How ICM Works in Tournament Poker

ICM, the Independent Chip Model, uses the stacks of all remaining players and the prize ladder to estimate each player's expected tournament payout, or tournament equity. Chips are not cash: doubling a stack usually does not double its expected prize, because some of its value comes from preserving the chance to finish in a paid position.

ICM helps compare the consequences of decisions and proposed deals. It does not guarantee a payout, predict a particular finishing position or tell you how to play every hand.

What does ICM actually measure?

ICM estimates the expected remaining prize from a snapshot of the current stacks, assuming equal player ability. Its results are expressed in the units used for the payouts.

In the classic model, a player's probability of finishing first equals their share of all chips. The winner is then removed from consideration, and the next finishing position is assigned using the remaining players' chip shares. These are model probabilities—not an observed tournament win rate or a claim that individual physical chips independently determine the finish.

Three different quantities matter:

Dollar payouts produce dollar tournament equity. That is a gross prize estimate, not profit after subtracting the buy-in.

The official ICMIZER explanation identifies stacks and payouts as inputs. The numerical examples below were calculated independently.

ICMIZER help excerpt identifying stacks and payouts as inputs to the model

Original English source excerpt from ICMIZER support, showing the model's inputs rather than verifying our numerical example.

How do you calculate ICM for three players?

Find each player's probability of each finishing position, multiply those probabilities by the corresponding prizes, and add the results.

Take stacks of 50,000 for A, 30,000 for B and 20,000 for C, with prizes of $500, $300 and $200. There are 100,000 chips and $1,000 in prizes, and all three players are already paid. We use the equal-ability snapshot model, retain unrounded values throughout, and display probabilities to nine decimal places and money to two.

A's probability of finishing first is:

For A to finish second, either B or C must finish first. Remove that winner's stack before calculating A's share of the remaining chips:

The remaining probability belongs to third place:

A's expected prize is therefore:

That gives A tournament equity of $383.93. Applying the same process to B and C produces:

The guaranteed $200 is already included through the prize ladder. Do not add it to each result again.

How do you reproduce this in the calculator?

Open the ICM tool, select ICM Calculator, and enter the stacks and prizes. Results update automatically; there is no Calculate button.

The calculator supports 2 to 10 players. To reproduce the example:

1. Keep three player rows. Remove an extra row with its × control, or use the add-player control if another row is needed.
2. Enter A, B and C in the name fields, and 50000, 30000 and 20000 in their chip-count fields.
3. Enter 500, 300 and 200 in the prize fields, in descending order. Use the add-place or remove-place controls to match the paid positions.
4. Check the updated tournament equities: $383.93, $327.50 and $288.57.

Enter positive payouts for paid places only; an unpaid place does not need a zero-prize row. Dollar inputs return dollar units. If you enter payouts as percentages of the pool, results use those percentage units instead—there is no separate percent checkbox.

ICM calculator inputs for A, B and C with stacks of 50000, 30000 and 20000 and prizes of 500, 300 and 200

Inputs for the three-player example. Both calculator screenshots were captured on localhost; the interface displays decimal points.

ICM calculator results showing tournament equity and its difference from chip chop for players A, B and C

Results for the same stacks and prizes, with monetary amounts displayed to two decimal places.

The ICM equity vs chip chop section compares the model estimate with a split proportional to chips. Its ICM vs chips column shows −$116.07 for A, +$27.50 for B and +$88.57 for C. These are monetary differences, not risk premiums. Risk premium measures a difference in required hand equity, in percentage points.

The bubble-factor panel averages symmetric transfers of 1% of all chips against each opponent. It is not an all-in outcome matrix or an exact calling threshold.

For a multi-table tournament, include every remaining player—not just your table. If more than ten players remain, this calculator cannot represent the whole field.

Why can a positive chip-EV call lose tournament equity?

A call can increase your expected chip count while decreasing your expected prize because the tournament-equity loss from busting can exceed the gain from doubling.

Consider an abstract equal-stakes model: four players each have 10,000 chips, and three prizes pay $500, $300 and $200. Ignore blinds, antes, ties, differences in skill and future hands. This is not a fully specified legal poker hand or a recommendation for a real table.

Folding preserves your 10,000 chips and $250 of tournament equity. Winning a 10,000-chip contest leaves stacks of 20,000, 10,000 and 10,000, giving you $383.33 of equity. Losing means finishing fourth for zero.

