{
 "schema_version": "1.3",
 "table_size": 4,
 "id": "raise-combo-draw-vs-turn-barrel",
 "format": "multi_option",
 "free": true,
 "rebuild": "v2",
 "provenance_class": "dealt_hand",
 "updated_at": "2026-08-09T00:00:00Z",
 "source": {
  "session_id": "CallingTurn",
  "verdict_audio": "quiz/audio/raise-combo-draw-vs-turn-barrel.m4a",
  "lesson_slug": null,
  "master_slug": "CallingTurn_master",
  "master_t": 2551.0,
  "src_t": 2822.3,
  "ser_sentences": [
   538,
   539,
   602
  ],
  "hand_record": {
   "lane": "FRAME_READ",
   "played": "A real hand, played and then reviewed in-session. From t~2588 to t~3160 the shared screen (Tyler's \u2014 he hands the share over at s385, t=2029: 'you should see the new screen') is a PokerTracker hand-history replayer with a typed action-log pane and a FLOP/TURN/RIVER transport bar. He steps the hand to the flop node (t=2590/2600, hero to act facing 8.62), then to the turn node (t=2685 onward, hero to act facing 63.76) and rules on THAT node; at t=2895-2909, after the ruling, he steps the replayer one action FORWARD and the log shows hero's real continuation ('Hero raises to 137.17' \u2014 recorded in hand_record.result, rendered and aired nowhere); at t=3153.9 the felt is stepped BACK to hero's turn CHECK (villain yet to bet) and he delivers the closing line. Frames are READ-ONLY provenance and are never aired: the title bar carries the operator's hand id and the felt carries an operator watermark.",
   "hand_id": "#4340843727 \u2014 read verbatim off the replayer title bar at t=2700 and again at t=3153.9 (identical string, two timestamps 454s apart).",
   "stake": {
    "source": "Replayer title bar, verbatim at t=2700: 'Playback for <room> Hand #4340843727 - $5 NL FAST (6 max) - Holdem' (the room name is withheld from the fixture per the no-operator-branding policy and kept in the audit trail). Corroborated INSIDE the hand by the log's own blind posts: '<P4> posts the small blind 2.50 / Hero posts the big blind 5.00' (crop at t=2590/2600).",
    "usd_per_bb": 5.0,
    "note": "DOLLAR-NATIVE replayer, bb-native fixture. Every displayed BB figure = the on-screen dollar amount / 5.00, the blind printed in this hand's own log. Conversions: open 12.50 -> 2.5; preflop pot 27.50 -> 5.5; flop bet 8.62 -> 1.7 (1.724); pot into the turn 44.74 -> 8.9 (8.948); turn bet 63.76 -> 12.8 (12.752); pot facing 108.50 -> 21.7; hero behind 553.70 -> 110.7 (110.74); villain behind 400.12 -> 80.0 (80.024). Every rounding is within 0.05 BB and the rounded chain reconciles exactly (5.5 + 1.7 + 1.7 = 8.9; 8.9 + 12.8 = 21.7)."
