{
 "schema_version": "1.3",
 "table_size": 6,
 "id": "flush-draw-no-fold-equity-check",
 "format": "multi_option",
 "free": true,
 "rebuild": "v2",
 "provenance_class": "dealt_hand",
 "updated_at": "2026-08-08T00:00:00Z",
 "source": {
  "session_id": "MoreHandReviewwithCurrentHands",
  "verdict_audio": "quiz/audio/flush-draw-no-fold-equity-check.m4a",
  "verdict_beats": "RE-CUT 2026-08-08 (dealt_hand rebuild v2). Old beats s406 + s413 + s425/s426 spliced a FLOP argument onto a TURN verdict: s406 ('upwards of 20 percent of the time on this board') is Tyler's flop check-raise frequency read off the Flopzilla range he had just typed (frames t=3610/t=3700) — a figure from a DIFFERENT node than the one this quiz grades — and s413 ('I'm guessing it's checked over 50 percent') is a self-declared guess with no referent on any screen. Both dropped. New beats: w9333-9354 + w9357-9385 (the mechanism: no fold equity, you block the folding region, J9s/T9s are the hands that would fold) + s425 + s426 (the ruling: 'always checking the turn with this hand… zero percent bet'). s425/s426 are cut as two beats so the 3.0s dead gap between them (3812.75-3815.77) never airs. Registered in tools/cut_quiz_audio.py; echo-checked by faster-whisper small.en re-ASR of the shipped m4a — see aired_transcript.",
  "lesson_slug": null,
  "master_slug": "MoreHandReviewwithCurrentHands_master",
  "master_t": 3219.4,
  "src_t": 3755.3,
  "registry_note": "registry/quiz_lessons.json currently carries src_t 3609.6 / master_t 3073.7 — the OLD flop anchor. The values here are recomputed from the same master cutlist (output/lono_staging/full_sessions/MoreHandReviewwithCurrentHands_master.cutlist.json) for the new first beat at src t=3755.31; no lesson cutlist covers this session's verdict, so the map stays MASTER-only. Regenerate the registry with tools/map_quiz_lessons.py --write after this rebuild lands.",
  "coherence_note": "COHERENCE FIX 2026-08-09 (option_feedback.check only; no felt, audio, math or provenance field moved). The panel used to open 'Two reasons, both his. You keep worse hands in, and you BLOCK the folding region…' — a paraphrase of s412 (t=3690.93-3697.05, 'it's a turn check because essentially you want to keep in worse hands and you block some of the folding region'). Tyler DOES say it and he says it about this node, so the line was not misattributed — but it does not survive this rendering: (a) no worse hand that stays in is ever named on the card, (b) the only sub-jack-high combos the quiz names anywhere are J♠9♠ and T♠9♠, which the very next clause says hero's own J♠T♠ removes, (c) option_feedback.bet describes the continuing range as pair-plus-draw combos, every one of which beats jack-high, and (d) the delivered 28.97s clip airs only the fold-equity/blocker reason (re-ASR 2026-08-09 reproduces the aired_transcript word for word), so a player who reads 'two reasons, both his' hears one. s412's first clause is therefore recorded here and deliberately NOT displayed; the blocker reason, which the clip does air, now carries the panel alone. s412 stays in ser_sentences as a sentence drawn on, not aired.",
  "ser_sentences": [
   389,
   393,
   412,
   414,
   415,
   416,
   417,
   418,
   419,
   425,
   426,
   429
  ],
  "hand_record": {
   "lane": "FRAME_READ",
   "played": "A real dealt hand, reviewed on the PokerTracker 4 replayer inside the session screen-share. The replayer is paused on this hand from t=3459 to t=3960: the hand loads with the action log scrolled to the TOP (t=3460-3462), sits on the flop node while Tyler and Eric argue the c-bet (t=3462-3536), cuts away to Flopzilla while Tyler builds the big blind's calling range (t=3541-3705), and returns at t=3712 on the TURN node — which is the node this quiz grades. The turn bet and the check-raise jam land on screen at t=3790, and Tyler's ruling (s425-s426, t=3804-3820) is spoken over that frame.",
   "hand_id": "Replayer title bar and log header both read 'Ignition Hand #4?75422?87' (10 digits, leading 4) — the individual digits are NOT resolvable: the session capture is 1324x720, so the hand id renders at ~3px per glyph. Averaging the 4 unique title-bar encodings across t=3455-3535 (own crop tb_avg.png) did not recover them. Recorded as PARTIAL rather than guessed. Identity is instead anchored by three independent corroborations, below.",
   "identity_corroborations": [
