ICM — The Mathematics of Tournament Chip Value

🏆 What is ICM?

In cash games, the relationship between chips and money is straightforward: chips equal money. Every chip you win adds exactly that amount to your profit, and every chip you lose subtracts the same.

In tournaments, this relationship fundamentally breaks. Doubling your stack does NOT double your expected prize money. This asymmetry exists because:

• Ceiling effect: You can never win more than the 1st place prize, no matter how many chips you accumulate.
• Floor effect: Losing all your chips means elimination and zero additional prize money.
• Concavity: The relationship between chips and prize equity follows a concave curve — each additional chip is worth less than the previous one (diminishing returns).

ICM (Independent Chip Model) is a mathematical model that converts tournament chip stacks into estimated prize money equity. It answers the question: "Given the current chip distribution and prize structure, how much money is each player's stack worth on average?"

📐 The Malmuth-Harville Model

The most widely used ICM model is the Malmuth-Harville model. It calculates each player's probability of finishing in each position based on their chip stacks.

Probability of Finishing 1st

The simplest assumption: each player's probability of winning is proportional to their chip stack.

$$P_i(1\text{st}) = \frac{s_i}{S}$$

where \(s_i\) is player \(i\)'s stack and \(S = \sum_{j=1}^{N} s_j\) is the total chips in play.

Probability of Finishing 2nd

Player \(i\) finishes 2nd when some other player \(j\) finishes 1st, and then player \(i\) wins among the remaining players:

$$P_i(2\text{nd}) = \sum_{j \neq i} P_j(1\text{st}) \times \frac{s_i}{S - s_j}$$

This is a conditional probability: for each possible 1st-place finisher \(j\), we remove \(j\)'s stack and calculate \(i\)'s chance of winning the remaining field.

General Formula

For finishing in position \(k\), we extend this recursion by summing over all possible orderings of the players who finish before position \(k\):

$$P_i(k\text{th}) = \sum_{\text{all orderings}} \prod \frac{s_{\sigma(m)}}{S - \sum_{n < m} s_{\sigma(n)}}$$

This summation grows factorially with the number of players, which is why ICM calculators use approximations for large fields.

ICM Equity

Once we have all finishing probabilities, the ICM equity for player \(i\) is the probability-weighted sum of all prizes:

$$\text{ICM}_i = \sum_{k=1}^{N} P_i(k\text{th}) \times \text{Prize}_k$$

This gives us the expected dollar value of each player's chip stack.

🔢 Worked Example

3 players remain in a tournament. The prize structure and chip stacks are:

Total prize pool = $100. Stacks: A = 5,000, B = 3,000, C = 2,000. Total chips = 10,000.

Step 1: Probabilities of Finishing 1st

$$P_A(1\text{st}) = \frac{5000}{10000} = 0.50$$ $$P_B(1\text{st}) = \frac{3000}{10000} = 0.30$$ $$P_C(1\text{st}) = \frac{2000}{10000} = 0.20$$

Step 2: Probabilities of Finishing 2nd

Player A:

$$P_A(2\text{nd}) = P_B(1\text{st}) \times \frac{5000}{10000 - 3000} + P_C(1\text{st}) \times \frac{5000}{10000 - 2000}$$ $$= 0.30 \times \frac{5000}{7000} + 0.20 \times \frac{5000}{8000}$$ $$= 0.30 \times 0.7143 + 0.20 \times 0.6250 = 0.2143 + 0.1250 = 0.3393$$

Player B:

$$P_B(2\text{nd}) = P_A(1\text{st}) \times \frac{3000}{10000 - 5000} + P_C(1\text{st}) \times \frac{3000}{10000 - 2000}$$ $$= 0.50 \times \frac{3000}{5000} + 0.20 \times \frac{3000}{8000}$$ $$= 0.50 \times 0.60 + 0.20 \times 0.375 = 0.30 + 0.075 = 0.3750$$

Player C:

$$P_C(2\text{nd}) = P_A(1\text{st}) \times \frac{2000}{10000 - 5000} + P_B(1\text{st}) \times \frac{2000}{10000 - 3000}$$ $$= 0.50 \times \frac{2000}{5000} + 0.30 \times \frac{2000}{7000}$$ $$= 0.50 \times 0.40 + 0.30 \times 0.2857 = 0.20 + 0.0857 = 0.2857$$

