ICM in poker: finally understand why your chips aren't worth their face value

You're on the bubble of a tournament, medium stack, and the chip leader shoves on you. You look down at A♠Q♠ — a hugely winning hand in this spot under normal circumstances — and yet, the right decision is to fold. Welcome to the world of ICM, the concept that separates profitable tournament players from those who "don't understand what just happened".

Chips are not money

In cash games, a chip is worth exactly its face value: doubling your stack means doubling your money. In tournaments, that equivalence breaks. Winnings are distributed in tiers, and part of the prize pool is reserved for the places you reach by surviving — not by accumulating.

The fundamental consequence: doubling your stack does not double your expected winnings. The chips you win are worth less than the chips you risk. That's the whole point of ICM.

What ICM actually calculates

ICM (Independent Chip Model) converts stacks into monetary value. The model rests on a simple assumption: each player's probability of finishing first is proportional to their stack; then, with that player removed, the probability of finishing second is proportional to the remaining stacks, and so on.

Take a 3-player Sit & Go, €100 prize pool, paid 50/30/20. The stacks: A holds 5,000 chips, B 3,000, C 2,000.

• A (50% of the chips): ICM equity ≈ €38.40
• B (30% of the chips): ICM equity ≈ €32.75
• C (20% of the chips): ICM equity ≈ €28.86

Look at those numbers closely: A holds half the chips but only 38% of the money. C holds just a fifth of the chips but almost 29% of the money. The first chips of a stack are worth far more than the last ones — it's mathematical, and it changes everything.

Risk premium: why we fold hands that are "too strong"

Since chips won are worth less than chips risked, every all-in demands an extra safety margin: that's the risk premium. Where a call needs 45% equity in pure chip-EV, ICM may demand 55%, 60%, sometimes more.

This premium isn't uniform — it depends on who covers you:

• Against a stack that covers you, your tournament is on the line: the premium explodes, you fold very strong hands.
• Against a shorter stack, you can't be eliminated: the premium is small, you can call wider.
• Big stacks attack almost risk-free: nobody can eliminate them, and everyone else faces the premium.

It's tournament poker's most counter-intuitive inversion: the same hand, at the same stack size, plays differently depending on who's across from you. On the bubble, these effects reach their peak.

The extreme case: the satellite

In a satellite where every paid place receives the same ticket, ICM produces a spectacular result: near the bubble with a comfortable stack, folding aces preflop becomes correct. Winning the hand barely increases your value (the ticket is already nearly locked up), losing it destroys it. No possible gain, real risk: the fold is mandatory, even with the best hand in the game.

If that logic shocks you, it's because you're still thinking in chips. ICM thinks in euros.

The model's limits

ICM is a model, not a law of physics. It ignores:

• Position: the same stack on the BTN and UTG doesn't have the same future — ICM values them identically.
• Skill: the model assumes everyone plays the same. A very good deep-stacked player can accept a bit more variance than ICM suggests.
• The rest of the tournament: rising blinds, table changes — the model photographs a single instant.

These limits are known and documented, but they don't overturn the conclusion: in high-stakes spots (bubble, final table, satellites), ICM is a far better compass than chip-EV.

What about Spin & Gos?

An interesting special case: in a classic winner-take-all Spin & Go, there's only one prize. No tiers, no survival to value: maximizing your chips and maximizing your money become the same thing again. Pure chip-EV is the right reference there — which is exactly why Nash push/fold charts apply directly, with no ICM adjustment.

How to integrate ICM into your game

1. Identify the hot zones: bubble approach, pay jumps at the final table, satellites. That's where ICM weighs.
2. Spot who covers whom before every preflop decision — it's the number one piece of information.
3. Tighten your calls far more than your shoves: ICM punishes the caller most, because the caller has no fold equity.
4. Train your preflop ranges away from the tables: in short-stacked spots, decisions come too fast to improvise. That's exactly what spaced repetition automates.

ICM can't be guessed at the table: it's understood beforehand, and it translates into ranges you know by heart. The players who "feel" the bubble are actually the ones who did the work in advance.