How LeakPrice counts

The method is the same for every report. The report itself, in “How we counted”, links here and gives your database's numbers: what was left out of the calculation, how many hands went into each review, the time zone.

• How to read the numbers
• Stats and results
• Preflop
• Postflop: spots and top regs
• Solver check
• Solver trees
• River: bluffcatchers
• Aggression vs the solver
• Pool exploit
• Pool review
• Price of findings and potential
• What the report doesn't know

How to read the numbers

• ± — 95% interval: within it, the result may be just variance. Color marks what falls outside ±; the rest is a direction, not a conclusion. On a short sample, ± for early positions can be wider than the win rate itself.
• The difference is not random — |z| ≥ 2.6 by z-test. The threshold is stricter than the usual 1.96: the report has hundreds of comparisons, and at 1.96 some differences would be random. Where z has its own column, it's shown as a number.
• Finding color — size of the difference: red — 15 points or more, yellow — 8–14, no color — under 8. Green — where you miss value, not where you lose.
• Pool skew — a difference from the solver that is not random (|z| ≥ 2.6) and is at least 3 points: the points cut out small gaps, the z-test cuts out randomness, both are needed. Over hundreds of thousands of pool decisions a z-test alone would pass even a one-point difference you can't play against.
• Each stat has its own sample threshold, shared by the pool and your database, and the number isn't printed until it's reached: VPIP needs a couple hundred hands, street aggression three hundred, WWSF three thousand.

Stats and results

• Heads-up hands (two players at the table) are not analysed: the ranges there are different and cannot be compared with 6-max. How many there are and their result — in a line at the top of the report; next to it — the result of the whole database including them, as in a tracker.
• Winnings graph — result by hand, in the order played. Blue — hands that went to showdown, red — the rest (folds and walks); together they make the total. Dashed — the result with every all-in before the river replaced by its expectation: equity × pot. EV diff — Net won minus All-in EV: negative — you ran below EV, positive — above. This is variance, not play.
• All-in EV — heads-up all-ins that reached showdown only. Equity on the flop and turn is enumerated exactly; preflop — over 2,000 random runouts. Multiway all-in pots are counted at Net won.
• Postflop stats: sample size in brackets; turn and river cbet — second and third barrel in a row, as in the tracker; delayed cbet — on a separate line. “Bettor acts first” and “bettor acts last” are summary rows over all heads-up position pairs: no conclusions are drawn from them — that's what the position pairs in the top-reg comparison are for.
• The “Solver” column in the stats — what the stats would be for a player who plays like the solver everywhere. Preflop — solver ranges for the whole 6-max table. If the line goes into a spot the ranges don't have (most often after a cold call: squeeze and overcall are not solved, hence the dash for squeeze), decisions up to that point are in the stats, later ones are not; the share of such hands is in the report. Cold 4bet spots the ranges don't have use a substitute spot for the solver: the same player vs the same 3bet with a different opener. Postflop and results — all solved heads-up spots, weighted by how often each spot occurs. Spots: SRP BTN vs BB · 3bet pot SB vs BTN · 3bet pot BB vs BTN · 3bet pot BB vs SB · SRP CO vs BB · SRP HJ vs BB · SRP SB vs BB · 3bet pot BTN vs CO · 3bet pot BTN vs HJ · 3bet pot CO vs HJ · 4bet pot BTN vs BB · 4bet pot BTN vs SB · 4bet pot SB vs BB. In the “Solver check” only spots where you have enough hands for a row are shown. Flop donk has a dash in the “Solver” column: a caller lead exists only in some trees (SRP BTN and CO vs BB and 3bet pot BB vs SB don't have it), and the solver number would come from different spots than yours. Trees where the caller has a lead: SRP HJ vs BB · 3bet pot BTN vs CO · 3bet pot BTN vs HJ · 3bet pot CO vs HJ · 4bet pot BTN vs BB · 4bet pot BTN vs SB. Your numbers (in the pool review, the pool's numbers) cover the whole database, with multiway pots and unsolved spots, so the “Solver” column is a reference point, not a threshold. Solver W$SD and WWSF — 50% by construction: the pot one player won, the other lost. For the solver a chop is half the pot; for you, as in a tracker, it's a win.
• Aggression by street — the share of hands where you bet or raised at least once on that street, out of all hands where you acted on it. Top regs are compared on comparable hands: 5–6-player tables, stacks from 100 bb, no ante; The “You” column next to top regs is computed by the same code and with the same filters as the top regs and, as everywhere postflop, without hands with a recreational player on the flop and without 4bet pots. 3–4-player tables aren't included: the position mix there is different, and aggression shifts with it, not with play. Band — the aggression norm across the whole database: flop 35–42%, turn 29–33%, river 33–36% (300+ hands per street). Above the band — overaggression, below — too many checks. Low river aggression in position loses money directly: hands that win only by betting get checked; out of position a check is almost free — the opponent will bet for you. Frequency doesn't show whether the difference costs money; where it shows — that's in the findings by street.
• Where the hand ends. The table has no hands where you made no move: everyone folded to your BB (walks). The table and the hands outside it (mostly walks) add up to the database result — the reconciliation is in the report.
• A session is a run of hands with no break longer than 30 minutes; typical — median. Session and playing-time win rate — after rake; rows under 2,000 hands — no win rate. How many tables were open at once can't be recovered from fast-fold history (Zoom, Rush & Cash): the table name there is either one for the whole pool or new every hand. The workload measure here is pace.
• Time — as recorded by the room; if the room doesn't record a time zone, it's the room client's time zone. Highlighted are the three hours and two days with the highest share of recs and, if the database has nicknames, those with a VIP at the table most often (a rec with VPIP of 50% or more). Playing time: win rate — over all your hands on that day or hour, as in “Win rate by position”, from 2,000 hands per row. Composition — the share of recs among opponents with a known class, and how many recs sit with you on average; from 300 hands and if the class is known for at least 30% of opponent seats, otherwise “—”. Opponent class (reg or rec) — over the whole database, by the same thresholds as in the pool review; opponents without a class are not counted in the share.
• Table line-up counts only regular tables of 6+ players, not fast-fold: there the line-up holds from hand to hand. EV bb/100 — win rate with all-ins before the river that reached showdown counted by equity, as in a tracker. Average pool reg — NL400–2000 6-max. Rake per seat is the table's rake divided by the players: what each one loses when there is nobody at the table to win from.
• Tilt. Extra cost — bb per 100 hands while down, vs normal play in the same part of the session; the conclusion is based on sessions down 300 bb or more; the “from 600 bb” row is for reference. Extra cost preflop — how much you lose vs the best play by the solver with the same hand in the same spot; on the river — the call's +EV for the folded hand vs the pool's betting range (as an interval: with and without your hands in the betting range). Normal play — the same part of the session (up to 1,000, 2,500, 5,000 hands and beyond): losses come late more often, and fatigue in a long session shouldn't look like tilt. The interval is 95% over sessions (400 resamples). River fold — to the opponent's first bet, difference from normal play in percentage points. Flop, turn, and river calls and bets aren't included.