Step 3: Probabilities of Finishing 3rd

Since there are only 3 players, the probability of finishing 3rd is simply:

$$P_i(3\text{rd}) = 1 - P_i(1\text{st}) - P_i(2\text{nd})$$ $$P_A(3\text{rd}) = 1 - 0.50 - 0.3393 = 0.1607$$ $$P_B(3\text{rd}) = 1 - 0.30 - 0.3750 = 0.3250$$ $$P_C(3\text{rd}) = 1 - 0.20 - 0.2857 = 0.5143$$

Step 4: ICM Equity

$$\text{ICM}_A = 0.50 \times 50 + 0.3393 \times 30 + 0.1607 \times 20 = 25.00 + 10.18 + 3.21 = \$38.39$$ $$\text{ICM}_B = 0.30 \times 50 + 0.3750 \times 30 + 0.3250 \times 20 = 15.00 + 11.25 + 6.50 = \$32.75$$ $$\text{ICM}_C = 0.20 \times 50 + 0.2857 \times 30 + 0.5143 \times 20 = 10.00 + 8.57 + 10.29 = \$28.86$$

Verification

$$\$38.39 + \$32.75 + \$28.86 = \$100.00 \checkmark$$

🫧 Bubble Factor

Bubble factor (BF) quantifies the asymmetry between winning and losing chips in ICM terms. It measures how much more losing hurts compared to how much winning helps.

$$BF = \frac{|\Delta\text{ICM}_{\text{lose}}|}{\Delta\text{ICM}_{\text{win}}}$$

Example

Suppose you are considering calling an all-in:

• If you win: your ICM equity increases by $2.50
• If you lose: your ICM equity decreases by $5.00

$$BF = \frac{5.00}{2.50} = 2.0$$

This means you need TWICE the chip-EV edge to justify the risk in ICM terms.

Bubble Factor and Required Equity

Bubble factor directly determines the minimum equity you need to call profitably:

$$\text{Required Equity} = \frac{BF}{1 + BF}$$

With \(BF = 2.0\):

$$\text{Required Equity} = \frac{2.0}{1 + 2.0} = \frac{2}{3} = 66.7\%$$

Instead of the normal 50% equity needed in a chip-EV pot, you now need 66.7% — a massive difference.

Typical Bubble Factors

🃏 When ICM Changes Your Decision

Here is a concrete hand example where chip-EV says call but ICM says fold.

Scenario

Bubble of a 100-player tournament. 15 players remain, 14 get paid. Hero has 25BB, villain shoves 20BB from the cutoff. Hero is in the big blind with A♠Q♥ — approximately 57% equity vs villain's shoving range.

Chip-EV Analysis

In a cash game or chip-EV context, this is an easy call. 57% equity is well above the ~43% breakeven point, making it a clearly profitable call:

$$EV_{\text{chip}} = 0.57 \times 20\text{BB} - 0.43 \times 20\text{BB} = +2.8\text{BB}$$

ICM Analysis

However, the bubble factor in this situation is 2.3. The required equity becomes:

$$\text{Required Equity} = \frac{2.3}{1 + 2.3} = \frac{2.3}{3.3} = 69.7\%$$

Our 57% equity is far below the 69.7% threshold:

$$57\% < 69.7\% \Rightarrow \textbf{FOLD}$$

Despite having a +2.8BB chip-EV edge, this call is significantly -EV in ICM terms. The cost of busting on the bubble (losing your equity in the remaining prize pool) far outweighs the benefit of doubling up.

📋 Practical Guidelines

Understanding ICM is one thing — applying it at the table is another. Here are the key principles for each stack size:

Big Stack

ICM pressure is on others, not on you. You have the least to lose relative to your stack. Use this leverage to accumulate chips aggressively — put medium stacks in tough spots where they cannot call without premium hands. The big stack is the "bully" of ICM situations.

Medium Stack

You are the most affected by ICM. You have enough chips that busting would cost you significant equity, but not so many that you can afford to gamble. Tighten significantly near bubbles and pay raises — avoid marginal spots against other medium-to-big stacks.

Short Stack

The ICM paradox: because you are already in danger of elimination, you actually have less to lose in ICM terms. Your remaining chips are worth less per unit than a medium stack's chips. This means you can (and should) be more aggressive — push wider ranges because the downside of busting is smaller relative to the upside of doubling up.

Early Tournament

ICM barely matters in the early stages. With hundreds of players remaining and payouts far away, the bubble factor is approximately 1.0. Play for chip-EV and focus on accumulating chips. ICM considerations become important only as you approach the money bubble or final table.