    "(1) Hero's hole cards: pixel zoom at t=3500 (own crop hero_3500.png, 10x) shows two GREY cards with SPADE pips reading T and J = J♠T♠ — and Tyler names the hand on tape three times while this window is up (s404 'and then we have jack ten of spades', s406 'jack ten of spades really wants to get check raised', s413 'it's going to be jack ten of spades there').",
    "(2) Board: the log line reads 'Flop: [Qs 5s 7h]' (own crop log_3806.png) and the felt renders it in the same order in a 4-colour deck — Q on a grey spade card, 5 on a grey spade card, 7 on a RED heart card (own crop board_3500.png, 10x). Tyler types that exact board into Flopzilla at s403: 'is the queen of spades and then what like the seven of hearts five of spades'.",
    "(3) Pot arithmetic closes on the felt's own chip labels: 0.50 (SB) + 1 (the folded limper) + 6 (hero) + 6 (BB) = 13.50 = the settled-pot chip printed at the flop node (t=3500); + 4.20 + 4.20 = 21.90 = the settled-pot chip printed at the turn node (t=3714/t=3750); + 18 + 183.30 = 223.20 = the 'Pot:' readout after the jam (t=3790/t=3806). Three street pots, three independent frames, no slack."
   ],
   "stake": {
    "units": "NO CONVERSION IS PERFORMED OR CLAIMED. This PokerTracker replayer is configured to display every amount natively in BIG BLINDS: the log reads 'posts the small blind 0.50 BB', 'posts the big blind 1 BB', 'Hero raises to 6 BB', 'Hero bets 4.20 BB', 'raises to 183.30 BB'; the felt reads 'Pot: 21.90 BB', 'Pot: 223.20 BB', '91.30 BB', '183.30 BB'. Every bb figure in this fixture is copied verbatim off those BB-native readouts.",
    "title_bar": "'Playback for Ignition Hand #4?75422?87 - $1 NL (6 max) - Holdem' (own averaged crop tb_avg.png from t=3455-3535). The '(6 max)' pins the table size; the dollar stake is NOT used anywhere in this fixture and no $ figure is displayed. The 'Avg Stack: $143' overlay top-left of the replayer is a table statistic, not a stake, and is likewise unused."
   },
   "video_t": [
    3459.0,
    3960.0
   ],
   "frames": [
    "t=3462 (own p_3462.png, log crop log_3462.png at 7x) — action log scrolled to the TOP: 'Ignition Hand #4?75422?87 begins at 2023/1?/?1 4:1? am / Seat 3 (P3-f971e3911…) is the dealer / P4-1abdec911… posts the small blind 0.50 BB / P5-aa8f90f05… posts the big blind 1 BB / P6-3c1ac658a… folds / P1-727ce72b5… calls 1 BB / Hero raises to 6 BB / P3-f971e3911… folds / P4-1abdec911… folds / P5-aa8f90f05… calls 5 BB / P1-727ce72b5… folds'.",
    "t=3500 (own p_3500.png) — FLOP node: board Q♠ 5♠ 7♥, settled pot chip '13.50 BB', hero's bet chip '4.20 BB', 'Pot: 17.70 BB', 'Odds: 4.21:1 - 19%' (the BIG BLIND's price to call, 4.20 into 17.70), hero seat '91.30 BB' (already net of the 4.20), villain P5 seat '187.50 BB' with two face-down cards. Hole-card zoom hero_3500.png = T♠ J♠.",
    "t=3714 and t=3750 (own p_3714.png / p_3750.png, felt crops felt_3714.png / felt_3750.png at 3x) — THE DECISION NODE this quiz grades: board Q♠ 5♠ 7♥ 3♥ (the 3 is a RED card with a HEART pip, board_3806.png at 9x), 'Pot: 21.90 BB' with NO bet in front of either player (the big blind has checked), hero '91.30 BB' on the clock, villain '183.30 BB' still holding cards. Tyler talks over this exact frame from s414 (t=3712.16) through s419.",
    "t=3790 and t=3806 (own p_3790.png / p_3806.png) — THE PUNISHMENT: hero's bet chip '18 BB', villain's chip '183.30 BB', villain seat '0 BB' (all-in), 'Pot: 223.20 BB', 'Odds: 1.79:1 - 36%' (hero's price to call the jam, 73.30 into 204.50 effective), hero '73.30 BB' behind. The replay advances to this state between t=3750 and t=3790 — exactly under Eric's 'and then this happens' (s420) and 'literally the worst thing happened' (s422).",
    "t=3806 log crop (own log_3806.png at 7x) — 'Flop: [Qs 5s 7h] / P5-aa8f90f05… checks / Hero bets 4.20 BB / P5-aa8f90f05… calls 4.20 BB / Turn: [3h] / P5-aa8f90f05… checks / Hero bets 18 BB / P5-aa8f90f05… raises to 183.30 BB'.",
    "t=3610 and t=3700 (own f_3610.jpg / f_3700.jpg) — the FLOPZILLA interlude that the pre-audit fixture mistook for this hand's evidence: a 13x13 starting-hand matrix with a range Tyler hand-marks live, the flop slots loaded Q♠ 7♥ 5♠, and a statistics pane. It is a range-construction tool driven by Tyler's own assumptions (s401-s402 'let's say he went suited broadway, pocket pairs and then serial connectors'), NOT this hand's node — which is why nothing read off it is displayed here."