Preflop

• Preflop is checked against a preflop solver: every dealt hand has a known EV for each action, and the loss adds up over hands. Raise — the spot's raising action: open, 3bet, 4bet or 5bet.
• Preflop by spot: “You play” — anything but fold; pool — the opponents at your tables; “You lose” — the cost of your decisions in the spot vs the best play, in bb/100 over the whole database. A spot with fewer than 30 decisions is shown without a conclusion. In spots you reach after having already acted (response to a 3bet, 4bet and 5bet, response to a raise after your own limp), the solver is counted on the cards you reached the spot with; so it differs from the same spot in the pool review — that uses the solver's full range: pool cards aren't visible. Decisions the solver has no spot for are not checked: facing a 3bet or 4bet with a stack under 70 bb, facing a 3bet or 4bet that a third player had already called before you (the solver's pot is heads-up), defending vs an open more than 0.3 bb from the solver's size, facing a min-3bet or any other 3bet or 4bet whose size differs from the solver's by more than 25%, spots with UTG and HJ at tables with other than 5–6 players — and folding TT+ and AK when first to act or AA, KK against a 3bet or 4bet: nobody plays this way, it looks like a timeout. They are not in the price, the frequencies or the conclusion; how many you have is in the report.
• A pool preflop response counts as skewed if its difference from the solver is not random (|z| ≥ 2.6) and is at least 3 points. “Justified” doesn't mean “more profitable than the solver”: the pool at these tables responds so that your deviation goes the right way. “Not justified” — the pool responds the opposite way. “Not explained by the pool” — the pool doesn't deviate noticeably from the solver, so the reason for your deviation isn't the pool.
• Range composition — matrices for spots where you have at least 100 decisions: left — you, right — the solver; each cell is split into raise, call and fold by how often each is played; clear differences are outlined. Clear divergence — a hand the solver plays almost always (90% or more), and you 50+ points less often, or the solver almost never plays (10% or less), and you 50+ points more often; a hand needs five deals or more. “You lose” — EV of the best action minus EV of the chosen one per the solver, for the cards of this hand dealt to you in this spot; this includes a wrong choice between call and raise. bb/100 — over 100 hands of the whole database, to compare directly with your win rate. “By hands vs fold” — how much the hand made over a plain fold in your hands: folding first in costs exactly what's invested (from SB −0.5 bb, from other positions — 0), so the hand result compares with folding directly.

