Tournament chips are not money. That single fact separates tournament strategy from cash-game strategy, and the Independent Chip Model (ICM) is the math that measures it. Players who ignore ICM make calls that look fine in chips and bleed real dollars. This guide explains what ICM is, walks through an actual calculation with three players and a $100 prize pool, and shows exactly how ICM pressure reshapes your preflop ranges.
This is the ground-floor article. Once the model clicks, our final table ICM strategy guide is the deep dive: ladder math, stack-by-stack adjustments, and the spots where ICM gets truly violent.
What Is ICM in Poker?
ICM, the Independent Chip Model, is a formula that converts tournament chip stacks into real-money prize equity. It takes two inputs, the remaining stacks and the remaining payouts, and outputs what each stack is worth in dollars right now. It ignores skill, position, and blinds entirely; it is a pure snapshot of stacks versus payouts.
Why does anyone need a model for this? Because tournament payouts are not proportional to chips. Bust 14th and you collect 14th-place money whether you went out with one big blind or a mountain. Chips only convert to cash through the payout ladder, and that conversion is lumpy.
Three consequences fall out of the model immediately:
- Chips lose marginal value as you stack them. Your first 5,000 chips are worth more dollars than your next 5,000. Doubling your stack never doubles your equity.
- Survival has cash value. Folding into a pay jump earns real money without playing a hand.
- Every all-in is a dollar decision, not a chip decision. A call that wins chips on average can still lose money on average.
Every serious ICM calculator, deal-making tool, and tournament solver runs some version of this model under the hood, usually the Malmuth-Harville method we will walk through below.
Chip EV vs ICM: What Is the Difference?
Chip EV (cEV) values every chip equally, so doubling your stack doubles your tournament value. ICM ($EV) weights chips against the payout structure, so doubling your stack always less than doubles your real-money equity. In cash games the two are identical. In tournaments they diverge, and the gap peaks near pay jumps.
The practical headline is an asymmetry: a chip you lose is worth more than a chip you win. Lose half your stack and you lose more dollar equity than you gain by winning the same amount. That asymmetry is why tournament players need extra equity to get chips in, and why "I was flipping, it's fine" is often wrong in a way it never is at a cash table.
Early in a tournament with flat stacks and distant payouts, standard GTO opening ranges apply almost unchanged. The divergence grows as the bubble approaches and explodes at final tables and in satellites.
A Worked ICM Example: 3 Players, $100 Prize Pool
In a three-handed tournament paying $50, $30, and $20, a player holding 50% of the chips holds only about 38% of the prize money under ICM. The example below shows why, using the Malmuth-Harville method that nearly every ICM tool runs internally.
Setup: 10,000 chips in play, three players left, everyone is already in the money.
| Player | Stack | Chip share | ICM equity |
|---|---|---|---|
| A | 5,000 | 50% | $38.39 |
| B | 3,000 | 30% | $32.75 |
| C | 2,000 | 20% | $28.86 |
The intuition behind the calculation is simple, even though the arithmetic gets tedious fast:
- Your chance of finishing 1st equals your share of the chips. Player A wins 50% of the time, B 30%, C 20%.
- For 2nd place, imagine each opponent winning, remove them, and re-run the logic on what remains. If B wins (30% of the time), A finishes 2nd with probability 5,000 / 7,000, A's share of the non-B chips. If C wins (20%), A is 2nd with probability 5,000 / 8,000. Add it up and A finishes 2nd about 34% of the time, and 3rd the remaining 16%.
- Multiply each finish probability by its payout. For A: 50% of $50, plus 34% of $30, plus 16% of $20, roughly $38.39.
Now look at what the table is telling you. Player A holds half the chips but only $38.39 of a $100 pool. Player C holds a fifth of the chips but $28.86, because even the shortest stack still cashes for at least $20 and wins outright 20% of the time. If A won every chip on the table, the prize is capped at $50: the second 5,000 chips would add only $11.61 of equity, while the first 5,000 carried $38.39.
Here is the same idea as a decision. Suppose A and B get all-in on a pure coin flip for B's 3,000-chip stack:
- A wins: A has 8,000 chips, B is out. A's new equity: $46.00.
- A loses: A drops to 2,000 chips. A's new equity: $29.50.
- The flip is 50/50, so A's average: $37.75, which is less than the $38.39 A had before the hand.