   },
   "frames": [
    "cache/CallingTurn.mp4 t=2590 \u2014 flop node, log scrolled to the TOP of the hand: first line 'Seat 3 (P3-9be6cf4917...) is the dealer'. Felt: 'Pot: $27.50', hero labelled CHECK, villain plate $472.50 (485.00 start - 12.50), P2 plate $1,504.26, P4 plate $522.13, two seat ovals empty.",
    "cache/CallingTurn.mp4 t=2600 \u2014 flop node facing the c-bet: 'Pot: $36.12', 'Odds: 4.19:1 - 19%', settled disc $27.50, villain chip $8.62, hero plate $562.32, villain plate $463.88. Log pane covers 'posts the small blind 2.50' through 'bets 8.62'.",
    "cache/CallingTurn.mp4 t=2685 \u2014 THE DECISION NODE, hero to act (hero plate ringed green): 'Pot: $108.50', 'Odds: 1.70:1 - 37%', settled disc $44.74, villain chip $63.76, hero $553.70, villain $400.12.",
    "cache/CallingTurn.mp4 t=2700 \u2014 same node; title-bar crop 'Playback for <room> Hand #4340843727 - $5 NL FAST (6 max) - Holdem'; log-pane crop covers 'P4 folds' through 'bets 63.76'; board crop = 8(heart) T(heart) K(diamond) 5(diamond) on the 4-colour deck; hero hole-card crop = 7 and 6, both on the diamond blue with the diamond pip.",
    "cache/CallingTurn.mp4 t=2790 \u2014 Tyler's EV work over the decision node (spoken s526-s534, i.e. BEFORE the ruling this fixture airs): the felt is unchanged, plus HoldEq v1.1.7 (Range 1 'KQo' 66.667%, Range 2 '7d6d' 33.333%, Board 'KdTh8h5d') and a Windows Calculator running the raise line \u2014 '180 + 180 + 44' / '404 * 0.33' -> 133.32 / '133.32 - 180' -> -46.68 (the pot his 180 raise builds, his 33% share of it, and the ~46 he is therefore risking; all reproducible single-hand math from figures printed on this replayer). A SECOND Calculator on the same screen multiplies the 107 facing pot by the fold frequency Tyler sources to database analysis at s530-s532. That factor, and every product of it on either calculator, is a measured population magnitude: it is edge-gated under rule 7 and is deliberately NOT transcribed here.",
    "cache/CallingTurn.mp4 t=2825 \u2014 THE FIRST VERDICT ANCHOR (s538/s539, t=2822.3-2828.3, the aired 'So I don't think this thing's terribly close here / I think it has to get raised'): the felt is still the decision node \u2014 'Pot: $108.50', 'Odds: 1.70:1 - 37%', hero $553.70 ringed green \u2014 with the HoldEq and Calculator windows over it. The ruling this fixture airs is spoken AT this node.",
    "cache/CallingTurn.mp4 t=2850 \u2014 the flat's arithmetic in a third Calculator: '63 + 44 + 63' / '126 + 44' -> 170 / '170 * 0.3333' -> 56.661 / '56.661 - 63' -> -6.339 (= -1.27 BB, the EV of calling; reproducible from the displayed price and the displayed 33%). The gated window has advanced by one step; its contents are not transcribed here either.",
    "cache/CallingTurn.mp4 t=2903 \u2014 AFTER the ruling and BEFORE the closer, Tyler steps the replayer forward one action: the log gains 'Hero raises to 137.17', the felt reads 'Pot: $245.67' / 'Odds: 3.35:1 - 23%', hero's plate is $416.53 with a $137.17 chip out and the BTN plate $400.12 is still to act. So hero's actual continuation IS on the frames \u2014 a check-raise to $137.17 = 27.4 BB, NOT the 36 BB Tyler argues hypothetically. Villain's cards are face-up on this replayer (9\u26606\u2660, both grey with the spade pip), which is what the felt's 'Hero - 32%' is computed against: hero has 14 winners of the 44 unseen = 31.8%, recomputed independently. The button's answer is never shown (last log line at t=2909 is still the raise; a solver window covers the replayer from t=2911). Nothing from this frame is rendered or aired \u2014 the fixture's node stays at the 63.76 bet.",
    "cache/CallingTurn.mp4 t=3153.9 \u2014 closing line (s602): the felt is back, stepped one beat BACK of the decision \u2014 'Pot: $44.74', hero labelled CHECK, villain plate $463.88 ringed green (yet to bet), log ending on 'Hero checks'. The screen at BOTH verdict anchors is this hand's replayer, not a solver.",
    "cache/CallingTurn.mp4 t=2980 (context only) \u2014 between the two anchors Tyler opens a solver for a generic BTN-vs-BB theory sidebar (s555-s600). It is cited AFTER the ruling and is NOT the source of this hand; the felt returns at 3153.9. Nothing from that sidebar is displayed by this fixture."