Postflop: spots and top regs

• Postflop — hands where you saw the flop, excluding hands with a rec on the flop (if the database has nicknames to classify opponents) and 4bet pots; how many you have is in the report. Postflop is rated only vs regs and unclassified players. Hands with a rec on the flop are a separate review. 4bet pots are too rare for a money estimate or a comparison with top regs; they are checked against the solver in their own spots (“Postflop solver check”). General stats and win rate by position — over all hands.
• Spot — pot type, role and the pair of positions. Loss in a spot vs top regs: (top regs' win rate − your win rate in the spot) × the spot's share of hands in the database, in bb/100 of the whole database. Spots with similar logic are merged for sample size: EP/MP — one position (UTG and HJ), SRP CO and BTN vs BB — one spot, 3bet pot SB and BB vs BTN — one spot (if still too few hands — vs CO and BTN together); same for top regs. “Not proven” — the difference from top regs is within ±: money in such a spot is only a direction (in a 3bet pot over a few hundred hands ± is hundreds of bb/100). Money by spot is a direction, not the price of a specific mistake: stats and frequencies show where you play differently from top regs, not which hand you misplayed; specifics — in the street findings and in lines. “Share of position losses” — what part of your losses in that position this spot accounts for; over 100% — you play the other spots of this position better than top regs. Win rates by position vs top regs — only hands heads-up on the flop vs regs and unclassified players, so they don't match “Win rate by position” at the start of the report. A multiway pot doesn't belong to a single position and isn't part of the by-position breakdown; how many you have and your winrate in them is in the report. Assumption: win rates are before rake on both sides; rake for one of the top regs (we don't have his hands) is taken from the other top regs' rake in the same spot; per-hand variance for top regs is taken equal to yours in the same spot: only their results are stored, not the variance.
• Spot win rates vs top regs are compared before rake — for you and for top regs. Top regs play limits where rake per pot is hundredths of a blind; yours is tens of times higher; after rake the difference in stakes would count as your loss, though it has nothing to do with play. The rake you pay in the spot is a separate column: it's the price of your stake, not a mistake.
• Lines vs top regs. Line — how you play from the flop to the end: “bet / bet / bet”, “bet / check / bet”, “check-call / check-call / check”. Next to you — top regs (5 players combined). “Frequency” — you choose the line more or less often; “execution” — you win less in the same line. The spot's win rate difference splits by line, with no remainder, into “frequency” and “execution”. EV — before rake and in flop pots, shown in bb at your average flop pot in the spot: otherwise rake differences between stakes and open or 3bet sizes would look like misplay; so the spot win rate here doesn't match the top 5. Row — from 30 of your hands, EV and “execution” — from 50, a spot under 300 hands — no line table: on a smaller sample it's noise. * — your line share differs from top regs and the difference is not random (|z| ≥ 2.6), or the line EV is outside ± with 50+ hands on each side. A line's EV depends on which hands go into it: if you check strong hands more often than top regs, one line gains EV and another loses it — that's a redistribution, not two mistakes; so the conclusion is made by decision point and by hands. Lines without showdown don't go into “execution”: the EV there is just the bet sizes. In “Choice” items the frequency difference is proven (|z| ≥ 2.6 for at least one option); money is an estimate from option EVs, not a proven price. An option's result is in full and before rake, including what was invested up to this point, so for a fold it's negative; facing a bet, a fold is worth 0 — these numbers can't be compared. An option's result is shown from 50 hands, otherwise “few hands”.
• Point EV (response to a bet and a raise) — from the decision, before rake, in bb per hand: hand result plus everything you invested before this move. A point is shown from 30 hands, EV — from 50. A fold is worth 0, so EV shows what your calls and raises make. Counted in pots at the decision and converted to bb at your average pot at that point — so differences in bet sizes and rake between stakes don't look like mistakes. Previous streets aren't distinguished: “turn: you bet, opponent raises” — regardless of how the flop was played. “Playing like top regs” — (top regs' EV − yours) × your hands at the point: plus — how much you lose vs top regs, minus — you respond better. * — your share differs from top regs and the difference is not random, or EV is outside ±. This is a direction: EV depends on the hands you reach the point with.
• Comparison with top regs by street: the most notable spots of the street are shown, no more than 10. The aggressor's turn bet includes delayed bets. “Fold to aggressor's bet” — only when they act before you.
• Your sizings: sample — your first bet or raise on the street, as aggressor and caller, heads-up and multiway (a bet took the pot only if everyone folded); without hands with a recreational player on the flop and without 4bet pots. Sizing findings are counted on it; the “Pool folds to your bet” table in the pool review uses a different sample, so the numbers differ. Green — the pool folds 8 points more than needed and 5 more than the solver in counted heads-up spots where the solver uses this size.
• Excess in the table of your folds — fold points above “folds needed”: the opponent's bluff pays off; the table doesn't say whether your fold is right.
• Folds needed — the fold share at which a bluff breaks even: B / (B + P), where B is the bet, P the pot before it. It's a threshold for either side, not a defense norm. Folds above needed mean the opponent's bluff pays off — but on their own they don't mean your fold is wrong: if the range came to the street weak, the mistake is earlier in the tree. Third pot — 25%, half pot — 33%, pot — 50%, two pots — 67%. MDF — 1 − folds needed. Counted for a pure bluff with no equity: a real hand has backdoors, so the edge is a bit bigger. A raise doesn't shift folds needed: an empty hand folds to a raise and loses the same bet.

Solver check

• Solver check: every spot is solved on 177 flops — one flop of each type, each pair of positions with its own tree. A row is shown from 50 of your hands; highlighted — the difference is not random. A flop type means the same suits, paired or not, the same top card, the same connectedness and the same number of cards ten or higher. Each of your hands follows the tree of the solved flop of its type, and at every point where you acted, the solver's frequency at the same point of the same flop is shown next to yours. On 123 of the 177 types the preflop aggressor has fewer cbet sizes on the flop than on the other 54: on monotone flops one size, 25% (in SRP vs the BTN — 33%), on paired flops the same, except in 3bet pots; for BTN and CO vs BB on the other flops of these types — 33/75% instead of 20/33/50/75%; in SRP SB vs BB the sizes are everywhere the same as on the 54. So where there is one size, a cbet of 40% or more (vs the BTN — 53% or more) is checked only as a bet of any size, and the hand does not go further down the tree. The solver plays against the solver — deviating from it isn't necessarily a mistake: vs the pool, playing differently is sometimes better. This is a check, not a price.
• Sizes in the solver check: bets and raises are snapped to the nearest tree size. Raise size as in the solver: the amount on top of the call as a share of the pot after the call (a 60% raise over a 33% bet is a raise to 1.33 pots). Your bet or raise more than 1.6 times away from the nearest size counts as a generic bet or raise (for the solver — all its sizes together), but such a hand goes no further down the tree.
• Sizing combinations: a combination is your flop and turn actions with the same line in between (cbet 33% → call → turn 75%, check-raise vs cbet → call → turn bet); the pool's most common combinations are shown, next to your shares and the solver's shares on the same flops. Compare action-row EVs only with your result, not with each other. Spots are merged as in the top 5 (CO/BTN vs BB, SB/BB vs BTN). “Your result” — average hand outcome from the start of the flop, after rake, with ± (1.96·sd/√n); "Solver EV" is from the start of the flop, in bb at your average flop pot in the spot. The first row is the whole spot: all your hands in the role and the solver's EV at the start of the flop on the same flops. A combination row has all your hands that reached this point and the solver's EV with its whole range there. Action rows have only the hands that chose this action: the solver bets stronger hands, so bet EV is almost always above check EV. Solver — on your hands in the spot, on your flops; if you have none — on pool hands (marked in the row). Actions that neither you nor the pool take and the solver takes less than 10% of the time are not shown. “Bet of another size” — more than 1.6 times away from the nearest tree size. The pool here isn't a model, just an answer to “what do people do most often”. Solver EV is vs the solver, your result is vs the pool and on your hand selection: this is a check, not a price.
• Sizing combinations flop → turn: at most 6 combinations per role; row — from 30 of your hands or 100 pool hands, result — from 50 of yours. Highlighted if the difference from the solver is not random (|z| ≥ 2.6) and the spot has 50+ hands. Tree sizes are the same as in the solver check.
• Solver by role. Spots: SRP BTN vs BB · 3bet pot SB vs BTN · 3bet pot BB vs BTN · 3bet pot BB vs SB · SRP CO vs BB · SRP HJ vs BB · SRP SB vs BB · 3bet pot BTN vs CO · 3bet pot BTN vs HJ · 3bet pot CO vs HJ · 4bet pot BTN vs BB · 4bet pot BTN vs SB · 4bet pot SB vs BB — the same trees as in the solver check (where only spots with enough hands for a row are shown). Opponents — your opponents in the same spots and role. You and the solver are combined with the same mix: the weight of each spot and role is how often the spot occurs for the solver, so the difference is strategy, not which spots you play more often. Probe — the preflop caller bets first after the previous street was checked through; donk — betting first into the player who bet on the previous street. Caller flop aggression — excluding hands where the caller led the flop in a spot with no lead in the solver tree: the solver doesn't have that option; same for opponents. Spots without a caller lead: SRP BTN vs BB · 3bet pot BB vs SB · SRP CO vs BB. A chop at showdown counts here as half a win in W$SD and WWSF, as for the solver (in the general stats, as in a tracker, it's a win). The solver plays against the solver, you play against the pool: against a pool that underbluffs, the correct exploit can move you away from the solver.

