⚠️ Review note — remove before publish. Seven hands selected, reconstructed, and solver-analyzed. All seven passed 2-independent-source verification (hole cards, full betting line, outcome). No fabricated first-person Patrik quotes — all Patrik-attributed content is verbatim from cited public sources. This draft lands on internals.quintace.ai for Patrik's review; v2 incorporates any cuts, corrections, and his own quoted reactions.
§1 — Seven hands across the solver era
Patrik Antonius has been at televised cash tables through the entire arc of the solver era. PioSolver shipped in 2015. The pool caught up through 2016-2018. By the time High Stakes Poker returned in 2022, everyone at the table had done solver work. By the time Triton's Cash Game Invitational ran in 2024, the default study tool for any seat at a seven-figure cash game was a laptop open to a GTO simulation.
Patrik has kept playing. And — more interesting — he's kept playing in ways the solver doesn't always endorse.
QuintAce's solver revisits seven of his televised cash decisions from 2014 through 2024. Four of them are clean GTO plays. Three of them are zero-frequency deviations — lines the solver rates as pure mistakes. Two of those three mistakes won the pot anyway, because at this level opponents make the larger mistake.
The seven hands span HSP, Triton, the record-breaking $1.978M pot on No Gamble No Future, and one Aussie Millions hand from 2014 — the year before PioSolver shipped, on the boundary where read-first poker and solver-first poker meet. The pattern across the set is the article's thesis: the modern Patrik plays two games inside the same session, and the solver can tell you which is which.
§2 — Hand 1: The nut-straight check-back (HSP S9E1, 2022)
High Stakes Poker Season 9 Episode 1, live from PokerGO Studio, $200/$400 NLHE cash, ~500bb effective. Koray Aldemir opens the button to $1,200 with A♦J♣. Antonius 3-bets from the small blind to $4,000 with K♣Q♣. Aldemir calls.
Flop 3♠J♠T♥. Antonius has flopped a gutshot and an overcard; c-bets $6,000. Aldemir calls with top pair.
Turn 9♦. Antonius just turned the nut straight. K-Q-J-T-9.
He checks. Aldemir checks behind.
River A♥. Antonius overbets $30,000. Aldemir calls with two pair. Antonius wins $80,800.
Solver verdict on the turn check-back. GTO prefers a bet 73% of the time with precise larger sizes (50bb at 22.2% mix, 37.5bb at 17.1%). Checking is a minority 27.1% line — a legitimate trap variation, but the sub-optimal branch. On a two-tone, straight-complete board with K♣Q♣ unblocking zero of villain's flush draws, the solver rates every hand in villain's continuing range as a value target worth charging. Pot-control is a tactic for marginal made hands. The absolute nuts, on a wet board, vs a range loaded with two-pair and draws, is a bet-bigger-bet-more-often spot by the modern reading.
The live-TV instinct — keep the pot manageable, let Aldemir catch up — is the pre-solver answer. The solver's answer is bet.
§3 — Hand 2: The $1.978M flop 3-bet, 3-way (No Gamble No Future, Feb 2023)
PokerGO's No Gamble, No Future, $1,000/$2,000 with $2,000 BB ante. Antonius opens SB to $30,000 with A♥K♥. Eric Persson calls with Q♥9♥. Richard Yong calls with A♣T♣.
Flop 8♥3♥3♣. Patrik has the nut-flush draw plus overcards. He bets $40,000. Persson raises to $140,000 with a pair plus worse flush draw. Yong has two overs and a backdoor club draw.
Antonius 3-bets to $250,000. Yong folds. Persson calls.
Turn A♠. Antonius bets $150,000. Persson jams $692,000. Antonius calls — now top pair top kicker plus the nut flush draw, ahead of Persson's hand. River J♠ changes nothing. Antonius wins $1,978,000 — the largest pot ever broadcast on a U.S. live stream.
Note: multi-way postflop spot. Solver output on 3-way postflop frequencies is HU-approximation; preflop ranges and equity math are strongly trusted, postflop mixed-strategy frequencies should be read as directional.
Solver verdict on the flop 3-bet. GTO prefers a call. The solver treats the 3-way flop raise-and-3-bet tree as a specialized spot where AhKh's equity is front-runner against the combined range of Persson (pair + draw) and Yong (equity hand). Calling extends the 3-way dynamic and retains the ability to outplay post-turn. 3-betting folds out Yong's equity but turns the hand semi-face-up to Persson, who now plays the rest of the tree knowing Antonius has an overpair or nut-ish draw.