   ],
   "log_verbatim": [
    "Seat 3 (P3-9be6cf4917ebdb888a326372197e57d7) is the dealer",
    "P4-205cf13734d5e5e5754f630ac4328c21 posts the small blind 2.50",
    "Hero posts the big blind 5.00",
    "P2-b76e48a37d91ca17b1409c652baf83f3 folds",
    "P3-9be6cf4917ebdb888a326372197e57d7 raises to 12.50",
    "P4-205cf13734d5e5e5754f630ac4328c21 folds",
    "Hero calls 7.50",
    "Flop: [8h Th Kd]",
    "Hero checks",
    "P3-9be6cf4917ebdb888a326372197e57d7 bets 8.62",
    "Hero calls 8.62",
    "Turn: [5d]",
    "Hero checks",
    "P3-9be6cf4917ebdb888a326372197e57d7 bets 63.76",
    "Hero raises to 137.17",
    "(concatenated across three scroll states: t=2590 covers the dealer line -> 'Hero checks' on the flop; t=2700 covers 'P4 folds' -> 'bets 63.76', the decision node this fixture renders; and t=2895-2909 \u2014 AFTER the aired ruling \u2014 adds the final line 'Hero raises to 137.17', which is hero's real continuation. The log never advances past that line on any frame: a solver window covers the replayer from t=2911, and when the felt returns at t=3153.9 it has been stepped BACK to hero's turn check. The button's response to the raise is therefore not on the tape.)"
   ],
   "seat_derivation": "The title bar says '(6 max)' and the felt draws six ovals, but only FOUR are dealt in (two ovals are empty on every frame). Positions are read off the log text, not inferred: 'Seat 3 (P3-...) is the dealer' => P3 = BTN; 'P4-... posts the small blind 2.50' => P4 = SB; 'Hero posts the big blind 5.00' => hero = BB; P2 is the only other dealt seat and acts first preflop => CO (4-handed). The dealer disc on the felt sits beside P3's seat, agreeing with the log. Rendered exactly 4-handed. The v1 fixture staged a 6-handed ring with invented UTG and HJ folds.",
   "villain_read": "NONE. Tyler states no read on this specific opponent anywhere in the window \u2014 his fold-frequency argument is explicitly about the population ('I know from database analysis...', s530), not about this seat. The v1 fixture's 'REG' name is therefore removed and the villain seat carries no name.",
   "stacks": "Hand start: BTN $485.00 (97.0 BB; 472.50 plate at t=2590 + 12.50 out), hero $574.82 (114.96 BB; 562.32 at t=2600 + 12.50), SB $524.63 (522.13 + 2.50), CO $1,504.26. At the decision: hero $553.70 behind (110.74 BB) = 574.82 - 12.50 - 8.62; BTN $400.12 behind (80.02 BB) = 485.00 - 12.50 - 8.62 - 63.76. Every intermediate plate reconciles (472.50 / 463.88 / 562.32). Note the two players are NOT the same depth, so no single 'N BB deep' line is stated anywhere in this fixture.",
   "result": "Hero's continuation IS on the frames; only the button's answer is not. At t=2895-2909 \u2014 after the aired ruling (s538/s539) and before the closer (s602) \u2014 Tyler steps the replayer one action forward and the log gains 'Hero raises to 137.17' ('Pot: $245.67', hero plate $416.53 with a $137.17 chip, BTN $400.12 still to act). So hero check-raised the turn to $137.17 = 27.4 BB. The log goes no further on any frame (a solver window covers the replayer from t=2911; the felt returns at t=3153.9 stepped BACK to hero's check), so the button's response is not shown and is not narrated. The PokerTracker hand-selector row at the top left reads '7 6  Hero  $84.38' (constant at t=2600, t=2700, t=2790 and t=2903) = the hand's net result, and with the raise now known only the fold reconciles: hero invests 12.50 + 8.62 + 137.17 = 158.29 into a 245.67 pot, so him taking it down is +87.38 gross = +84.38 after a $3 rake, while any called line pays out far more. That last step is arithmetic, not pixels, and is flagged as an inference in unverifiable_claims. NOTHING about the raise or the result is rendered or aired: this fixture's node is the 63.76 bet and it narrates no outcome.",
   "verdict_screens": "At the first anchor (s538/s539, t=2822.3-2828.3; frame t=2825, with the surrounding EV work running t=2790-2850) the felt stays on the decision node with HoldEq and three Calculator windows layered over it \u2014 every input Tyler types is a number printed on this replayer (44.74 pot, 63.76 bet, 107/108.50 facing pot) plus his own proposed 180 raise and the population fold frequency this fixture gates (that one factor is withheld everywhere, including from the frame list above). At the closing anchor (t=3153.9) the felt is alone on screen."