Solver trees

• Solver check: BTN vs BB, single-raised pot. Solver on 177 flops: BTN opens, BB calls; 100 bb, rake 5% capped at 1.38 bb, as at NL200. Your BTN vs BB hands are compared on your flops, not “on average”. Tree sizes: flop: BTN cbet 20/33/50/75% (on paired flops — only 20%), BB has no lead, check-raise 60%, then raises 60%/all-in; turn: BTN 33/75/135%, BB donk 15%, probe and bet after own check-raise 33/75%, raises BB 60%/all-in, BTN 100%/all-in; river: BTN 50/75/125%/all-in, BB 33/75/135%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone and paired flops the BTN cbets only 25%, on the others 33/75%. Hands with a stack outside 80–130 bb aren't checked: the tree is solved for 100 bb.
• Solver check: CO vs BB, single-raised pot. Solver on 177 flops: CO opens 2.5 bb, BB calls; 100 bb, rake 5% capped at 1.38 bb, as at NL200. Your CO vs BB hands are compared on your flops, not “on average”. Tree sizes: flop: CO cbet 20/33/50/75%, BB has no lead, check-raise 60%, then raises 60%/all-in; turn: CO 75/135% (no 33%), BB donk 15%, probe and bet after own check-raise 33/75%, raises BB 60%/all-in, CO 100%/all-in; river: CO 50/75/125%/all-in, BB 33/75/135%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone and paired flops the CO cbets only 25%, on the others 33/75%. Hands with a stack outside 80–130 bb aren't checked: the tree is solved for 100 bb.
• Solver check: HJ vs BB, single-raised pot. Solver on 177 flops: HJ opens 2.5 bb, BB calls; 100 bb, rake 5% capped at 1.38 bb, as at NL200. Your HJ vs BB hands are compared on your flops, not “on average”. Tree sizes (as players bet): flop: BB lead 25%, HJ cbet 33/75%, raise 60% (vs lead and vs cbet), then 60%/all-in; turn: HJ 33/75/135% (after cbet and call — only 75/135%), BB donk 25%, probe and bet after own raise 33/75/135%, raises 60%/all-in; river: BB 33/75/135%/all-in, HJ 33/75/125%/all-in, raises 60/120%/all-in; flop types added on 19.10.2026: on monotone and paired flops the HJ cbets only 25%. Hands with a stack outside 80–130 bb aren't checked: the tree is solved for 100 bb.
• Solver check: UTG vs BB, single-raised pot. Solver on 177 flops: UTG opens 2.5 bb, BB calls; 100 bb, rake 5% capped at 1.38 bb, as at NL200. Your UTG vs BB hands are compared on your flops, not “on average”. Tree sizes (as players bet): flop: BB lead 33%, UTG cbet 33/50%, raise 60% (vs lead and vs cbet), then 60%/all-in; turn: UTG 33/75/150%, BB donk 33%, probe and bet after own raise 33/75%, raises 75%/all-in; river: UTG 75/150%/all-in, BB 33/75%/all-in, raises 75%/all-in; flop types added on 19.10.2026: on monotone and paired flops the UTG cbets only 25%. Hands with a stack outside 80–130 bb aren't checked: the tree is solved for 100 bb.
• Solver check: SB vs BB, single-raised pot. Solver on 177 flops: SB opens 3 bb, BB calls; pot 6 bb, stack 97 bb; rake 5% with a 1.38 bb cap, as at NL200. Your SB vs BB hands are compared on your flops, not “on average”. Tree sizes (as players bet): flop: SB cbet 33%, BB after a check 33/75%, raise 60% (vs cbet and check-raise), then 60%/all-in; turn: both 33/75/135%, donk SB 33%, raises 60%/all-in; river: SB 33/75/125%/all-in, BB 50/75/125%/all-in, raises 60%/all-in. Hands with a stack of 80–130 bb and a flop pot of 4.5–8 bb are checked (open SB 2.25–4 bb, the min-raise is not included): the tree was solved for an open of SB 3 bb and 100 bb before the deal.
• Solver check: 3bet pot, BB vs BTN. Solver on 177 flops: BTN opens 2.5 bb, BB 3bets 11.5 bb, BTN calls; pot 23.5 bb, stack 88.5 bb; rake 5% with a 1.38 bb cap, as at NL200. Your BB vs BTN hands are compared on your flops, not “on average”. Tree sizes: as in 3bet SB vs BTN: flop: BB cbet 33/50%, BTN after a check 33/50%, raise 60%, re-raise all-in; turn: both 33/50/75%, donk BB 33%, raises 60%/all-in; river: BB 33/75%/all-in, BTN 50/75/110%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone flops the BB cbets only 25%. Hands checked: stack 80–130 bb and flop pot 16–30 bb; the tree is solved for a 3bet to 11.5 bb with 100 bb starting stacks.
• Solver check: 3bet pot, SB vs BTN. Solver on 177 flops: BTN opens 2.5 bb, SB 3bets 10.5 bb, BTN calls; pot 22 bb, stack 89.5 bb; rake 5% with a 1.38 bb cap, as at NL200. Your SB vs BTN hands are compared on your flops, not “on average”. Tree sizes (as players bet): flop: SB cbet 33/50%, BTN after a check 33/50%, raise 60%, re-raise all-in; turn: both 33/50/75%, donk SB 33%, raises 60%/all-in; river: SB 33/75%/all-in, BTN 50/75/110%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone flops the SB cbets only 25%. Hands checked: stack 80–130 bb and flop pot 16–30 bb; the tree is solved for a 3bet to 10.5 bb with 100 bb starting stacks.
