Call excludes the blinds by definition, because they already have money in. Steal only counts from the cut-off, button and small blind. A dot in those cells means the situation is impossible there, not that it never happened.
Distance 1. One seat before the button, and the second-best seat. A common place to start attacking the blinds.
Distance 2. Named for hijacking the cut-off's steal before they get the chance.
Distance 3. One seat earlier again. At a six-handed table this is the earliest non-blind seat — the one most players would call "under the gun" there.
Distances 4, 5 and 6. These only exist at bigger tables: UTG+2 needs seven dealt, UTG+1 eight, and UTG nine. UTG is the tightest seat on the table by a clear margin, because six players still get to act behind it.
Posts the smaller forced bet. Acts second-to-last before the flop and first on every street afterwards, which makes it the hardest seat to play. It keeps its own name rather than a distance because having money already committed changes the decision entirely.
Posts the larger forced bet. Acts last before the flop, then first afterwards. Already has money in the pot, so it defends far more hands than any other seat.
Poker's traditional vocabulary is not consistent across table sizes: six-handed players call the earliest seat "UTG", while nine-handed players call that exact same seat "LJ" and save "UTG" for a seat two further out. Following convention meant two seats with identical position sat in different rows, while three genuinely different nine-handed seats were merged into one. Naming by distance means a label always identifies exactly one position.
In Position means acting last after the flop; Out Of Position means acting first. Information is the whole game, and position is how you get it.