The solver's preferred line wins less often but wins bigger when it does. Antonius's preferred line wins this hand at $1.978M because Persson mis-evaluated the jam on the ace turn — the deviation traded some EV for a clean commitment point, and the clean commitment happened to extract everything.
§4 — Hand 3: The 4-way bottom-boat river jam (HSP S9E13, 2022)
High Stakes Poker Season 9 Episode 13, $500/$1,000 with $1,000 BB ante. Four-way flop. Board runs out 5♣ 9♦ T♥ / T♣ / Q♦.
Antonius has 5♠5♦. Bottom full house on the turn.
Daniel Negreanu has 9♠9♥ — the bigger full house after the turn ten pairs.
Phil Ivey has K♣T♥ — also a full house (tens full of kings).
The river queen adds a fourth plausible boat (tens full of queens) to the board. Everybody who made a full house on the turn or river now has a target on their back.
Antonius checks. Negreanu checks. Ivey bets. Antonius check-raise jams for $153,000. Negreanu folds 99. Ivey folds KT. Antonius wins $311,500 with the WORST full house on the board.
Note: 4-way postflop spot collapsing to HU on the river jam. Solver output strongly trusted for preflop ranges + the HU river jam itself; flop/turn mixed frequencies in a 4-way tree are approximation.
Solver verdict on the river jam. GTO prefers the call, not the raise. The river 4-bet jam with bottom boat is a zero-frequency line — Antonius's jamming range should structurally contain tens-full-of-queens (the one card that made on the river), quads, and the occasional straight. Fives-full is the worst boat on the board; jamming with it folds out every boat that loses to you (fives-over-the-pair) while isolating every boat that beats you.
Negreanu folded the hand that beat Antonius. The two players made cancelling GTO mistakes — Negreanu's fold was a larger error than Antonius's jam, and the chips moved to Antonius. The solver doesn't track "who should win the chips once the mistakes are locked in"; it tracks equilibrium. At equilibrium, this jam is a leak.
§5 — Hand 4: The disciplined AQ fold vs Tsang's bluff (Triton Montenegro 2024)
Triton Cash Game Invitational I, Maestral Resort, Budva. Stakes escalated to $2K/$5K with $5K BB ante across the session. Antonius has A♠Q♠ and plays to a river where Elton Tsang overbets all-in with J♥9♥ on a board that gave Tsang nothing but the right leverage profile.
Antonius folds.
Solver verdict. Fold at 100% GTO frequency. A-high on a paired board vs a polarized overbet jam is a pure fold — no bluff-catcher math saves it. The "pot odds are pot odds, let's see a showdown" reflex that's baked into generations of cash-game players is a leak here, and the solver is unambiguous about it.
Antonius lost the pot. He didn't lose EV. Tsang was bluffing — the hand goes to Tsang because the jam was sized to generate the fold — but "getting shown a bluff on a disciplined river fold" is exactly what a GTO defense range produces. The solver endorses the fold; the result is orthogonal.
This is the cleanest hand in the set for the "Patrik plays GTO when GTO is the right answer" narrative. No read-based deviation, no zero-frequency audacity. The discipline fold.
§6 — Hand 5: The middle-set crying call vs Robl's flopped nuts (Cash of the Titans, 2023)
PokerGO's Cash of the Titans, $500/$1,000 with $1,000 BB ante. Antonius 99 on a Q♠9♥8♠ flop — middle set. Rob Yong and Andrew Robl at the table. Persson limps preflop; Antonius 3-bets; Robl calls with J♣T♣ (flops the nut straight); Persson folds.
Flop action with Antonius's middle set vs Robl's flopped straight is the cooler that writes itself. Turn 3♠ doesn't change anything. River 4♥ bricks. Robl bets $375K into a giant pot. Antonius crying-calls. Robl shows the straight. Antonius loses $1,269,000.
Solver verdict on the river call. The call is correct. More interestingly, against Robl's entire range (not his actual hand), the solver says Antonius should have been stacking off on the turn at 100% — his middle set on a drawy board vs Robl's flop-calling range is a range-vs-range value stack-off. Robl happened to have the single worst hand in his continuing range for Antonius, but that's distribution noise, not a mistake.
The editorial reading is important. Antonius didn't play this hand badly. He played it somewhat conservatively relative to solver-optimal (the solver wanted earlier aggression; Antonius crying-called the river), and he ran into the cooler. Range-vs-range stack-offs don't change their EV because the runout was bad — the fact that Robl had the exact hand that beat 99 was a ~6% occurrence in his range. The loss is variance, not leak.