• Solver check: 3bet pot, BB vs SB. Solver on 177 flops: SB opens 3 bb, BB 3bets 9 bb, SB calls; pot 18 bb, stack 91 bb; rake 5% with a 1.38 bb cap, as at NL200. Your BB vs SB hands are compared on your flops, not “on average”. Tree sizes: flop: no lead for SB, BB cbet 33/50%, check-raise 60%, re-raise 60%/all-in; turn: BB 33/50/75%, SB 33/75%, donk SB 33%, raises 60%/all-in; river: SB 33/75%/all-in, BB 50/75/110%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone flops the BB cbets only 25%. Hands with a stack of 80–130 bb and a flop pot of 13–24 bb are checked: the tree was solved for an open of SB 3 bb, a 3bet to 9 bb and 100 bb before the deal.
• Solver check: 4bet pot, SB vs BB. Solver on 177 flops: SB opens 3 bb, BB 3bets 9 bb, SB 4bets 23 bb, BB calls; pot 46 bb, stack 77 bb; rake 5% with a 1.38 bb cap, as at NL200. Your SB vs BB hands are compared on your flops, not “on average”. Tree sizes (as players bet): flop: SB cbet 25/50%, BB after a check 25%, raise 60%/all-in (vs a 50% cbet — all-in only); turn: both 33/50%/all-in (after a 50% cbet and a call — 33%/all-in), SB donk 33%/all-in, raises all-in only; river: SB 50%/all-in, BB 33%/all-in, raises all-in only; flop types added on 19.10.2026: on monotone and paired flops the SB cbets only 25%. Hands with a stack of 80–130 bb and a flop pot of 34–60 bb are checked: the tree was solved for an open of SB 3 bb, a 3bet to 9 bb, a 4bet to 23 bb and 100 bb before the deal.
• Solver check: 3bet pot in position (BTN vs CO and HJ, CO vs HJ). Solver on 177 flops: open 2.5 bb, 3bet in position 8.2 bb, call; pot 17.9 bb, stack 91.8 bb; each pair has its own tree and its own solver ranges; rake 5% with a 1.38 bb cap, as at NL200. Tree sizes (as players bet), the same for all three pairs: flop lead OOP 25% (leads 20–35% of pot), IP cbet 33/50%, check-raise 60%/all-in; turn IP 33/50/75% (after cbet and call 50/75%), OOP 33/75%, donk 33%; river OOP 33/75/125%/all-in, IP 50/75/110%/all-in; raises 60%/all-in (a raise on a lead too); flop types added on 19.10.2026: on monotone flops the IP cbets only 25%. Hands with a stack of 80–130 bb and a flop pot of 13–24 bb are checked: the trees were solved for an open of 2.5 bb, a 3bet to 8.2 bb and 100 bb before the deal.
• Solver check: 3bet pot, SB/BB vs CO and HJ. Solver on 177 flops: CO or HJ opens 2.5 bb, BB 3bets 11.5 bb or SB 3bets 10.5 bb, call; pot 22–23.5 bb, stack 88.5–89.5 bb; each pair has its own tree and its own solver ranges; rake 5% with a 1.38 bb cap, as at NL200. Your SB and BB vs CO and HJ hands are compared on your flops, not “on average”. Tree sizes (as players bet), the same for all four pairs: flop: blind cbet 33/50%, CO/HJ after a check 33%, raise 60%; turn: blind 50/75%, CO/HJ 33/50%, blind donk 33%, raises 60%/all-in; river: blind 33/75%/all-in, CO/HJ 50/75%/all-in, raises 60%/all-in; flop types added on 19.10.2026: on monotone flops the blind cbets only 25%. Hands with a stack of 80–130 bb and a flop pot of 16–30 bb are checked: the trees were solved for an open of 2.5 bb, a 3bet to 10.5–11.5 bb and 100 bb before the deal.
• Solver check: UTG/HJ/CO vs BTN, single-raised pot. Solver on 177 flops: UTG, HJ or CO opens 2.5 bb, BTN calls; pot 6.5 bb, stack 97.5 bb; each pair has its own tree and its own solver ranges; rake 5% with a 1.38 bb cap, as at NL200. Your UTG, HJ and CO vs BTN hands are compared on your flops, not “on average”. Tree sizes (as players bet), the same for all three pairs: flop: UTG/HJ/CO cbet 33/50%, BTN after a check 33/50%, raise BTN 50%, check-raise UTG/HJ/CO 60%; turn: UTG/HJ/CO 33/75/150%, BTN 50/75/150%, donk UTG/HJ/CO 33%, raises 75%/all-in; river: UTG/HJ/CO 33/75%/all-in, BTN 75/150%/all-in, raises 75%/all-in; flop types added on 19.10.2026: on monotone and paired flops the UTG/HJ/CO cbets only 33%. Hands with a stack outside 80–130 bb aren't checked: the tree is solved for 100 bb.
• Solver check: 4bet pot, SB/BB vs BTN. Solver on 177 flops: BTN opens 2.5 bb, BB 3bets 11.5 bb or SB 3bets 10.5 bb, BTN 4bets 26 or 25 bb, call; pot 51–52.5 bb, stack 74–75 bb; each pair has its own tree and its own solver ranges; rake 5% with a 1.38 bb cap, as at NL200. Tree sizes (as players bet): flop: BTN cbet 25/50%/all-in, blind — lead all-in only, raise vs cbet all-in only; turn: after check-check both 33/50%/all-in, blind's donk 33%/all-in, BTN after cbet and call 33%/all-in (vs SB also 50%); river: blind 33%/all-in, BTN 50%/all-in if the turn went check-check, otherwise all-in only; raises on the turn and river all-in only; flop types added on 19.10.2026: on monotone and paired flops the BTN cbets only 25%. Hands with a stack of 80–130 bb and a flop pot of 38–68 bb are checked: the trees were solved for an open of 2.5 bb, a 3bet to 10.5–11.5 bb, a 4bet to 25–26 bb and 100 bb before the deal.

