The first time ICM cost me money, I didn't even know it existed. Four of us left, three getting paid, and I looked down at pocket jacks with a middling stack. I shoved, the chip leader called with ace-ten, and I busted on the bubble for nothing. In a cash game that shove is fine. On that bubble it was a disaster — and the reason why is the single most important idea in tournament poker.
Chips in a tournament are not money. You only ever win one first prize, so doubling your stack never doubles what you're actually worth. ICM — the Independent Chip Model — is the math that turns your pile of chips into the real dollars it represents, and once you see it, calls and folds that felt wrong suddenly make sense. This guide takes you from "what does ICM stand for" all the way to splitting a final-table deal, with every number worked out so you can check it yourself.
ICM lives inside the 
tournament game specifically — it's the reason late-stage MTT play looks nothing like a cash game.
ICM at a glance
What Is ICM in Poker?
ICM (the Independent Chip Model) converts a chip stack into its real prize-money value, using the remaining payouts and every player's stack size. It answers one question: if the tournament ended right now with these stacks, what is my slice of the prize pool actually worth in dollars?
It works by estimating how often each player finishes in each paying position — first, second, third, and so on — from their share of the chips, then multiplying those probabilities by the payouts. The bigger your stack, the more often you finish high; but because the top prize is capped, extra chips buy less and less money.
The key mental shift: in a cash game a chip is a dollar, full stop. In a tournament a chip is a lottery ticket on a fixed set of prizes. ICM prices that ticket. It applies to tournaments and sit-and-gos only — 
never to cash games, where your chips already equal their face value.
Why Your Chips Aren't Worth Their Face Value in Money
Because prize money is "top-heavy" and locked in below you. Say three prizes pay $50 / $30 / $20. The moment you're in the money, you're guaranteed at least $20 — so the chips protecting that $20 are precious, while the chips reaching for first are chasing a prize you can only win once.
That makes the chip-to-money curve bend: the first chips (survival) are worth a lot, the last chips (going for the win) are worth less. A player with half the chips does not own half the prize pool — they own noticeably less, because they can't finish better than first but they can still bust.
Flip it around and the short stack is the winner of this math. They already have a real claim on the pay jumps below them, so each of their chips is worth more than face value. This single asymmetry — big stack overvalued in chips, short stack undervalued — drives every ICM decision you'll ever make.
How Is ICM Calculated? (The Malmuth–Harville Model)
ICM assigns each player their probability of finishing in each position based purely on stack size, then multiplies by the payouts. The method is often called the Malmuth–Harville model — the finish-probability math comes from David Harville's 1970s work on horse-racing odds, which Mason Malmuth applied to poker.
The rule is simple and recursive:
- •Your chance of finishing 1st = your stack ÷ total chips.
- •Your chance of finishing 2nd = the sum, over every other player who could finish 1st, of (their chance of winning) × (your stack ÷ the chips left without them).
- •Keep going for each lower position.
| Finish | Leader (5,000 · 50%) | Middle (3,000 · 30%) | Short (2,000 · 20%) |
|---|---|---|---|
| 1st | 50.0% | 30.0% | 20.0% |
| 2nd | 33.9% | 37.5% | 28.6% |
| 3rd | 16.1% | 32.5% | 51.4% |
Take the leader's 2nd-place number so you can see the recursion: if Middle wins first (30% of the time), the leader then takes 5,000 of the 7,000 chips left = 71.4%, and 0.30 × 0.714 = 21.4%; if Short wins first (20%), the leader takes 5,000 of 8,000 = 62.5%, and 0.20 × 0.625 = 12.5%. Add them: 33.9% of the time the leader finishes 2nd.
Now multiply each row by the payouts and you get the dollar value of every stack:
| Player | Chip % | ICM value | ICM % | vs chips |
|---|---|---|---|---|
| Leader | 50.0% | $38.39 | 38.4% | −11.6 |
| Middle | 30.0% | $32.75 | 32.8% | +2.8 |
| Short | 20.0% | $28.86 | 28.9% | +8.9 |
There it is in numbers: the leader has half the chips but only 38.4% of the money, while the short stack's 20% of chips is worth 28.9%. You don't have to run this by hand at the table — the ICM calculator does it instantly — but seeing the machinery once is what makes the strategy stick.
ICM vs Chip EV — What's the Difference?
Chip EV measures a decision in chips won or lost; ICM (or "$EV") measures it in real prize money. They agree early and split hard late. Early in a tournament, with tiny pay jumps far away, a chip is basically a chip — you play 
chip EV, accumulating relentlessly. Near the money and the final table, ICM takes over.
The classic clash is a marginal all-in. In chip EV, a coin flip for a big pot can be fine or even good — you gain as many chips as you lose. In ICM it can be a clear fold, because busting costs you locked-in prize money you can't get back, while the chips you'd win are worth less than face value.
