Cash-game intuition breaks in tournaments
In a cash game, a $100 chip is worth $100. Double your stack and you double its cash value. Tournament chips do not work that way because prizes are awarded by finishing position, not by cashing out each chip.
Once payouts matter, losing chips usually hurts more than winning the same number helps. Busting can cost your entire remaining prize equity, while doubling up never guarantees that you will receive twice as much money. The Independent Chip Model—ICM—is a way to estimate that changing relationship between stacks and the remaining prize pool.
ICM does not tell you how many chips a play wins. It estimates how a chip decision changes your share of the remaining prize money.
What ICM does in plain English
ICM starts with every remaining stack and the payout schedule. It assumes each player's chance of finishing first is proportional to their share of the chips, then works through the possible finishing orders to estimate each player's prize equity.
The result is not a prediction of the exact finish. It is a dollar-value estimate for the current stack. That estimate lets you compare actions by money expected value—often written as $EV—instead of looking only at tournament-chip expected value, or chip EV.
You do not need to calculate every finishing order at the table. You need to recognize when the payout structure creates enough pressure that a chip-profitable gamble can still lose prize equity.
A three-player example
Three players remain. First place pays $500, second pays $300, and third pays $200. Their stacks are 50%, 30%, and 20% of the chips in play.
The chip leader owns half the chips, but ICM values that stack at about $384—not $500—because first place is not guaranteed. The middle stack's estimate is about $328, and the short stack's is about $289. The shortest stack owns only 20% of the chips but nearly 29% of the prize equity because third place already pays $200.
Now the key idea is visible: gaining 10% of the chips does not add a fixed amount of money, and losing 10% does not remove that same amount. Stack value curves because everyone is competing for several different payouts.
Risk premium changes your calling decisions
In a chip-EV calculation, a call is profitable when your equity clears the price offered by the pot. Under ICM pressure, a call may need additional equity because losing the all-in damages your prize equity more than winning it improves your prize equity. That extra requirement is commonly called risk premium.
This is why late-tournament calling ranges can become much tighter than jamming ranges. The player who moves all-in can win immediately when everyone folds. The caller has no fold equity and must accept the full downside of losing.
Do not turn this into “always fold near the money.” The correct adjustment depends on the payout jump, every relevant stack, who covers whom, and the ranges involved. ICM rewards selective aggression as often as it rewards disciplined folds.
The medium stack often feels the most pressure
A very short stack may need to gamble before blinds destroy its fold equity. A large stack can pressure players it covers because their tournament life is at risk. The medium stack can be trapped between them: it has enough chips to lose meaningful prize equity but not enough to threaten the leader comfortably.
Before entering a large pot, look beyond your own big-blind count. Identify the shortest stacks, the players who cover you, and the next payout jump. A hand that is routine against the short stack may become a fold against the one player who can eliminate you.
- Big stack: pressure stacks that cannot call without risking a valuable finish.
- Medium stack: avoid unnecessary collisions with covering stacks while attacking carefully selected targets.
- Short stack: preserve fold equity and do not wait so long that opponents can call without meaningful risk.
Where ICM pressure is strongest
ICM still exists before these moments, but its practical effect is often smaller when the field is large and the next payout change is distant. MIT's tournament-poker material emphasizes that the divergence between chip value and dollar value becomes especially important around the money bubble and final table.
- The money bubble, when one elimination changes every survivor from unpaid to paid.
- Large payout jumps, especially at final tables.
- Satellites, where several finishing positions award the same seat and extra chips may add almost no value.
- All-ins involving uneven stacks, particularly when one player can eliminate another.
Pressure is a weapon only when someone can fold
Spotting ICM pressure is not enough. You must identify who actually feels it. A player with one blind left may be forced to call. A huge stack may welcome the confrontation. A medium stack with several shorter players remaining is often the player with the most to protect.
Before jamming as a bluff, ask whether the target understands the payout pressure and has enough stack to fold. Before calling, ask whether the aggressor is capable of exploiting the pressure with a wider range. Good tournament poker is not simply tight; it is aware of how the same payout structure affects each player differently.
ICM is a model, not a commandment
Traditional ICM uses stack sizes and payouts. It does not directly know who is more skilled, who posts the next big blind, who has position, how the table is playing, or whether a particular stack can exploit future opponents better than the others.
A 2025 empirical study covering more than 10,000 tournaments found that ICM performed better than its comparison baseline, while also finding a tendency to underestimate larger-stack performance and overestimate shorter-stack performance. That is useful perspective: ICM is a strong reference model, not perfect reality.
Bounties also change the calculation because eliminating a player can create immediate cash value. Progressive knockout events require both prize-pool equity and bounty value; applying ordinary ICM without the bounty component can produce the wrong answer.
A ten-second tournament scan
Run the scan before major late-stage decisions, not after seeing the result. Away from the table, use an ICM calculator or tournament solver to review the full stack distribution and payout schedule. Studying only your two cards misses the reason the spot was difficult.
- Payout: What is the next jump, and how close is it?
- Stacks: Who is shortest, who covers me, and whom do I cover?
- Consequence: What happens to my prize equity if I fold, lose, or win?
- Leverage: Can my opponent realistically fold, or am I only creating risk?
- Format: Are bounties, satellite seats, or unusual payouts changing the incentive?
Common ICM mistakes
- Treating ICM as a rule to fold everything near the bubble.
- Looking only at your stack instead of every remaining stack.
- Calling with a chip-profitable range without accounting for risk premium.
- Attacking a short stack that is too committed to fold.
- Ignoring who covers whom when stacks are close.
- Using standard-tournament ICM in a bounty or satellite format without adjusting the value model.
Actionable takeaways
- Tournament chips do not convert to dollars at a fixed rate.
- Losing an all-in can cost more prize equity than winning it gains, creating a risk premium.
- Calling ranges often tighten more than shoving ranges because only the aggressor has fold equity.
- Check every stack, the next payout jump, and who covers whom before a major decision.
- Use big-stack pressure against players who can actually fold; do not confuse pressure with automatic aggression.
- Treat ICM as a valuable estimate whose assumptions become less complete when skill, future position, or bounties matter.
- Review bubble and final-table hands with the full payout and stack information—not just the cards.