River: bluffcatchers

• Showdowns next to a river fold — what the pool actually showed in the same spot: same pot type (SRP or 3bet pot), same bet and pot size, same number of barrels before the river. What's compared is not the hand's name but its strength on its board — the share of hands it beats.
• The bluffcatcher cheat sheet counts whether your call pays off, not how many bluffs the pool's bet has. “Weak pairs” are the same as bluffs in the pool review: third pair and lower and pocket pairs below the middle board card. A pocket pair above the middle card and second pair on a paired board count as second pair here, an overpair on a paired board as top pair (in the pool review the first two are merge, an overpair on a paired board is value).
• Bluffcatcher table. “Air” — high card and a board pair without a pair of your own; «weak pairs» — any pair below middle pair: third pair and below and a pocket pair below the middle card of the board (on K-T-5 — 5x and 99–22), on a paired board the same; a pocket pair between the middle and top card is second pair. Bluff — air and weak pairs: the solver rarely bets a weak pair on the river, in a small pot too. “Required” — the equity a call needs, B/(P+2B−rake), median across your folds. “Folds with top hands” — your folds with hands the cell says to call with: second and top pair, and in “call: weak pair and better” cells also weak pair (in “borderline” cells — second and top pair). “Call with top hands, bb” — the sum of call EV in these folds, winning and losing, vs the range without your hands and vs the full one. The finding's price is in the card, by the threshold rule.
• The threshold rule: call a river bet if the hand's strength on this board is not below the threshold for this bet size; the threshold is fitted on one half of your river folds and tested on the other; the split into halves is repeated 20 times, and the price is the median. The threshold rule looks at a hand's strength on its board, the cheat sheet at its name. So the calls the rule adds include weak pairs too: on some boards they are stronger than the threshold. What these calls are made of in your database, what calling with weak pairs would have given, and how many folds are unrated (too few showdowns in that pot type) — in the report.
• River-fold examples are picked by the second calculation. It doesn't count the price: it runs through all opponent hands on this board, varies their weights many times around what showdowns show, and once more with a computed bluff share for this bet size. “Confirmed” — calling is +EV at any weights. “Checked by price only” — not enough showdowns for this bet size. Folds the second calculation considers correct, or where the answer depends on weights, don't go into the examples. How many folds it confirmed, how many it considers correct and how many are unstable — in the report.
• What you check and bet the river with: your cards are known in every hand, so the breakdown is exact. Bet EV and check EV — from the decision point: hand result + amount invested before the river, bb. Caveat: you choose check or bet yourself, so the hands within a type differ between them — EV of bet vs EV of check is an estimate, not a price. The solver is checked in two spots: BTN vs BB in SRP and BTN vs a 3bet from SB or BB. After the opponent checks, BTN bets two pair and better eight or nine times out of ten, top pair about half the time or more, and missed draws less often than it checks them in SRP, more often in a 3bet pot. Acting first (BB in SRP, SB or BB after their own 3bet), the solver checks strong hands too: checking them costs you value only if you check noticeably more of them than the solver. Under 30 decisions — “—”.

Aggression vs the solver

• Underused aggression and overaggression — only heads-up spots solved by the solver. A play makes the list when you make it less (or more) often than the solver, the difference is not random (|z| ≥ 2.6) and you have 100+ opportunities. Order — by the number of missed (extra) plays over the database: a frequent spot comes first, not a rare one with a big gap. Missed over the database — (solver − you) × your opportunities; extra over the database — (you − solver) × your opportunities.
• Aggression vs the solver: a spot frequency can't prove a price in bb — only the potential rows have one. “Bet or raise on the street” — overall aggression; doesn't make the list if the same difference on the same street is already explained by a specific play. Probe — the preflop caller bets first after the previous street was checked through; donk — betting first into the player who bet on the previous street; flop donk — only in spots where the caller has a lead in the solver tree. The solver plays vs the solver; you play vs the pool. Extra plays can be value too: the frequency doesn't tell whether they are bluffs or thin value bets.
• Pool response to an aggressive play (the last column of the table) — how often opponents in this database fold to that play in the same spot and role, over all hands in the database: whether you played the hand, folded preflop or weren't at the table (anyone makes the play, opponents respond). “Needed” — what the play puts in, divided by that amount plus the pot before the play, averaged over their sizes. Solver fold — to the same play in the same spot; the solver uses its own sizes. “Few” — fewer than 30 responses. Green — the pool folds more than needed and more than the solver; red — less than needed; color — only where the difference is not random (|z| ≥ 2.6); a pool at the solver's level is no reason to make the play more often than the solver (where there's no solver — compared with folds needed). This applies to bluffs: a bet with a hand that's ahead pays off even without folds.
• Aggression potential — turn and river only, where the pool folds more than needed and more than the solver (where counted), and the difference is not random (green rows); it's an estimate, not an exact cost. Each missed turn and river play is valued as a zero-equity bluff vs this database's pool — pool folds: the pot before the play minus rake; pool doesn't fold: minus what the play put in. A value bet or a bet with equity wins differently, so this is an estimate. The flop is not counted: hands have a lot of equity there, so valuing them as zero-equity bluffs would be wrong.