§7 — Hand 6: The $730K bluff jam vs KK with T4 (Triton Montenegro 2024)
Same table as Hand 4. Antonius is on the straddle — $2,000 posted — with T♥4♥, functionally junk. Andy Ni in the big blind raises to $20,000 with K♦K♣. The table folds to Antonius. He calls.
Flop Q♣7♦7♥. Antonius has a gutshot-ish nothing. Ni c-bets $27,000. Antonius calls.
Turn 5♥. Antonius turns a flush draw. Ni bets $77,000. Antonius calls — priced in on the flush draw plus enough equity from any four, any non-club heart, the straight draws.
River 4♦. Bricks the flush draw, pairs the four. Antonius has bottom pair and no showdown value against any sensible Ni betting range.
Ni bets $265,000 — a 109%-pot river overbet. Antonius jams $730,000 total.
Ni folds pocket kings face-up. Antonius wins $1,269,000.
Solver verdict on the river jam. Zero-percent GTO. Pure mistake. The solver's reasoning is methodical and devastating:
- The straddle's jamming range on this runout structurally contains full houses (56%), sets turned boats (19%), strong two-pair (11%), straights (8%), and quads (4%). T♥4♥ fits none of these categories.
- Blocker profile is catastrophically inverted. A good bluff blocks value combos and unblocks bluff-catchers. T♥4♥ blocks zero of Ni's value combos (no Q, 7, 5, or K in hand) and blocks 22 of Ni's trash combos (the missing-heart-flush-draws Ni might fold). It's the worst possible blocker profile for a bluff.
- Ni needs only 27.2% equity to call at 3.9 SPR with 2.68:1 pot odds — priced in to defend any overpair.
- T♥4♥ has effectively zero showdown value beyond the paired 4.
The solver rates Antonius's jam at 0% frequency. Ni's fold is also a GTO mistake — facing a jam with kings at these pot odds, the fold is worse than the call against Ni's read of Antonius's range. Triton's own recap called Ni "wrong to do so." Two GTO errors, same hand; the larger error was the fold.
This is the article's centerpiece audacity hand. It's the exact opposite of Hand 4's discipline. The solver says you don't do this. Antonius did it, and Ni made the larger mistake. Live read-based cash poker at the outer edge of what equilibrium sanctions.
§8 — Hand 7: The 57.7% hero call vs Polk's bluff-raise (Aussie Millions 2014)
Aussie Millions 2014, live $1,000/$2,000 cash, Crown Casino Melbourne. Effective stacks ~235bb — the largest live-cash sample in the set. Antonius has 8♥6♥ in the big blind. Doug Polk has 7♠4♦ in the small blind.
Polk opens small. Antonius defends 86s. Flop 7♣5♣4♦ — Antonius has an open-ended straight draw and a club flush draw; Polk has top-two-pair. Action gets heavy through the flop and turn (the turn pairs the 5). River Q♠ blanks. Antonius has rivered a straight, 4-5-6-7-8.
He bets $90,000. Polk raises to $302,000. Antonius calls. Polk shows 74o. Antonius wins.
2014 is the boundary. PioSolver hadn't shipped. Patrik was playing a read-based defense with a made hand against a heavily leveraged bluff-raise on a paired board.
Solver verdict on the river call. Solver calls at 57.7% frequency — a clear mixed strategy but firmly on the call side. The reasoning: Polk's range at this node contains genuine value (7-x-paired hands, 44-or-better two pair, straights), but the paired board is a classic board for a bluff-raise because it blocks some of the straights that beat value. Antonius has the exact hand (a rivered straight, not just a bluff-catcher) that needs to call. His range has more low-end-straights + sets than bluff-catchers, so the straight is closer to a pure value call than a bluff-catch.
Antonius made a GTO-approved hero call in 2014 — a year before PioSolver was available for anyone to verify it was the right play. The solver-era tools only caught up to the read.
§9 — The pattern
Seven hands. Four solver-approved, three zero-frequency deviations. Three of the four approved hands are Antonius winning or losing with GTO lines (H1's 27% minority-trap aside — he took the minority branch, but it was a legitimate branch; H4's discipline fold; H5's crying call that the solver actually preferred as a stack-off; H7's hero call at 57.7%). Three of the deviations (H2's 3-bet, H3's boat jam, H6's T4hh jam) are lines the solver rates at 0% or low frequency.