  }
 },
 "context": "$5 NL ONLINE \u00b7 6-MAX TABLE, 4 DEALT \u00b7 BB vs BTN, SINGLE-RAISED POT",
 "hero": {
  "pos": "BB",
  "cards": [
   "7d",
   "6d"
  ],
  "stack_bb": 110.7,
  "avatar": "lono_ghoul",
  "button": false
 },
 "seats": [
  {
   "pos": "CO",
   "avatar": "ghoul_87",
   "folded": true
  },
  {
   "pos": "BTN",
   "avatar": "ghoul_897",
   "stack_bb": 80.0,
   "villain": true,
   "button": true
  },
  {
   "pos": "SB",
   "avatar": "ghoul_138",
   "folded": true
  }
 ],
 "streets": [
  {
   "name": "preflop",
   "actions": [
    {
     "pos": "CO",
     "act": "fold"
    },
    {
     "pos": "BTN",
     "act": "open",
     "to_bb": 2.5
    },
    {
     "pos": "SB",
     "act": "fold"
    },
    {
     "pos": "hero",
     "act": "call"
    }
   ],
   "pot_bb": 5.5
  },
  {
   "name": "flop",
   "board": [
    "8h",
    "Th",
    "Kd"
   ],
   "actions": [
    {
     "pos": "hero",
     "act": "check"
    },
    {
     "pos": "BTN",
     "act": "bet",
     "to_bb": 1.7
    },
    {
     "pos": "hero",
     "act": "call"
    }
   ],
   "pot_bb": 8.9
  },
  {
   "name": "turn",
   "board": [
    "8h",
    "Th",
    "Kd",
    "5d"
   ],
   "actions": [
    {
     "pos": "hero",
     "act": "check"
    },
    {
     "pos": "BTN",
     "act": "bet",
     "to_bb": 12.8
    }
   ]
  }
 ],
 "prompt": "7\u26666\u2666 IN THE BIG BLIND \u00b7 OPEN-ENDER + FLUSH DRAW \u00b7 THE BUTTON OVERBETS THE TURN, 12.8 BB INTO 8.9 \u2014 FOLD, CALL OR CHECK-RAISE?",
 "pot_bb": 21.7,
 "options": [
  {
   "id": "fold",
   "label": "FOLD"
  },
  {
   "id": "call",
   "label": "CALL"
  },
  {
   "id": "raise",
   "label": "CHECK-RAISE"
  }
 ],
 "correct": "raise",
 "option_feedback": {
  "fold": "Too much hand to surrender. Any nine or any four brings the straight and any diamond brings the flush \u2014 fifteen cards improve you on the last card. Tyler never even prices a fold here; the whole question he works through is raise versus call, and he calls that one not terribly close.",
  "call": "The trap answer. At 12.8 to call into 21.7 you need 37% just to break even, and against the offsuit king-queen Tyler assigns the button here you hold 33% \u2014 he runs that exact matchup in an equity tool on screen before he argues the spot. So the flat is a slight loser, and flatting seven-six of diamonds is the precise play he names as the mistake.",
  "raise": "Correct \u2014 'I think it has to get raised.' The size Tyler argues for is a raise to about 36 BB, and it earns twice over. He gives it up and you take the 21.7 already in front of you. He continues and you still hold eight straight cards plus the diamonds \u2014 his own caveat is that the four is the one genuinely clean card, but you are never drawing dead, and that is what caps what the raise can cost. Run his model out and the check-raise breaks even at a fold rate near 30%. His database work on this board type says the real figure sits comfortably past that."