   ],
   "log_verbatim": [
    "Ignition Hand #4?75422?87 begins at 2023/1?/?1 4:1? am",
    "Seat 3 (P3-f971e3911…) is the dealer",
    "P4-1abdec911… posts the small blind 0.50 BB",
    "P5-aa8f90f05… posts the big blind 1 BB",
    "P6-3c1ac658a… folds",
    "P1-727ce72b5… calls 1 BB",
    "Hero raises to 6 BB",
    "P3-f971e3911… folds",
    "P4-1abdec911… folds",
    "P5-aa8f90f05… calls 5 BB",
    "P1-727ce72b5… folds",
    "Flop: [Qs 5s 7h]",
    "P5-aa8f90f05… checks",
    "Hero bets 4.20 BB",
    "P5-aa8f90f05… calls 4.20 BB",
    "Turn: [3h]",
    "P5-aa8f90f05… checks",
    "Hero bets 18 BB",
    "P5-aa8f90f05… raises to 183.30 BB",
    "(log scrolled to the top at t=3462 and to the bottom at t=3500/t=3806; the two scroll states together cover preflop through the turn jam. The P-hashes are truncated here at the resolution they are legible.)"
   ],
   "positions": "'Seat 3 (P3) is the dealer' + the log's own action order pins every seat with no guesswork: P3 = BTN, P4 = SB (posts 0.50), P5 = BB (posts 1), then clockwise P6 = UTG (folds first), P1 = HJ (limps 1, then folds to the raise), and HERO acts third = CO. Hero is therefore the isolator in position; the big blind is the caller and acts first on every postflop street, which is what the log shows.",
   "hud_reads_gated": "PIPELINE-SIDE ONLY, never displayed (G9). Villain P5 HUD at t=3500: 'VP 18 / PR 16 / 3B 13 / CB 50', stack line 'P5-aa / 193 BB / FCB 100 / (103)'. Hero HUD: 'VP 23 / PR 19 / 3B 17 / CB 89 / Hero / 101 BB / FCB 33 / (103)' — the 101 BB independently corroborates hero's start-of-hand stack (91.30 behind + 4.20 + 6 = 101.50). The fixture displays the LABEL 'REG' only, and that label is Tyler's own word on tape (s389).",
   "result": "Hero bet 18 BB into 21.90 on the turn and the big blind check-raised ALL-IN to 183.30 BB (log t=3806; felt t=3790). HERO FOLDED. CORRECTED 2026-08-09 at the S5 fidelity gate: the pre-gate record claimed the outcome was 'NOT shown' and that the replay 'stays frozen on the decision from t=3790 through at least t=3940'. It is shown. The FELT does stay frozen on the jam, but the action LOG advances — own temporal-averaged crops of the log box show its last line as 'Hero folds' by t=3800 (tmix of 3799.5-3801.0 at 8x), and by t=3830 the box reads '… P5-aa8f90f05… raises to 183.30 BB / Hero folds / 165.30 BB returned to P5-aa8f90f05… / Action has ended / P5-aa8f90f05… wins 57.90 BB' (own crops fulllog_3830.png, wins.png at 10x). The arithmetic closes: 183.30 - 18 = 165.30 returned, 21.90 + 18 + 18 = 57.90 awarded. There is still NO SHOWDOWN — hero folded, so the big blind's cards were never turned over and remain unknown; the replayer's 'Hero - 43%/44%' turn readout is therefore equity versus an UNKNOWN hand, not versus the villain's range, which is one more reason it is displayed nowhere in this fixture. Nothing displayed or aired moves: the quiz stops at the turn decision and grades Tyler's ruling, and the check-raise is displayed only as the wrong-answer consequence, which is what the frames show."
