ICM Explained: How It Changes Tournament Decisions

I spent two years in underground poker rooms in Queens observing how players made decisions near the end of tournaments. The decision-making changed dramatically when the game shifted from play-to-win to play-to-be-alive. That shift is what ICM attempts to model.

ICM stands for Independent Chip Model. It calculates what your expected payout should be based on your current chip stack relative to all other stacks, assuming the blinds get larger and eventually someone wins. The calculation produces an "ICM value," which is your expected value in the tournament.

Here is how it works in a simple three-player example:

Three players remain. The prize structure is $100 for first, $50 for second, $20 for third. Player A has 5 million chips, Player B has 3 million, Player C has 2 million. The total is 10 million chips. Using ICM:

• Player A's ICM value is 50 + (30-50) * (3/10) * (2/10) = roughly $58
• Player B's ICM value is roughly $42
• Player C's ICM value is roughly $20

The rough numbers are less important than the concept: your expected payout based on chip equity is different from your chip percentage. You have 50% of the chips but an expected payout higher than 50% of the money. This is the chip leader advantage.

Where ICM Changes Decision-Making

ICM becomes relevant when players discuss chopping a tournament or when a player considers folding to make a deal. Instead of playing until one person wins, players might agree to split the remaining prize pool based on ICM calculations.

A player with an ICM advantage (more chips than their fair share of expected value) wants to negotiate based on ICM. A player with an ICM disadvantage wants to keep playing. The negotiation revolves around ICM as the reference point.

ICM also affects risk tolerance. If you are deep in the money and considering a large bet, ICM tells you whether the bet is justified by your chip equity. A bet that costs you 30% of your stack needs a better success rate if your ICM value is high than if it is low.

"ICM is a way of saying, 'Given what we know right now, this is what you should expect to win if you play it out.' It does not predict perfectly, but it is mathematically sound."

The limitation of ICM is that it assumes all players are equal in skill. It does not account for one player being significantly better or worse than the others. In real tournaments, skill variation is substantial. A skilled player should be willing to take more risk than ICM suggests. A weak player should be willing to take less risk.

The other limitation is that ICM assumes stacks will not change dramatically in relative terms due to skill gaps. If a player is much better, their chip percentage will increase. ICM does not model this.

The value of ICM is not in its precise calculation but in its framework. It gives players a language to discuss what their reasonable expectation is. Without ICM, negotiations are just arguments. With ICM, negotiations have a mathematical anchor.

In the underground rooms I observed, the older players knew ICM intuitively without knowing the term. They understood chip value and negotiating position. The younger players were reading it off calculators on their phones. Both groups understood that chips are not equally valuable once you near the end of a tournament.