 },
 "verdict_text": "I don't think this thing's terribly close here. I think it has to get raised. I think what would have been a mistake here is for me to flat seven-six of diamonds.",
 "principle": "With a big combo draw, a turn overbet is a raise and not a call: at that price the flat is a slight loser, while the check-raise takes the pot outright often enough to print on its own \u2014 and the draw keeps you live the times he continues.",
 "concept": "check-raise-combo-draw-turn",
 "concepts_secondary": [],
 "topic": "check-raising",
 "difficulty": 4,
 "aired_transcript": "So, I don't think this thing's terribly close here. I think it has to get raised. I think what would have been a mistake here is for me to flat seven-six of diamonds.",
 "aired_transcript_note": "faster-whisper small.en re-ASR of the delivered quiz/audio/raise-combo-draw-vs-turn-barrel.m4a (9.685s, re-cut 2026-08-08), verbatim except two renderings: the ASR writes the ranks as digits ('flat 7 -6') and writes Tyler's reduced 'of' as 'at' ('at diamonds') \u2014 the session SER hears the same reduced vowel at w8300 (conf .86), and 'seven-six at diamonds' is not a phrase, so the display renders 'of'. Beats: s538 + s539 (the ruling, spoken over the replayer parked on the 63.76 turn bet, frame t=2825) + w8285-8301 (the closer at t=3153.9, same felt stepped one beat back). On the re-ASR the lowest word probability is 0.49 (the '7' of '7-6'); every other token is >= 0.56. The third beat is cut as w8285-8301 and SER labels w8285 'Think' at conf .02, with a separately stranded 'I' at w8284 (conf .44) that the cut does NOT reach \u2014 the 0.12s head handle still picks up the swallowed 'I' onset, because SER's boundary on that low-confidence word runs early; both the delivered audio and the re-ASR come back 'I think what would have been\u2026', which is what the display renders. OLD CLIP REPLACED: it aired s533-s534 \u2014 'So even if I never get paid on river' plus Tyler's spoken EV for the 36 BB raise, the figure he reads off the calculator that multiplies the facing pot by the database-sourced fold frequency of s530-s532. With this fixture displaying the price and the raise size, that EV inverts straight back to the population magnitude the rebuild gates, so it is not quoted here either (rule 7); it also had no referent anywhere on the felt, and s533's 'even if' dangled once its premise sentence was cut in the 2026-08-03 rule-7 pass. No number, no size word, no position, no card size and no population magnitude is aired now. [CORRECTED 2026-08-09] This cleanliness assertion was TRUE of this fixture and FALSE of the bundle: the concept card this quiz declares (check-raise-combo-draw-turn in quiz/concepts.json, shipped by build_quiz_bundle.py and rendered on the reveal) carried 'you fold out ~half his range' \u2014 the word-form rounding of the gated figure \u2014 plus 'flatting leaves ~6 BB on the table', the call-versus-raise swing named above. Both were removed from the concept card on 2026-08-09 and validate_quiz.py gained invariant I9 so a concept card can no longer ship a magnitude without declared provenance. The assertion now holds across the whole delivery surface, not just this file.",
 "pct_provenance": {
  "33": {
   "role": "equity_vs_assigned_hand",
   "referent": "hero's 7\u26666\u2666 against the ONE hand class Tyler assigns the button at this node \u2014 offsuit king-queen (s513 'that's basically the hand we're going after is king queen') \u2014 on 8\u2665T\u2665K\u26665\u2666 with one card to come. It is NOT hero's equity against the button's whole barrelling range, and it is not a population figure.",
   "source": "HoldEq v1.1.7 on the shared screen at t=2790/2850: Range 1 'KQo' 66.667%, Range 2 '7d6d' 33.333%, Board 'KdTh8h5d'. Spoken twice: s526 'I'm gonna get back 33 percent of it roughly' and s545 'the reason why this is so damn good is because we have 33 equity'. Reproducible by hand: KQo is 9 combos with K\u2666 on board; 6 of them leave 15 outs and the 3 containing Q\u2666 leave 14, so (6\u00d715 + 3\u00d714)/(9\u00d744) = 132/396 = exactly 1/3."
  },
  "37": {
   "role": "fixture_math",
   "referent": "the equity a CALL needs to break even at the price this fixture displays \u2014 12.8 BB to call into a 21.7 BB pot. A property of the price, not of anyone's hand or of any pool.",
   "source": "recomputed from the fixture: 12.8 / (21.7 + 12.8) = 37.1%. The replayer prints the same figure itself at this node \u2014 'Odds: 1.70:1 - 37%' (t=2685, t=2700), i.e. 63.76 to call into 108.50."