  }
 },
 "context": "6-MAX ONLINE CASH · HERO RAISED OVER A LIMPER · HEADS-UP vs THE BIG BLIND",
 "hero": {
  "pos": "CO",
  "cards": [
   "Js",
   "Ts"
  ],
  "stack_bb": 91.3,
  "avatar": "lono_ghoul"
 },
 "seats": [
  {
   "pos": "BTN",
   "avatar": "ghoul_2348",
   "folded": true,
   "button": true
  },
  {
   "pos": "SB",
   "avatar": "ghoul_599",
   "folded": true
  },
  {
   "pos": "BB",
   "avatar": "ghoul_4",
   "name": "REG",
   "stack_bb": 183.3,
   "villain": true
  },
  {
   "pos": "UTG",
   "avatar": "ghoul_3260",
   "folded": true
  },
  {
   "pos": "HJ",
   "avatar": "ghoul_290",
   "folded": true
  }
 ],
 "streets": [
  {
   "name": "preflop",
   "actions": [
    {
     "pos": "UTG",
     "act": "fold"
    },
    {
     "pos": "HJ",
     "act": "call"
    },
    {
     "pos": "hero",
     "act": "raise",
     "to_bb": 6
    },
    {
     "pos": "BTN",
     "act": "fold"
    },
    {
     "pos": "SB",
     "act": "fold"
    },
    {
     "pos": "BB",
     "act": "call"
    },
    {
     "pos": "HJ",
     "act": "fold"
    }
   ],
   "pot_bb": 13.5
  },
  {
   "name": "flop",
   "board": [
    "Qs",
    "5s",
    "7h"
   ],
   "actions": [
    {
     "pos": "BB",
     "act": "check"
    },
    {
     "pos": "hero",
     "act": "bet",
     "to_bb": 4.2
    },
    {
     "pos": "BB",
     "act": "call"
    }
   ],
   "pot_bb": 21.9
  },
  {
   "name": "turn",
   "board": [
    "Qs",
    "5s",
    "7h",
    "3h"
   ],
   "actions": [
    {
     "pos": "BB",
     "act": "check"
    }
   ]
  }
 ],
 "prompt": "J♠T♠ · SPADE FLUSH DRAW · HE CHECKS THE 3♥ TURN — SEMIBLUFF OR CHECK?",
 "pot_bb": 21.9,
 "options": [
  {
   "id": "bet",
   "label": "SEMIBLUFF"
  },
  {
   "id": "check",
   "label": "CHECK"
  }
 ],
 "correct": "check",
 "option_feedback": {
  "bet": "That is what happened — hero fired the turn and the big blind check-raised all-in. Tyler's test comes first: name the hands that fold. Nothing here does. He called the flop with pair-plus-draw combos, and the 3♥ is a bad card for you — it hands those combos a heart draw instead of a reason to give up. A semibluff that folds nothing out is just building a pot you are behind in.",
  "check": "Correct — Tyler: 'always checking the turn with this hand… zero percent bet.' His reason is blockers. The hands he could get to fold this turn are jack-nine of spades and ten-nine of spades — and your own J♠T♠ removes both, so the folds you would be betting for are sitting in your own hand. Everything else he called the flop with is a pair-plus-draw combo that does not fold, and the one you would most like to push out — ace-five of hearts — just picked up a heart draw on the 3♥ and calls forever. No folds means no fold equity, so check behind in position and take the free card to your flush."