Of the three deviations, three won the pot. None lost.
That's not coincidence, but it's also not a claim about Patrik being "above GTO." The solver's 0% label means the line is -EV at equilibrium — against an opponent whose own strategy is optimized to exploit that line. In each of the three deviation hands (H2, H3, H6), the opponent's response was the larger GTO error: - Persson's jam on the ace turn mis-evaluated Antonius's range-defense (H2) - Negreanu's fold of the bigger full house mis-evaluated Antonius's jamming range (H3) - Ni's fold of pocket kings mis-evaluated Antonius's blocker profile (H6)
Read-driven deviation from GTO is a bet on your opponent making a larger mistake than you. At this level, against a cash-game-invitational field, Antonius has historically been the player whose opponents fold at the wrong moments. The solver can tell you that move is zero-frequency at equilibrium. It can't tell you your opponent is operating at equilibrium.
The discipline hands (H4, H5, H7) are the portfolio's anchor. They're where the solver and Antonius converge and the EV is paid cleanly — the disciplined AQ fold, the conservative 99 crying call that was actually closer to optimal than an overplayed shove would have been, the 2014 hero call that the solver would only validate years later.
The audacity hands (H1 minority-trap, H2 3-bet, H3 bottom boat, H6 T4hh) are the portfolio's optionality. They're negative-EV in isolation, positive-EV against opponents who fold at the wrong moments. Whether they're sustainable at the top of the modern game is a different article.
§10 — Takeaway
The intermediate-to-pro reader's lesson isn't "learn to make the Hand 6 jam." The blocker profile was catastrophically wrong; the solver's 0% frequency is a well-calibrated label. Learning to make that jam means learning to identify the specific game state where your opponent will make a larger mistake than you — and that state is extremely rare outside the top 0.1% of live cash.
The more actionable lesson comes from the discipline hands. Hand 4's AQ fold on a paired board vs an overbet jam is solver-correct and a standard leak in the population. Hand 7's hero call with a made hand on a paired board is solver-correct and a standard fold-to-pressure leak in the population. The EV gap between "reading opponents correctly" (the audacity game) and "making GTO-correct folds and calls" (the discipline game) is smaller at the top than the stories suggest. Most readers will capture more EV by fixing their discipline than by trying to replicate the audacity.
Patrik plays both games. The solver says he gets away with the audacity often enough to make it worth it, at his seat, against his opponents. Your seat probably isn't his seat.
Methodology and caveats
Hand verification. All 7 hands were cross-checked against at least 2 independent sources (primary broadcast + editorial coverage) before inclusion. Source citations per hand: H1 — Upswing + PokerNews + VIP-Grinders; H2 — PokerCoaching + PokerNews + Poker News Daily; H3 — PokerNews + VIP-Grinders; H4 — Triton official + PokerNews; H5 — PokerNews + NYC Poker Tour; H6 — Pokerarena.cz + Triton official + Joey Ingram YouTube + Zeros Poker; H7 — Pokerlistings + fr.pokernews + Upswing. An earlier verification round identified two famous hands in circulation as mis-attributed — the "$273K set-vs-flush" and the "Polk 74 bluff online 2023" (actually Aussie Millions 2014 with the outcome inverted) — both corrected here before publication.
Solver. QuintAce's decision_analysis_tool run on each hand's key decision node. Store IDs in the per-hand solver-output markdown files under articles/patrik-modern-cash/solver-output/.
Multi-way reachability. Hands 2 and 3 involve 3-way and 4-way postflop trees where QuintAce's solver provides HU-approximation outputs rather than pure multi-way equilibrium. Preflop ranges, pot-odds reasoning, and blocker math remain strongly trusted; postflop mixed-strategy frequencies should be read as directional. The per-hand markdown files carry the full reachability notes.
No fabricated quotes. All Patrik-attributed content in this draft is paraphrased narration of his known lines, not quoted speech. Direct Patrik quotes land in v2 after his post-deploy reactions.
Era boundary. Hand 7 (Aussie Millions 2014) sits on the pre-PioSolver boundary; PioSolver shipped 2015. Including 2014 in "modern cash era" is a defensible editorial choice — live $1K/$2K cash with 235bb stacks was structurally the game the solver era optimized for, and the verdict on Antonius's hero call stands up to 2024-era analysis without modification.
Seven hands, five wins, two losses. Solver-approved where he was disciplined. Zero-frequency deviation where he was audacious. Two different games inside the same decade. Patrik reviews the live URL; v2 carries his own reactions in his own voice.