Pool exploit

• Pool exploit: the solver solved the spot vs a pool with a skew of known size; on the turn and river — for several skew sizes. The gain is the EV of the best response against the pool minus the EV of equilibrium play against the same pool: whatever equilibrium play already wins from the skew is not counted here. If you play like the solver, that part is already in your win rate; where you bet less often than the solver does, it is partly in the underused aggression section. Preflop, the gain is scaled to your pool's skew proportionally, at most 1.5 times the calculated one.
• On the turn and river it uses solver curves for each spot, your role and the flop line (cbet and call, or check-check; the lines are mixed in your own proportion). The gain usually grows faster than the skew, but not everywhere. Between the computed skews (−10 to +20 points) it is a straight line; beyond the ends, the end value. The pool's fold is how opponents respond to bets by every player in your database in this spot and role; we assume the pool folds to your bets the same way. The pool's fold is compared with the solver's fold to the same bet sizes: the solver folds more often to a big bet, and without this a pool that faces bigger bets than the solver uses would look like it overfolds more than it does. So the pool and solver folds here can differ from the underused aggression section: there all sizes are compared at once. The turn and river are separate rows and add up; in the calculation this sum differs from the exploit on both streets at once by −13% to +36%. In the calculation the flop is a small cbet and call, or check-check; other flop lines and all bet sizes are reduced to these two. So the row is approximate.
• The pool's raise against your river bet and the pool aggressor's first river bet use solver curves in which the opponent's frequency of this action is multiplied by 0.75 and 0.5; your multiplier is the pool's frequency divided by the solver's: interpolated linearly between the computed multipliers; below 0.5, the 0.5 value is used. In the calculation, the pool takes this action with the same mix of hands as the solver, just less often with each hand; the removed frequency goes to that hand's other actions: from a raise to call and fold, from a bet to check. The frequency does not show which hands the pool actually bets and raises with. The rows "Pool raises your bet" and "Fold to your bet or raise" in the same spot and role describe the pool's response to the same bet; they are calculated separately and added up. When the pool raises less often, its fold in the calculation rises by 0.7–1.2 pp, so the total is slightly overstated; at the table, both rows point the same way: bet the river more often. Against a pool that bets the river less often, the solver also changes the turn and bets it more often; the advice covers only the river, where the skew is.
• The gain is multiplied by your number of opportunities. The potential takes half of the full exploit: half is realistic to reproduce at the table, the full exploit is harder to reproduce, and the pool may adjust. A notable skew counts: |z| of 2.6 or more and at least 3 points; preflop, only in the same direction as in the calculation. Exploit and underused aggression add up, they are not counted twice: aggression means catching up to the solver's frequency, exploit means going further than the solver vs the pool. The “per calculation …” mark — this spot isn't in the calculation, so the curve of the nearest solved spot is used: an out-of-position aggressor or different ranges respond to the skew differently, so such a row is rougher.
• Pool folds on the turn and river. Calculation: in the solver, the opponent's fold to your turn and river bet or raise is locked and shifted by −10 to +20 points, and the solver found the best response; in SRP the all-in was not shifted: on the river it is an overbet of several pots, and shifting it would inflate the gain several times over; SRP and 3bet pot spots, both roles, two flop lines: cbet and call, check-check; 9 to 14 flops per spot; SRP river by turn line: turn checked through, your turn bet called, his turn bet and your call; each line has its own solver fold, pool skew and gain; turn likewise by flop line: cbet and call, check-check; the "Solver" column shows the solver's fold to the same bet sizes the pool faces, where the solver uses them. Your opportunities: how many times you acted on the turn in this role in this spot; the river uses the same opportunities, because the river gain is computed per time you reach the turn.
• Pool raises on the river and the pool aggressor betting first on the river. Calculation: in the solver, the opponent's frequency of this action is multiplied by 0.75 and 0.5 equally across all of their hands, and the solver found the best response; SRP spots; 14 flops per spot; the "Solver" column shows the solver's frequency in this spot and role. Your opportunities: how many times you acted on the turn in this role in this spot; the river uses the same opportunities, because the river gain is computed per time you reach the turn.
• Preflop, BB 3bet vs an SB open. Calculation: preflop solver, 100 bb, three-handed: BB 3bet vs SB open 18.6% at equilibrium and 13.2% with that frequency locked, SB re-solved; in the “Solver” column — solver 6-max ranges. Your opportunities: how many times you were in the SB with everyone folded to you.
• The pool exploit is played vs all opponents, including those with a small sample (under 500 hands): that is 100% of your encounters at the table.

