Final Table Deals: ICM vs. Chip-Chop vs. ICM-with-a-Save

A final-table deal can swing thousands of dollars in seconds — far more than hours of play. Here's how chip-chop, ICM, and ICM-with-a-save actually differ, and how to negotiate from either side of the table.

You ground twelve hours through a 2,000-runner field, dodged three flips, and now you're three-handed for life-changing money. Somebody at the table says the word every pro wants to hear and every amateur fears: "Do you guys want to talk numbers?"

The next ninety seconds may be worth more EV than the entire day of play. A deal can move four figures from one stack to another based purely on which model the table agrees to use — and most players don't even know which model favors them. This is one of the most underpriced skills in tournament poker: you don't have to out-play anyone to win the negotiation, you just have to know the math better than they do.

There are three models you'll see at a final table. Let's make each one precise, run the actual numbers, and then talk about how to argue your corner.

The setup we'll use throughout

To keep everything concrete, we'll work one example the whole way down. Three players left, the remaining prize pool to be divided is:

Chip counts (10,000,000 chips in play):

| Player | Chips | Share of chips | |---|---|---| | Big Stack | 5,000,000 | 50% | | Middle | 3,000,000 | 30% | | Short | 2,000,000 | 20% |

These figures are illustrative — the structure of the result is what generalizes, not the exact dollars.

Model 1: Chip-Chop (proportional to chips)

A chip-chop is the simplest possible deal: each player takes a share of the remaining pool proportional to their current chip stack.

Clean, fast, and the table can do it in their heads. There's just one problem: it treats chips as linear money. It assumes that owning 50% of the chips means owning 50% of the money. In a winner-take-all sit-and-go that would be roughly defensible. In a tournament with a flat-ish payout ladder, it is simply wrong — and it is wrong in a directional way that always benefits the same player.

Notice what the pure chip-chop did to our short stack: it valued his 20% of the chips at exactly the 3rd-place prize, $4,000. But the short stack isn't guaranteed third. He can ladder up. He can double through, he can win. A model that hands him the bottom prize and calls it fair is quietly transferring his ladder equity to the chip leader.

That's why a straight chip-chop is the chip leader's favorite proposal — and why you'll often hear it come out of the biggest stack's mouth first.

Model 2: The ICM Deal (proportional to ICM $-equity)

The Independent Chip Model fixes exactly that flaw. Instead of treating chips as money, ICM converts each stack into its dollar expectation — the average payout that stack would earn if the tournament were played out from here, assuming every player's probability of finishing in a given position is proportional to their chips at each step (the Malmuth-Harville formulation).

Running ICM on the same stacks and the same $10k / $6k / $4k ladder gives each player's ICM $-equity:

| Player | Chips | Chip-Chop | ICM $-equity | ICM − Chip-Chop | |---|---|---|---|---| | Big Stack | 50% | $10,000 | $7,679 | −$2,321 | | Middle | 30% | $6,000 | $6,550 | +$550 | | Short | 20% | $4,000 | $5,771 | +$1,771 | | Total | 100% | $20,000 | $20,000 | $0 |

Both models distribute exactly $20,000 — the pool is conserved. But look at how differently they slice it.

The big stack's 50% of the chips is worth only $7,679 under ICM, not $10,000. Why? Because he can't actually win $10,000 worth more often than his chips suggest — the payout ladder is compressed. The jump from 3rd ($4k) to 1st ($10k) is only 2.5×, while his chip lead over the short stack is 2.5×, but a huge chunk of his stack's "extra" chips are spent buying him a marginally better shot at a prize that isn't proportionally bigger. Meanwhile the short stack is guaranteed at least $4,000 no matter what, and has real equity in finishing 2nd or 1st. ICM prices that survival correctly.

The headline: chip-chop overpays the leader by ~$2,300 and underpays the short stack by ~$1,800 versus ICM. That gap is the entire negotiation. It is not a rounding error — it is more than 40% of the short stack's "fair" number sitting on the line, decided by which word the table agrees to.

ICM is the correct baseline. It's what every serious tournament player, every solver, and every reputable deal-calculator uses. If you take nothing else from this article: when someone proposes a chip-chop and you are not the chip leader, you are being asked to donate your ladder equity.

Why the shorts gain and the leader loses — the intuition

Chips win you tournaments; money is what survival buys. Each additional chip is worth less than the last one (diminishing returns), because no matter how many chips you pile up, you can only win first place once. ICM captures this concavity; chip-chop assumes a straight line. The bigger your stack, the more your chips are over-valued by the linear assumption — so the leader always prefers chip-chop, and everyone shorter than the leader always prefers ICM.

Model 3: ICM-with-a-Save (leave money on the table for 1st)

There's a cultural objection to pure ICM deals: if you ICM everything, nobody is actually "the champion" — the trophy, the title, and the bragging rights get divorced from a meaningful payday. Players who came to win don't love locking up a number and then playing a meaningless freeroll for a title that pays the same as the deal.

The fix is the save (also called "leaving money on the table"): take an ICM deal on most of the pool, but set aside a chunk to still be played for by the finalists — usually so that 1st place still means something.

Mechanically, here's the clean way to do it with our example. The table agrees to set aside $2,000 of the $20,000 to keep playing for, and to ICM the remaining $18,000 now. To ICM the reduced pool, you scale the ladder down proportionally (×0.9): $9,000 / $5,400 / $3,600. Running ICM on $18,000:

| Player | Locked (ICM of $18k) | Plays on for share of the $2k save | |---|---|---| | Big Stack | $6,911 | + still live for the $2,000 | | Middle | $5,895 | + still live for the $2,000 | | Short | $5,194 | + still live for the $2,000 | | Total locked | $18,000 | + $2,000 played out = $20,000 |

Everyone pockets their locked ICM number immediately, then plays on for the remaining $2,000 — typically winner-take-all, so 1st place is now worth their locked amount plus $2,000. The size of the save is itself negotiable: a small save ($1k–$2k here) keeps variance low and is basically an ICM deal with a trophy attached; a large save is closer to "deal the min-cash, play for the rest."

The save is the most common deal you'll see among strong, ego-aware players, because it solves the human problem (someone wants to be champion) without re-introducing the math problem (chip-chop robbing the shorts). It's the diplomatic middle: ICM-fair on the money that matters, with just enough left on the felt to keep the title alive.

How to negotiate — from either side

When to deal at all

Deal when these line up:

Play on when:

If you're the big stack

Your equity is highest in chip-EV terms, and you want to capture as much of it as possible:

If you're the short or middle stack

Pitfalls — read this before you shake hands

The core takeaway

At a final table you are frequently negotiating for more EV than hours of play could ever produce, and the entire result hinges on which model the table adopts. Chip-chop pays the leader; ICM pays survival; the save splits the difference with a trophy on top.

Know which one favors you before you open your mouth, run the real ICM numbers rather than trusting the table's arithmetic, and — whichever seat you're in — don't leave money on the table by accident. Leave it there only on purpose, as a save, with the figures in front of you.