 },
 "verdict_text": "You're not going to generate a ton of fold equity — in fact, in some sense, you actually block the folds here. A hand I'd expect to fold turn would be something like jack-nine of spades, ten-nine of spades; those hands I think I could get to fold. Thousand buy-ins, three thousand buy-ins — always checking the turn with this hand, Eric. Zero percent bet. It's not a mandatory bet.",
 "principle": "Before you barrel a draw, name the hands that fold. When his continuing range is pair-plus-draw combos that never fold — and your own cards remove the two combos that would — the semibluff has no fold equity, so check and keep your equity alive.",
 "concept": "semibluff-negative-fold-equity",
 "concepts_secondary": [],
 "topic": "semi-bluffing",
 "difficulty": 3,
 "aired_transcript": "And you're not going to generate a ton of fold equity. In fact, in some sense, you actually block the folds here, Eric. A hand I'd expect a fold turn would be something like Jack, nine of spades, 10, nine of spades. Those hands, I think I could get to fold. Thousand buy-ins? 3000 buy-ins? Always checking the turn with this hand, Eric. Zero percent bet. It's not a mandatory bet. Yeah.",
 "aired_transcript_note": "Verbatim faster-whisper small.en re-ASR of the DELIVERED quiz/audio/flush-draw-no-fold-equity-check.m4a after this rebuild's re-cut (28.97s, beats w9333-9354 + w9357-9385 + s425 + s426; re-run 2026-08-08). Two ASR artefacts, both benign: 'a fold turn' is the model's rendering of Tyler's 'to fold turn' (SER w9361 reads 'to'), and 'Jack, nine of spades, 10, nine of spades' is its numeral rendering of 'jack-nine of spades, ten-nine of spades'. Echo check vs the intended SER words: every other word matches. Content check — nothing aired is off-node or gated: no percentage with a figure ('zero percent' carries no number), no population or database magnitude, no cards beyond the two combos the blocker argument is about, no positions, no stack sizes, no bet sizes, and no street the rendered hand does not show (the clip names only the turn, which is the decision street). The two figures the OLD clip aired — the flop check-raise frequency (s406) and the 'over 50 percent' guess (s413) — are gone.",
 "pct_provenance": {},
 "size_claims": {},
 "claims": [
  {
   "text": "units: the replayer prints every amount in BB, so no conversion exists in this fixture",
   "check": "log lines 'posts the small blind 0.50 BB' / 'posts the big blind 1 BB' / 'Hero raises to 6 BB' / 'Hero bets 4.20 BB' / 'Hero bets 18 BB' / 'raises to 183.30 BB' (own crops log_3462.png, log_3806.png) and felt readouts 'Pot: 21.90 BB', 'Pot: 223.20 BB', '91.30 BB', '183.30 BB' (own frames p_3714/p_3750/p_3790/p_3806)"
  },
  {
   "text": "preflop: HJ limps 1, hero raises to 6 from the CO, only the big blind calls; settled pot 13.5",
   "check": "log at t=3462 scrolled to top: P6 folds, P1 calls 1 BB, Hero raises to 6 BB, P3 folds, P4 folds, P5 calls 5 BB, P1 folds. 0.50 + 1 + 6 + 6 = 13.50 = the settled-pot chip printed at the flop node (t=3500)"
  },
  {
   "text": "flop Qs 5s 7h: the big blind checks, hero bets 4.2, the big blind calls; settled pot 21.9",
   "check": "log 'Flop: [Qs 5s 7h] / P5 checks / Hero bets 4.20 BB / P5 calls 4.20 BB'; the felt's 'Pot: 17.70 BB' at t=3500 = 13.50 settled + 4.20 in front, and after the call the turn frame prints the settled chip '21.90 BB' (t=3714, t=3750). 13.50 + 4.20 + 4.20 = 21.90"
  },
  {
   "text": "decision node: the 3h turns, the big blind checks, hero has 91.3 behind and the big blind 183.3",
   "check": "own frames p_3714.png / p_3750.png — board Q♠ 5♠ 7♥ 3♥, 'Pot: 21.90 BB', no chips in front of either player, hero seat '91.30 BB', villain seat '183.30 BB' with cards still live; log line 'Turn: [3h] / P5 checks'"
  },
  {
   "text": "the 3h is a heart, so it adds a heart draw to combos like ace-five of hearts",
   "check": "board suits are pixel-read on a 4-colour deck (own crop board_3806.png at 9x): Q and 5 grey with spade pips, 7 and 3 RED with heart pips. With 7♥ and 3♥ on board, A♥5♥ — the exact combo Tyler names at s416 — holds four hearts. Tyler's own summary of the card: s417 'the turn card is just bad for you'. (s417 spans words 9301-9357, t=3743.71-3761.81, and carries BOTH quoted fragments — 'where the turn card is just bad for you' AND 'in some sense you actually block the folds here', the latter cited in the next claim. Re-checked against the word bank 2026-08-09: the shared sentence id is correct, not a bookkeeping slip.)"