A chip-neutral gamble costs the chip leader $0.64 in real money. Run the numbers and A needs about 53.9% equity just to break even, because A risks $8.89 of equity to win $7.61. That extra 4% is ICM pressure in its mildest form: three players, modest pay jumps, no bubble. It only gets heavier from here.
What Is ICM Pressure? The Bubble Factor
ICM pressure is the gap between chip odds and dollar odds in a confrontation. It is measured by the bubble factor: the dollars you lose by losing an all-in divided by the dollars you gain by winning it. A bubble factor of 1.5 means losses hurt 1.5 times more than wins help, so you need 60% equity in a spot that chip math prices at 50%.
In the three-handed example above, Player A's bubble factor against B was 8.89 / 7.61, about 1.17. In a cash game the bubble factor is always exactly 1. In tournaments it ranges from barely above 1 in the early levels to numbers so large that no starting hand can call.
| Bubble factor | Equity needed (even chip odds) | Typical situation |
|---|---|---|
| 1.0 | 50% | Cash games, winner-take-all |
| 1.2 | ~55% | Early and mid MTT stages |
| 1.5 | 60% | Approaching the money, medium stack |
| 2.0 | ~67% | Money bubble vs a stack that covers you |
| 3.0+ | 75%+ | Satellite bubbles, brutal final-table ladders |
The conversion is mechanical: at even chip odds, required equity = bubble factor / (bubble factor + 1). What makes bubble factors interesting is that they are pairwise. You do not have one bubble factor; you have one against each opponent, and they are wildly unequal.
The worst seat in tournament poker is the medium stack facing a stack that covers you: your bubble factor against the chip leader might be 2.0 while it is only 1.2 against a short stack. The chip leader enjoys the mirror image, risking chips priced near 1.1 while charging you 2.0. That one-way leverage is why competent chip leaders raise relentlessly on bubbles.
The Classic Satellite Hand: Folding Pocket Aces Preflop
In satellites, where every seat pays exactly the same, ICM becomes so extreme that folding pocket aces preflop is often the standard play. When a fold locks in roughly 95% of a seat and calling an all-in offers 85% equity for that same seat and nothing more, calling burns money with the best hand in poker.
Concrete version. Six players remain in a satellite that awards five identical $1,000 seats. You have 60,000 chips, second in chips. One player has 8,000 and pays the big blind in two hands. The chip leader, with 100,000, shoves into you. You look down at A♠A♥, roughly 85% against whatever they are doing this with.
- Fold: you keep 60,000 chips with a micro stack about to blind out. Your seat equity is around 95%, call it $950.
- Call and win: you get a seat worth $1,000. Not $2,000, not a bonus, just the same seat with surplus chips that buy nothing.
- Call and lose: $0.
Calling is worth 0.85 × $1,000 = $850. Folding is worth about $950. Snapping off the shove with aces costs you roughly $100, a tenth of a seat. The upside is capped and the downside is total, which is the satellite condition in one sentence. Flat payouts make chips above "enough to survive" nearly worthless, and our satellite strategy guide covers how each stack size should play these endgames.
Most tournaments are not satellites, and aces are almost never a fold outside them. But the example proves the principle at its limit: hand strength is half the equation, and the payout structure is the other half.
How Does ICM Change Preflop Ranges?
ICM tightens every range that puts your own stack at risk and widens every range that threatens an opponent's stack. Calling ranges shrink the most, shoving ranges shrink somewhat, and covering stacks attack wider. Medium stacks fold hands they would snap with in a cash game while chip leaders open junk profitably.
The concrete shifts, roughly in order of importance:
- Calling all-ins tightens hardest. Calling has no fold equity, so it eats the full ICM tax. In the big blind facing a 10bb button jam, a chip EV strategy calls somewhere near 30% of hands. On a money bubble where the button covers you, the range collapses toward strong pairs and big broadways, often a third of its former size. AJo and 66 go from snap calls to easy folds.
- Opens tighten when you are covered. A middle stack on the bubble folds hands like ATo, KQo, and 55 from early position, hands that are standard opens in chip EV, because getting jammed on is a disaster at a bubble factor near 2.
- Big-stack aggression widens. The covering stack jams and raises hands like K9s, Q9s, and weak offsuit aces that would be chip EV neutral at best, because every opponent's defending range is crushed by their own bubble factor.
- Flatting nearly disappears at short stacks. ICM rewards taking the betting lead and punishes passive chip-risking, so play converges on jam-or-fold trees. Our short stack strategy guide covers those ranges in detail.
