Alex Theologis folded bottom pair on ten-ten-seven. In chips it was close. ICM made it easy.

Second in a series: one hand from a stream that aired in the last few days, checked against real numbers rather than opinions. Every figure below is either read off the broadcast graphics or computed here, and the assumptions are stated.

The hand

Triton Poker Series Jeju II, $15,000 NLH 8-Handed, Event #1, final day. Seven players left from 206 entries. Blinds 150,000/300,000 with a 300,000 big-blind ante.

• Punnat Punsri, hijack, 6,700,000 (22 big blinds), holding J♦J♥.
• Alex Theologis, big blind, 5,200,000 (17 big blinds), holding 7♦5♦.

Punsri opens to 600,000. Theologis defends his big blind. The pot is 1,650,000 and the broadcast has Punsri a 79% favourite.

Flop: 10♦ 7♠ 10♠. Theologis has flopped bottom pair with a backdoor flush draw. He checks. Punsri bets 375,000, a shade under a quarter of the pot. Theologis check-raises to 925,000. Punsri makes it 1,600,000.

That is the decision. The pot is 4,175,000 and Theologis owes 675,000 with 4,275,000 behind. He folds. Punsri scoops and moves to 9,300,000 and the chip lead; Theologis slides to fifth of seven.

The commentary called it a sharp fold and moved on. It is worth stopping, because the fold is much closer than it looks and the reason it is correct has almost nothing to do with the cards.

What the chips said

Call 675,000 to win the 4,175,000 already in the middle. That is the whole of the immediate arithmetic:

675,000 ÷ (4,175,000 + 675,000) = 13.9%.

Theologis needs to win this pot 13.9% of the time to break even in chips. Against the hand Punsri actually held, the broadcast put him at 12%. Against Punsri's whole re-raising range — which on a paired board contains some number of spade bluffs alongside the tens, sevens and overpairs — bottom pair with a backdoor flush draw is plausibly a little over that bar rather than a little under it.

In other words: if the chips in front of him were dollars, this is a coin-flip of a decision. Somewhere around break-even, decided by how many bluffs you credit the hijack with. That is not a fold anyone would call sharp.

What ICM said

They are not dollars. Seven players are left and the payouts, from the broadcast's own payout tab, run $108,000 for seventh up to $642,000 for the winner, with jumps of roughly $50,000 at every step from sixth to third. Sixth place had $147,000 locked up.

Run that ladder through the Independent Chip Model — the same Malmuth–Harville calculation behind our final-table ICM charts — and price the three outcomes in prize equity rather than chips:

Folding is worth $275,600. Winning gains him $88,000; losing costs him $18,100. The equity he needs to justify the call is no longer 13.9% but

$18,100 ÷ ($18,100 + $88,000) = 17.1%.

Three percentage points more than the chips demanded, on a bar 23% higher. That is the entire distance between a marginal call and a fold nobody has to think about, and it is created by the payout ladder alone — the cards on the table never enter into it. Note where it leaves him against the hand Punsri actually had: 12% equity clears neither bar, but it is the ICM number that turns "close" into "not close".

That number is stubborn. The other five stacks are not visible on the broadcast, so we reconstructed them three ways, from a flat field to a top-heavy one, and ran both the broadcast's payout ladder and the solver's standard final-table shape. Every combination returns between 17.0% and 17.3%. It is also conservative: the model treats the call as ending the hand, when in reality Theologis would face more bets on the turn and river with 3,600,000 behind. The true requirement is higher still.

Why ICM squeezes the defender and frees the aggressor

The usual summary of ICM — "play tighter near the money" — is wrong in a way this hand exposes. ICM does not make everybody tighter. It makes the player facing aggression tighter, and that is precisely what lets the player applying it get wider.

Our computed jam-or-fold charts show the effect cleanly. At 15 big blinds, eight-handed, here is how the first-in shoving range moves between chip EV and the two ICM situations:

The early seats tighten. The late seats, the ones acting last with nobody able to afford a call behind them, get dramatically wider — the button goes from a third of hands to 84% on the bubble. Same force, opposite signs, depending on which side of the bet you are standing.

Theologis was on the wrong side of it. Punsri did not need a strong hand for the re-raise to be profitable; he needed Theologis to be a player with 17 big blinds and $50,000 pay jumps behind him. As it happens Punsri had jacks and did not need the help.

The board mattered too

One more thing worth naming, because it is what makes the check-raise tempting in the first place. On a paired board the preflop aggressor bets nearly his entire range for a small size. Our library's paired-board c-bet page has the solver betting 99% of hands on K♠K♦3♣, and 98% of those bets are the small one.

Punsri's 375,000 into 1,650,000 is exactly that bet. It means almost nothing, which is why raising it looks like free money. But the same fact cuts the other way: when a range that bets everything small then re-raises, what is left is not a range you get to bluff-catch against with bottom pair. The small bet is weak; the raise behind it is not.

What to take from it

• Work out the chip price first, then add the tax. 13.9% was the honest starting point. Everything after that is the payout ladder, and it was worth three points.
• ICM has a direction. It tightens whoever is facing the bet. If you are the one betting with players behind you who cannot afford to be wrong, you are being handed a discount — take it.
• A wide small bet is not a wide raise. Ranges that bet everything cheaply on paired boards are still narrow when they put in a second raise.
• Marginal in chips is a fold at a final table. If the chip-EV decision is close, the ICM decision is not close, and it is not close in the direction of folding.

Run the numbers yourself

The pot-odds half of this is one line of arithmetic: our pot odds calculator turns any bet and pot into a required-equity percentage. The prize-equity half is what the ICM calculator is for — feed it the stacks and the payouts and it prices each stack in dollars, which is the number every final-table decision is actually made in.

For the ranges: the final-table and bubble push/fold charts show the chip-EV and ICM jam ranges side by side for every seat and stack from 5 to 20 big blinds, and the page for 75 suited covers how the hand plays from every position before any of this starts.

Source: Triton Poker's official stream of the $15K NLH 8-Handed Event #1 final day, Jeju II 2026, published 5 September 2026; the hand begins at 6:20:20. Stacks, blinds, hole cards, bet sizes and the equity percentages are read from the broadcast graphics. Payouts are from the payout figures stated on the broadcast, with second place interpolated. The five stacks not shown on screen are reconstructed; the required-equity result varies by less than 0.3 points across every reconstruction we tried. ICM is computed with the Malmuth–Harville model.