  },
  "30": {
   "role": "fixture_math",
   "referent": "the fold frequency at which the CHECK-RAISE breaks even, using Tyler's own on-screen model of it: raise to 36 BB, 33% equity the times he continues. Nothing measured \u2014 it is what the displayed price and the displayed equity imply.",
   "source": "recomputed in BB from the fixture's own numbers: called, the pot becomes 8.9 + 36 + 36 = 80.9 and hero's share is 0.333 \u00d7 80.9 = 27.0 against 36 invested, i.e. \u22129.0; folded, hero wins 21.7. Break-even fold rate = 9.0 / (21.7 + 9.0) = 29.3%. Tyler runs the identical model in dollars on screen at t=2790 ('180 + 180 + 44' = 404, '404 \u00d7 0.33' = 133.32, '133.32 \u2212 180' = \u221246.68), giving 46.68 / (107 + 46.68) = 30.4%. The MEASURED population fold rate he then substitutes into this model is edge-gated (rule 7): it is not displayed, not aired, not in the slug, and not transcribed anywhere in this fixture \u2014 including in the frame list, which records that the on-screen calculator carrying it exists and withholds its numbers. [CORRECTED 2026-08-09] This cleanliness assertion was TRUE of this fixture and FALSE of the bundle: the concept card this quiz declares (check-raise-combo-draw-turn in quiz/concepts.json, shipped by build_quiz_bundle.py and rendered on the reveal) carried 'you fold out ~half his range' \u2014 the word-form rounding of the gated figure \u2014 plus 'flatting leaves ~6 BB on the table', the call-versus-raise swing named above. Both were removed from the concept card on 2026-08-09 and validate_quiz.py gained invariant I9 so a concept card can no longer ship a magnitude without declared provenance. The assertion now holds across the whole delivery surface, not just this file."
  }
 },
 "size_claims": {
  "_note": "The only size adjective in any displayed string or in the aired clip is 'overbets' (prompt). The remaining entries document the real sizes read off the replayer.",
  "overbet": {
   "street": "turn",
   "actor": "BTN",
   "to_bb": 12.8,
   "pct_of_pot": 142.5,
   "source": "felt chip $63.76 into the settled $44.74 disc ('Pot: $108.50'), t=2685/2700 \u2014 142.5% of pot in dollars, 143.8% after the fixture's BB rounding. Either way an overbet."
  },
  "overbets": {
   "street": "turn",
   "actor": "BTN",
   "to_bb": 12.8,
   "pct_of_pot": 142.5,
   "source": "same claim as 'overbet' \u2014 the prompt uses the inflected form, and the lexicon matches both."
  },
  "preflop open to 2.5": {
   "street": "preflop",
   "actor": "BTN",
   "to_bb": 2.5,
   "source": "log '<P3> raises to 12.50' at 5.00/bb, t=2590"
  },
  "flop c-bet 1.7": {
   "street": "flop",
   "actor": "BTN",
   "to_bb": 1.7,
   "pct_of_pot": 31.3,
   "source": "log '<P3> bets 8.62' into the settled 27.50 ('Pot: $36.12'), t=2600 \u2014 31.3% of pot. No adjective for it appears anywhere in the fixture."
  }
 },
 "claims": [
  {
   "text": "the decision price: the button bets 12.8 BB into a settled 8.9 BB pot, so a call is 12.8 to win 21.7",
   "lhs": "hero's price at the node",
   "op": "=",
   "rhs": "12.8 / (21.7 + 12.8) = 37.1% break-even; in dollars 63.76 to win 108.50, and the replayer prints 'Odds: 1.70:1 - 37%'",
   "source": "pixels t=2685/2700 (chip, settled disc, pot box, odds strip); arithmetic recomputed"
  },
  {
   "text": "pot chain: 5.5 preflop -> 8.9 into the turn -> 21.7 facing the barrel",
   "lhs": "recomputed street pots",
   "op": "=",
   "rhs": "2.5 + 2.5 + 0.5 dead = 5.5; 5.5 + 1.7 + 1.7 = 8.9; 8.9 + 12.8 = 21.7",
   "source": "every on-screen pot box matches after \u00f75.00: 27.50 (t=2590), 36.12 (t=2600), 44.74 and 108.50 (t=2685/2700, t=3153.9)"
  },
  {
   "text": "the turn bet is more than the pot \u2014 12.8 into 8.9 is 143% of it",
   "lhs": "villain's turn sizing / pot before the bet",
   "op": ">",
   "rhs": "63.76 / 44.74 = 142.5% in the replayer's own dollars; 12.8 / 8.9 = 143.8% after rounding",
   "source": "pixels t=2685/2700"
  },
  {
   "text": "fifteen cards improve hero on the last card: eight straight cards (four nines, four fours) plus nine remaining diamonds, minus the two counted twice",
   "lhs": "hero outs with 7\u26666\u2666 on 8\u2665T\u2665K\u26665\u2666",
   "op": "=",
   "rhs": "8 + 9 \u2212 2 = 15 of 46 unseen; the replayer's own equity readout at this node is 32% against the button's actual holding",