That is exactly the mistake I made with those jacks: a fine chip-EV shove, a losing ICM shove. Chip EV asks "will this build my stack?" ICM asks "will this build my bankroll?" — and only the second one pays out.
The "ICM Tax": Why Losing Chips Hurts More Than Winning Helps
The "ICM tax" is the gap between your chip percentage and your real money percentage — value that vanishes the moment stacks get top-heavy. In the worked example, the leader's chips say 50% but the money says 38.4%: an 11.6-point ICM tax on being the big stack.
The tax shows up in every all-in as a risk premium — the extra equity you need on top of the chip-EV break-even before a call is actually profitable in dollars. If chip math says you need 40% to call, ICM might demand 48-50%, because the downside (busting, losing pay-jump equity) outweighs the upside (chips worth less than face).
The player who feels this most is the medium stack on the bubble — big enough to have real equity to lose, not short enough to be forced in. They carry the highest risk premium and should play the tightest. The big stack carries the lowest risk premium, which is the whole engine behind ICM pressure.
Bubble Factor & Risk Premium: How ICM Changes Your Shoves and Calls
"Bubble factor" measures how much more losing your chips costs you than winning the same chips helps — and it spikes right before every pay jump. A bubble factor of 1.0 means chips and money move together (early game). A bubble factor of 1.5 means a lost pot hurts 1.5× as much as an identical won pot helps — so you need a much bigger edge to get involved.
Two practical rules fall out of it:
- •Big stack: attack. Your low risk premium lets you open and 3-bet relentlessly against players who can't call without risking their tournament life. This is "applying ICM pressure," and it's the most reliable way to win chips at a final table.
- •Medium and short stacks: tighten your calling range, but keep shoving first. Being the one who moves all-in (with fold equity) is far better than being the one who has to call off. Under pressure, your calling range should shrink hard while your open-shoving range stays aggressive.
ICM Deal vs Chip Chop: How to Split a Final-Table Prize Pool
A chip chop splits the remaining money by raw chip percentage; an ICM deal splits it by each player's ICM dollar value. The chip chop favors big stacks, the ICM deal is fairer to short stacks. When players agree to end a tournament early and divide the prizes, these are the two methods on the table — and knowing the difference is worth real money.
Say three players with 50% / 30% / 20% of the chips are dividing a $1,500 remaining pool (paying $900 / $400 / $200):
| Player | Chip chop | ICM deal | Difference |
|---|---|---|---|
| Leader (50%) | $750 | $618 | −$132 |
| Middle (30%) | $450 | $485 | +$35 |
| Short (20%) | $300 | $397 | +$97 |
The short stack gets $97 more from an ICM deal than a chip chop, because ICM credits the pay jumps they've already earned. So the rule is easy: if you're short, ask for an ICM deal; if you're the chip leader, propose a chip chop. In practice the chip leader often negotiates a bit above their ICM number (and short stacks accept a bit below) in exchange for the certainty of locking up money — that's fine, as long as you know your ICM figure first. Run your own stacks and payouts through the ICM deal calculator before you agree to anything.
When Does ICM Matter Most — and When Should You Ignore It?
ICM matters most near pay jumps and least when they're far away. Lean on it in these spots:
- •The
money bubble — the biggest jump of all is from $0 to a cash, so risk premiums peak. - •The final-table bubble and every pay jump on the final table — each ladder step is real money.
- •Satellites — the extreme case: every qualifying seat is worth the same, so once you have enough chips to win a seat, extra chips are worth almost nothing and you fold nearly everything.
- •Early and middle stages, where the next pay jump is a distant abstraction and accumulating chips is what wins tournaments.
- •Deep-stacked play with tiny blinds, where you have room to outplay opponents rather than get it in.
- •Heads-up for the title, where only two prizes remain and ICM stops changing your strategy — it's effectively chip EV again.
How Accurate Is ICM? Its Limitations
ICM is the best simple model we have, but it's an approximation — it assumes every player is equally skilled and ignores almost everything except stack sizes. Be honest about what it leaves out:
- •Skill. ICM treats a world champion and a first-timer with equal stacks as equal. A better player's chips are worth more than the model says.
- •Position. A 3-big-blind stack on the button (about to see cheap flops) is worth more than the same stack in the big blind (about to be forced in). ICM can't see the seats.
- •Blinds and future play. ICM freezes the tournament at this instant; it ignores rising blinds, antes, and how the next few orbits will actually play out.
FAQ
The 3 Things to Remember
1. Chips aren't money. You win only one first prize, so the chip leader is worth less than their chip share and the short stack is worth more. That single gap is all of ICM. 2. Late game, switch from chip EV to $EV. Near pay jumps, a call needs extra equity (a risk premium) to be profitable. The medium stack folds hands a cash game would snap-call. 3. Know your number before you deal. Short stacks want an ICM deal, big stacks want a chip chop — run the calculator first.
From here, see how ICM pressure fits the wider tournament strategy, or go back to the foundation with poker equity and pot odds.