ICM

ICM stands for independent chip model. It is a method of translating a stack of tournament chips into an expected cash value, and it is the reason correct tournament play diverges so sharply from correct cash-game play.

The problem it solves

In a cash game a chip is worth its face value — a player can stand up and bank it. In a tournament, chips cannot be cashed out. They only matter in so far as they lead to finishing positions, and the prize structure pays those positions unevenly.

That produces a fundamental asymmetry. Doubling a stack does not double its monetary value, because the prize for finishing first is not double the prize for second. Meanwhile losing the last chip ends the tournament entirely. Chips you win are therefore worth less per chip than the chips you risk to win them.

What the model does

ICM takes the current stack sizes and the payout structure and estimates each player's probability of finishing in each paid position, from which it computes an expected prize. Its central assumption is that a player's chance of finishing first is simply their share of the chips in play, with the remaining places worked out recursively from there.

The practical output is a dollar figure per stack. Comparing those figures before and after a hypothetical all-in is what makes ICM usable: a call can win chips and lose money.

How it changes play

The consequences are large and counterintuitive.

Close to the money, and especially on the bubble, calling ranges tighten dramatically. A coin-flip for a player's tournament life is usually a losing proposition in money terms even though it is break-even in chips, because busting forfeits a payout that was nearly secured.

At the same time, pressure becomes extremely profitable. A big stack can open and shove far wider than usual, because the short and medium stacks cannot call without risking their own equity. The medium stack, with a comfortable position it does not want to surrender, is often the most constrained player at the table — more so than the short stack, which has less to lose.

Satellites are the extreme case. Where every qualifying place pays the same seat, chips above the amount needed to qualify are worth almost nothing, and play becomes almost entirely about survival.

The model's limits

ICM is a simplification and it is worth being clear about what it ignores.

It takes no account of position, blind levels or the speed of the structure, so it does not know that a short stack about to be swallowed by rising blinds is in worse shape than its chip count implies. It assumes no skill difference between players, so it undervalues a strong player's stack and overvalues a weak one's. And it assumes play ends immediately rather than continuing, which is why more elaborate models exist that try to account for future hands.

Used as a precise instruction, ICM will sometimes be wrong. Used as a correction to cash-game instincts — risk is more expensive than it looks, and applying pressure is cheaper — it captures something true about how tournaments actually pay out.