- Postflop follows the same logic. Fewer thin bluff-catches, more pot control, smaller sizings in stack-threatening spots. The final table ICM article works through these postflop adjustments.
One warning: ICM does not simply mean "play tighter." It means play tighter when your stack is at risk and looser when only your opponent's is. Players who absorb half the lesson turn into bubble nits that good chip leaders farm for blinds.
Common ICM Mistakes
Most ICM errors come from running cash-game math in a tournament. These four cover the bulk of the lost money:
- Calling because of pot odds alone. "I was getting 2-to-1" is chip math. Your pot odds need an ICM tax added before they mean anything near a pay jump.
- Ignoring the stack distribution. A 20bb stack plays completely differently when a 2bb stack is about to blind out. Always know who covers whom and where the desperate stacks sit.
- Missing the aggressive half. If you cover the table on the bubble and your open frequency looks the same as level two, you are leaving the most profitable spots in tournament poker untouched.
- Reverting after a pay jump. One ladder secured does not reset the model. The next jump starts mattering immediately, and stacks rarely rebalance right away.
How to Study ICM Spots
The fastest way to learn ICM is to compare chip EV and ICM solutions for the same spot side by side, because the hands that change action are exactly the lesson. Reading theory tells you ranges tighten; seeing AJo flip from call to fold in a specific stack setup makes it stick.
This is what GTO Gecko's ICM tooling is built around. The MTT library includes ICM-aware preflop ranges as part of the Elite plans, plus an ICM compare view that puts the chip EV range and the ICM range for the same scenario next to each other, so the risk premium becomes a visible band of hands switching color. The trainers then drill those spots with an ELO rating and adaptive repetition that re-serves whatever you misplay.
You can start without paying anything: browse the free preflop range library on GTO Gecko and use the daily free trainer hands to test how your bubble instincts hold up. No credit card involved. If you want the broader context on reading solver output, our guide on how to use a poker solver is the right companion piece.
Two study habits round it out. First, ICM magnifies variance, so size your buy-ins with the free MTT bankroll calculator and the principles from bankroll management for pros; a thin roll turns correct ICM folds into scared-money over-folds. Second, after every deep run, pull the hands where you felt squeezed and check what an ICM-aware solution actually does there.
ICM Poker FAQ
What does ICM stand for in poker?
ICM stands for Independent Chip Model. It is a mathematical model, usually computed with the Malmuth-Harville method, that converts tournament chip stacks into shares of the remaining prize pool. Solvers, deal-making calculators, and staking arrangements all use it as the standard way to price a tournament stack in dollars.
Is ICM only relevant at the final table?
No. ICM matters anywhere a pay jump is close relative to stack sizes: the money bubble, final table bubbles, and satellites are the heaviest spots. Early in a tournament it is nearly invisible and chip EV play is fine. Winner-take-all formats stay chip EV throughout; standard Spin & Go strategy, for example, runs on pure chip EV until a multi-prize structure appears.
Should you ever fold pocket aces preflop in a tournament?
Yes, but almost exclusively in satellites and extreme final-table ladder spots. When every payout is identical and folding preserves nearly locked-up equity, even an 85% favorite can be a clear fold, as the worked satellite example above shows. In a regular tournament with normal pay jumps, folding aces preflop is virtually never right.
What is the difference between chip EV and ICM?
Chip EV treats every chip as equally valuable, so a decision is good if it wins chips on average. ICM converts chips into prize money first, so a decision is good only if it wins dollars on average. The two agree in cash games and winner-take-all formats and diverge in any tournament with a payout ladder.
Does ICM apply to cash games?
No. In a cash game your chips are literally money, so chip EV and dollar EV are the same number and the bubble factor is always 1. That is exactly why cash-game instincts misfire in tournaments: the call that is automatic at a cash table can be a serious error on a bubble.
How accurate is the ICM model?
ICM is deliberately simplified: it ignores blinds, position, skill edges, and who is about to be forced all-in. Extensions like future game simulation patch some of that. It remains the industry standard because it is close enough in the spots that matter most, and far closer than chip EV.
Where to Go From Here
You now have the intro toolkit: what ICM is, why a chip won is worth less than a chip lost, how to read a bubble factor, and the concrete ways ranges shift under pressure. The next layer is the final table ICM strategy guide: ladder-by-ladder adjustments, playing each stack size, and deal-making math.
Until then, the cheapest improvement available is simple awareness. Before your next all-in decision near a pay jump, ask one question: am I pricing this in chips or in dollars? The players who ask it consistently earn more from the same cards.