Pool review

• Reg — all four conditions at once: VPIP 17 to 35%, VPIP–PFR gap no more than 12 pp, limp no more than 1%, 3bet 8 to 15%. Fail any one of them — recreational player. Class is unknown if there wasn't enough sample on the player. Players at 9-max (more than half of their hands at 7–9-seat tables) have no lower VPIP bound: at a full table a reg plays tighter, and VPIP below 17% is normal for them; the other three conditions are the same. Their class stats count only hands where UTG, UTG1 and UTG2 folded: an open or limp from the early seats of a full table lowers anyone's VPIP and 3bet.
• Steal defense — someone called or 3bet after the open; the SB fold and the BB fold taken separately do not show it. Threshold — the pot before the open / (the amount put in by the open + the pot before the open): if they defend less often than the threshold, the steal pays off immediately, even if you give up against any response.
• Pool street aggression — share of hands where the opponent bet or raised at least once on that street, out of all hands where he acted on it; only pots where you two are heads-up. The first column is SRP and 3bet pot together; limped pots and 4bet pots are excluded. The stats table includes all of their pots, including those without you. Normal bands are practical benchmarks for the whole game, not solver ones; in a 3bet pot the pool is more active (narrower ranges, bigger pot), so a 3bet pot is not compared with the band directly.
• Pool vs the solver: shown are spots with a pool sample of 300+ decisions where the skew is noticeable; entering — anything but fold. Red means the pool does this more often than the solver, blue less often; this is a pool tendency, not your mistake. Only decisions the solver has a spot for are compared — defense vs an open no more than 0.3 bb from the solver's size, response to a 3bet at a stack of 70+ bb, response to a 3bet or 4bet within 25% of the solver's size. In responses to 3bet, 4bet and 5bet and to a raise after a limp, the solver's whole range is used here (pool cards are hidden), while your database's preflop uses the hands you reached the spot with; so in the same spot the solver numbers differ.
• Pool folds and the solver: the solver itself may fold more than needed (SRP BTN vs BB on the flop — by 8–9 points), and a pool at its level is no reason to bluff more than the solver: the bluff pays off immediately, but it does for the solver too. “Bluff more” — only where the pool folds more often than both folds needed and the solver, by 5 points or more. Solver fold — to the preflop aggressor's bet of this size in the same spots, averaged over your bets; “—” — the solver barely bets this size or is solved for under half your bets; then compared only with folds needed. Color and the “pays off” conclusion — only where the difference is not random. Where there is no solver fold, the comparison is only with “folds needed”, so “more often” there is a direction, not a proven exploit. Pool folds to your bet: on the flop the sample is your heads-up flop cbet, SRP and 3bet pot; on the turn and river — your first bet on the street as the preflop aggressor (barrel or delayed), heads-up, SRP and 3bet pot; all opponents respond. The review of your game uses a different sample, so the numbers there differ.
• Showdowns on turn lines: the hand as it stood by the river — a draw that got there shows as a made hand.
• The bluff line in pool showdowns is middle pair. Value — top pair and better. Merge — second (middle) pair and pocket pairs above the middle board card. Bluff — air (high card, playing the board, a board pair without a pair of your own), third pair and lower, a pocket pair below the middle card. On a paired board your own pair is compared with the unpaired board cards. The boundary is the same in any pot: the solver rarely bets a weak pair on the river, even in a small pot. Third pair and lower is a bluff by nature: betting it beats almost nothing that calls and works only through folds.
• Showdowns are biased. The call itself does not skew the composition: the caller does not see the cards. But on lines before the river some hands never reach showdown — they're folded to a bet on the next street, or they take the pot with a bet (air can't win the pot without betting), and the loser at showdown may not show cards. All this understates the share of weak hands, more so on lines across several streets; for a called river bet only the muck remains. So the pool's bluff share is more likely understated: the real underbluff is smaller and the overbluff larger than in the table. This kind of review on an individual player needs around forty thousand hands vs him — it's a pool-level tool.
• River bluff by bet size — river bets only (a raise is its own line), bluff — as in “River bluff composition”; sizes with fewer than 120 showdowns are not shown.
• Pool-exploit rules are checked against the solver. “You” — your stat over the whole database that shows whether you exploit the skew (vs overfolding to steals — your steal), in brackets — the same stat for the pool; in “Reason” the pool's postflop is only in solver spots, so the numbers may differ. Preflop — solver ranges seat by seat (steal defense — the share of steals from CO, BTN and SB where someone responded after the open; each position has its own number; 3bet vs steal — the share of steals where the first response was a 3bet: where it is below the solver, the advice for that position is to open wider, even if the pool calls more often; 3bet and fold to 3bet — by position), Solver frequencies are combined with the weight of pool decisions at each seat. Limp — only from UTG–BTN, as in the general stats: a SB limp (the solver has one) is excluded for both the pool and the solver, so the solver shows 0% here. Postflop the solver exists only in solved heads-up spots, so the pool here is also only in those spots, in the same role and with the same spot mix. The rule triggers when the difference is not random (|z| ≥ 2.6) and is at least 3 points. Folds by sizing and spot are compared with folds needed for the bet size and, where the solver has the spot, with its fold to a bet of the same size — whichever is larger.
• Reads and exploit rules are measured against the solver, not the pool average: the pool is the average. Preflop — solver ranges by seat and position; postflop — only in heads-up spots solved by the solver, in the same role; no solver, no conclusion.

Price of findings and potential

• Only mistakes, “losing” spots and not-best moves have a price in bb; the report shows how many there are, and each has its own price, in bb over the database and in bb/100. The prices are not added up into one total. Symptoms, opportunities and deviations are measured in points and have no price. Interval — where the price is only partly checked by calculation: the low end is the rule's price, the high end is the top of the ± over the database hands. Tilt is part of preflop and river folds, so it is not counted among the mistakes with a price. The difference with top regs in the spots with the most money is a direction, not the price of a mistake.
• Pool exploit is the gain from playing against your pool's skew: the solver computed it for a skew of known size (best response against the pool minus equilibrium play against the same pool), here it is scaled to your pool's skew and your number of opportunities. The potential takes half the full exploit: half is realistic to repeat at the table, the full one is harder and the pool may adjust. The same spot can appear in both sections, but the gains are different: aggression means catching up to the solver's frequency, exploit means going beyond the solver against the pool. Underused aggression — aggressive turn and river plays you make less often than the solver where the pool folds more than needed and more than the solver: each missed play is valued as a zero-equity bluff vs this database's pool — pool folds: the pot before the play minus rake; pool doesn't fold: minus what the play put in.

What the report doesn't know

• The sample can't settle the price of most spots: ± is wider than the difference itself. “You lose” — vs the best play; it doesn't have to mean a negative result; “about X bb” — an estimate; only the frequency difference is proven.
• Top regs show how strong regs play, not equilibrium. A difference from them is a reason to think, not proof of a mistake.
• Top-reg data is frozen: “fold to aggressor's bet” is only there when the aggressor acts before you. BB and SB defense vs the opener's cbet is not in the top-reg comparison — only in the stats.
• Multiway pots aren't counted everywhere. River-fold price, the solver check and lines vs top regs — heads-up only. In the top-reg comparison and top 5, multiway is a separate spot; in the sizing tables and folds vs needed it goes together with heads-up pots (a bet took the pot only if everyone folded). Multiway pots are included in the stats.
• Opponents' cards are hidden except at showdown, so here you get frequencies, not ranges. They don't mean “he defends BB too wide” — only “he folds the blind to a steal less often than the solver”.
• Opponent win rates are left out on purpose: over a thousand hands the ± is hundreds of bb/100.
• The pool changes: players leave, adjust, move limits. Rerun the pool review on a fresh database every few months.