   "source": "cards read at t=2700 (hero 7\u26666\u2666, board 8\u2665T\u2665K\u26665\u2666); 'Hero - 32%' felt label; Tyler's caveat that not all of them are clean is s547-s548 ('this thing's actually a hot mess\u2026 the four is about the only card that's actually clean')"
  },
  {
   "text": "the flat is a slight loser at this price against the hand class Tyler assigns",
   "lhs": "EV(call) in BB",
   "op": "=",
   "rhs": "0.333 \u00d7 (21.7 + 12.8) \u2212 12.8 = \u22121.3 BB; Tyler runs it in dollars on screen \u2014 '63 + 44 + 63' -> 170, '170 \u00d7 0.3333' -> 56.661, '56.661 \u2212 63' -> \u22126.339, i.e. \u22121.27 BB",
   "source": "pixels t=2850 (third Calculator window); recomputed from the fixture's own values"
  },
  {
   "text": "the raise size argued is about 36 BB",
   "lhs": "Tyler's proposed check-raise",
   "op": "=",
   "rhs": "'let's say I do raise to 180 here' (s520) at 5.00/bb = 36.0 BB; he keeps using it (s544 'you're never going to lose the full 180 here even if he jams')",
   "source": "spoken s520/s544; the same 180 is the first term of the on-screen Calculator chain at t=2790"
  },
  {
   "text": "the check-raise hero actually made is 27.4 BB, not the 36 BB Tyler argues",
   "lhs": "hero's real turn raise",
   "op": "=",
   "rhs": "$137.17 / 5.00 = 27.43 BB (Tyler's hypothetical is 180 / 5.00 = 36.0 BB)",
   "source": "pixels t=2895-2909 \u2014 log line 'Hero raises to 137.17', felt 'Pot: $245.67', hero plate $416.53 with a $137.17 chip. Provenance only: no raise size is rendered or aired by this fixture, whose node is the 63.76 bet"
  },
  {
   "text": "the ruling is delivered at THIS node",
   "lhs": "screen at the verdict beats",
   "op": "=",
   "rhs": "this hand's replayer, parked on the 63.76 turn bet at t=2685-2850 and stepped one beat back at t=3153.9 \u2014 same title-bar hand id at both anchors",
   "source": "pixels t=2700 and t=3153.9 (identical title bar, 454s apart)"
  }
 ],
 "unverifiable_claims": [
  "Whose hand it is. The replayer labels the studied seat 'Hero' and Tyler argues it in the first person ('let's say I do raise to 180', 'for me to flat seven six of diamonds'), on his own shared screen \u2014 but the frames cannot tell a hand he played from a hand loaded into his database. The fixture claims only that the hand was dealt and reviewed; it names no player.",
  "How the hand FINISHED. Hero's own continuation is not unverifiable and is no longer claimed to be: the log at t=2895-2909 reads 'Hero raises to 137.17' (a 27.4 BB check-raise) and hand_record.result records it. What no frame shows is the button's answer \u2014 the replayer is covered by a solver window from t=2911 and returns at t=3153.9 stepped BACK to hero's check, and no sentence narrates an outcome. The hand-selector row '7 6  Hero  $84.38' reconciles with that raise ONLY if the button folded (+87.38 gross, -$3 rake), but that is an inference from arithmetic, not something any frame states. Nothing about the continuation or the result is displayed or aired.",
  "The measured population fold frequency Tyler substitutes into his EV model (s530-s532, sourced by him to database analysis), and every figure that inverts to it \u2014 his spoken EV for the raise, the call-versus-raise swing, and the calculator products that carry it \u2014 are edge-gated under rule 7. They are not displayed, not aired, not in the slug, and not transcribed in this fixture or its provenance: where the frame list has to mention the calculator windows that carry it, it records that they exist and withholds their numbers. What this fixture teaches is the framework \u2014 the break-even fold rate the DISPLAYED price and equity imply, plus the DIRECTION of his database finding. [CORRECTED 2026-08-09] This cleanliness assertion was TRUE of this fixture and FALSE of the bundle: the concept card this quiz declares (check-raise-combo-draw-turn in quiz/concepts.json, shipped by build_quiz_bundle.py and rendered on the reveal) carried 'you fold out ~half his range' \u2014 the word-form rounding of the gated figure \u2014 plus 'flatting leaves ~6 BB on the table', the call-versus-raise swing named above. Both were removed from the concept card on 2026-08-09 and validate_quiz.py gained invariant I9 so a concept card can no longer ship a magnitude without declared provenance. The assertion now holds across the whole delivery surface, not just this file.",
  "Seat identifiers on the felt plates are truncated ('P3-9be6cf491\u2026'); the full anonymised ids come from the log lines, which are transcribed in log_verbatim."