  },
  {
   "text": "hero's J♠T♠ removes both combos Tyler names as the folds (jack-nine of spades, ten-nine of spades)",
   "check": "J♠9♠ needs the J♠ and T♠9♠ needs the T♠; hero holds both (hole-card zoom hero_3500.png). Tyler s418 'a hand I'd expect to fold turn would be something like jack nine of spades, ten nine of spades — those hands I think I could get to fold', immediately after s417 'in some sense you actually block the folds here'"
  },
  {
   "text": "the wrong answer's consequence: hero bet 18 into 21.9 and was check-raised all-in to 183.3",
   "check": "log 'Hero bets 18 BB / P5 raises to 183.30 BB'; felt at t=3790 shows hero's chip '18 BB', villain's chip '183.30 BB', villain seat '0 BB', 'Pot: 223.20 BB' = 21.90 + 18 + 183.30"
  },
  {
   "text": "villain label REG is Tyler's spoken word, not a HUD read",
   "check": "s389 (t=3459.5, spoken as this hand loads on the replayer) 'pretty reg caller okay'; corroborated in kind by s393 'his range is going to be really tight, this is closer to a small blind range'. The HUD numerals exist in the frames and are deliberately NOT displayed (hand_record.hud_reads_gated)"
  }
 ],
 "unverifiable_claims": [
  "The operator hand id is not resolvable at the session capture's 1324x720 resolution ('#4?75422?87' — 10 digits, leading 4). Identity rests on the three corroborations in hand_record.identity_corroborations instead; nothing displayed depends on the id.",
  "The big blind's hole cards are never shown — hero folded to the jam, so there is no showdown and no way to know what he check-raised with. (Hero's OWN action is shown and is recorded in hand_record.result: the log reads 'Hero folds' from t=3800. The pre-gate version of this entry wrongly listed hero's fold as unverifiable; corrected 2026-08-09 at the S5 gate. Nothing displayed or aired depends on either fact.)",
  "Tyler's second no-fold example at s415 transcribes as 'seven six two' in both the SER and an independent small.en re-ASR; whether the trailing token is a suit, a second hand or the word 'too' cannot be resolved from the audio, so that example is neither displayed nor aired. The unambiguous example he gives at s416 (ace five of hearts) is the one used.",
  "The dollar stake in the title bar reads '$1 NL (6 max)' but only the '(6 max)' is used. No dollar figure appears anywhere in this fixture and no bb↔$ conversion was performed.",
  "Tyler's flop-side figures — s406's check-raise frequency and s413's 'I'm guessing it's checked over 50 percent' — belong to the Flopzilla range he typed and to a street this quiz does not grade. They are recorded here as deliberately EXCLUDED, not as claims."
 ],
 "cosmetic_invention": [
  "Ghoul avatars on all six seats (lono_ghoul for hero) — art only; the real table showed anonymised P-hash labels and HUD boxes.",
  "Nothing else. Every displayed fact — the six-handed structure, the dealer seat and every position, both hole cards and all four suits, the full board with its suits, every bet size (1 / 6 / 4.20 / 18 / 183.30), both stacks at the decision, all three street pots and the check-raise result — is read off this rebuild's own frames."
 ],
 "sourced_not_invented": [
  "Hero J♠T♠ — pixel zoom hero_3500.png (grey cards, spade pips) + Tyler names 'jack ten of spades' at s404/s406/s413. The spades are LOAD-BEARING (the flush draw, and the J♠/T♠ blockers in the aired verdict) and are sourced twice over.",
  "Board Q♠ 5♠ 7♥ 3♥ — log lines 'Flop: [Qs 5s 7h]' and 'Turn: [3h]' plus the 4-colour felt (board_3500.png, board_3806.png). The heart suits are LOAD-BEARING (the ace-five-of-hearts draw) and come from the log, not from us.",
  "Positions (hero CO, villain BB, BTN/SB/UTG/HJ) — derived from 'Seat 3 (P3) is the dealer' plus the log's action order; see hand_record.positions.",
  "Pot type: an isolation raise over one limper, NOT a 3-bet pot — 'P1 calls 1 BB / Hero raises to 6 BB / P5 calls 5 BB' in the log.",
  "Sizes 6 preflop, 4.2 on the flop, 18 on the turn and the 183.3 jam — log text, cross-checked against the three printed pots.",
  "Stacks at the decision (hero 91.3, villain 183.3) — seat readouts at t=3714/t=3750.",
  "Villain label 'REG' — Tyler s389 'pretty reg caller'."
 ],
 "decision": {
  "pot_settled_bb": 21.9,
  "bets": {}
 }
}