 ],
 "cosmetic_invention": [
  "Ghoul avatars (lono_ghoul hero, ghoul_897 BTN, ghoul_87 CO, ghoul_138 SB) \u2014 art only; the room shows anonymised ids.",
  "Rendering the table as 4-handed. The real table is a 6-max layout with two seats not dealt in on this hand; the fixture draws the four dealt players and drops the empty ovals. No action is invented for the missing seats \u2014 the v1 fixture's UTG and HJ folds are removed.",
  "The context line's '$5 NL ONLINE' wording \u2014 the stake is sourced from the title bar, the room name is deliberately withheld (no-operator-branding policy).",
  "NOTHING else is invented. The villain carries no name (the v1 'REG' is gone), and every card, size, stack, position label and street action below is sourced \u2014 see sourced_not_invented."
 ],
 "sourced_not_invented": [
  "Stake and blinds: title bar '$5 NL FAST (6 max)' (t=2700) + log 'posts the small blind 2.50' / 'Hero posts the big blind 5.00' (t=2590).",
  "Positions: 'Seat 3 (P3-\u2026) is the dealer' => BTN; 'P4-\u2026 posts the small blind' => SB; 'Hero posts the big blind' => BB; P2 is the only other dealt seat and acts first => CO. Dealer disc on the felt agrees.",
  "Hero's cards 7\u26666\u2666: both cards on the diamond blue with the diamond pip at t=2700, and spoken by Tyler at s602 ('seven six of diamonds'). LOAD-BEARING (the flush draw) and sourced twice.",
  "Board 8\u2665T\u2665K\u26665\u2666: log text 'Flop: [8h Th Kd]' and 'Turn: [5d]' (t=2590/2700), matching the 4-colour felt and HoldEq's own board field 'KdTh8h5d' (t=2790). The two diamonds are LOAD-BEARING and come from the log text, not from card colour alone.",
  "Every size: open 12.50, hero's 7.50 call, flop bet 8.62, hero's 8.62 call, turn bet 63.76 \u2014 all log lines, each cross-checked against a felt chip or pot box.",
  "Both stacks at the decision: hero $553.70 and BTN $400.12, read off the plates at t=2685/2700 and reconciled through every earlier plate.",
  "Hero's two checks (flop and turn): log lines 'Hero checks' twice, plus the felt CHECK label at t=2590 and t=3153.9.",
  "The price: settled $44.74 disc, $63.76 chip, 'Pot: $108.50' box and the replayer's own 'Odds: 1.70:1 - 37%' strip.",
  "The 33% equity figure and its referent: HoldEq window at t=2790 with 'KQo' typed as Range 1 by Tyler, plus s513 naming king-queen as the hand class."
 ],
 "decision": {
  "pot_settled_bb": 8.9,
  "bets": {
   "BTN": 12.8
  }
 },
 "rule7_scope_fix_2026-08-09": {
  "finding": "three assertions in this file certified the gated magnitude was 'not displayed, not aired, not in the slug' \u2014 each true of the fixture, none true of what shipped with it. The leak was one indirection away, in the concept card the fixture names.",
  "root_cause": "no gate had ever read quiz/concepts.json even though it ships in every bundle and renders on the reveal card. Fixture-scoped cleanliness checks cannot see a sibling file.",
  "fix": "concept description rewritten to direction-only; validate_quiz.py I9 now errors on any magnitude in a concept label/description without a declared magnitude_provenance; 5 seeded leaks caught in fault injection; fixes ported into the frozen exporter.",
  "unchanged": "hand, board, sizes, stacks, correct answer, displayed strings, audio"
 }
}