At a 55% probability of winning, the expected chip count after calling is:

That is 1,000 chips more than folding. But the expected prize after calling is:

The call produces $210.83 rather than $250, a difference of −$39.17. It is positive in chip EV relative to folding but negative in ICM EV. These calculations use unrounded intermediate values; displayed monetary results are rounded to two decimals.

How do you find the calling threshold, bubble factor and risk premium?

For a two-outcome contest, the break-even win probability comes from tournament equity after folding, winning and losing:

Here $E_f$ is equity after folding, $E_w$ after winning and $E_l$ after losing. If losing earns a paid finishing prize, include that cash in $E_l$ rather than automatically using zero.

The bubble factor for this particular confrontation is the equity lost divided by the equity gained:

The ICM calling threshold is 65.22%, compared with a chip threshold of 50% in this model. The risk premium is therefore 15.22 percentage points. This distinction between ICM and chip-based required equity is also explained in GTO Wizard's discussion of risk premium.

Percentages are displayed to two decimals, but calculations retain unrounded values. Do not always subtract 50%: posted chips and dead money can change the chip-based threshold.

For a real hand, first construct the full remaining stack vectors after folding, winning and losing, including blinds and antes. A tie needs its own outcome. Hand equity includes proportional pot shares from ties, so it cannot automatically replace win probability when ICM values the resulting stacks nonlinearly.

A shove also needs the probability that the opponent folds. A calling threshold is not a shoving threshold.

How does the prize ladder change required equity?

With the same stacks and stakes, different prize distributions produce different calling thresholds—even when the total prize pool is unchanged.

Keep the idealized four-player, equal-stakes model with 10,000 chips each, no blinds, no antes and no ties. Each format has a $1,000 pool, $250 of equity after folding and zero after losing.

Each satellite ticket is valued at exactly one third of $1,000. The displayed $333.33 is rounded; entering three prizes of 333.33 would create a different total pool.

This chart compares only these payout structures under the stated assumptions. It is not a set of universal factors for early play, the bubble or the final table. GTO Wizard's bubble-factor explanation likewise relates pressure to the confrontation and payout structure.

How should different stacks respond to ICM pressure?

ICM often discourages risky calls, but it does not provide a universal strategy for big, medium or short stacks.

A big stack can pressure players who have strong reasons to avoid elimination. However, a clash with another large stack can cost substantial tournament equity. Covering an opponent reduces immediate bust risk; it does not eliminate equity risk.

A medium stack should consider the short stacks, upcoming pay jumps and which opponents cover it. Losing chips to the leader creates a different final state from losing chips to a short stack.

A short stack should not automatically fold everything. Winning can greatly improve its position, while blinds and antes make waiting costly. The actual outcomes determine whether folding is preferable.

Tighter calls do not necessarily imply equally tight shoves, because a shove can win without showdown. That fold probability is still an assumption that needs support.

ICM applies from the start of a tournament, but its practical impact varies. There is no fixed early-stage cutoff before which it can always be ignored. Postflop betting also transfers chips and changes tournament equity, although this calculator and trainer do not analyze postflop play.

Which tournament formats need different treatment?

Winner-take-all valuation is linear, equal satellite prizes can create strong elimination pressure, and bounty tournaments require more information than a conventional prize ladder.

When the winner receives the entire pool, chip share directly determines tournament equity. The nonlinear valuation associated with a multi-place prize ladder disappears.

Heads-up, second prize is already guaranteed and the difference between first and second is distributed linearly. With stacks of 6,500 and 3,500 and prizes of $650 and $350:

The equities are $545 and $455. The risk premium for a chip trade between these two players is zero. ICM still provides a value; that value simply depends linearly on the chips.

Equal satellite tickets can make preserving a paid position especially valuable. That does not establish a universal rule to fold pocket aces: stacks, stakes and outcomes still matter.

PKO and mystery bounty tournaments need bounty values, stack coverage, bounty availability and an appropriate model. A classic finishing-prize ladder alone is insufficient. A bounty does not automatically make every risky call worthwhile.

How do ICM deals compare with chip chops?

Compare an ICM deal, chip chop and adjusted chip chop using the same pool. They allocate it differently, and none is automatically the fairest agreement.

Return to A, B and C with stacks of 50,000, 30,000 and 20,000 and a $1,000 pool. A chip chop divides the whole pool by chip share. An adjusted chip chop first assigns the guaranteed $200 to each player, then divides the remaining $400 by chips.

The ordering differs by player: for A, the raw chip chop of $500 exceeds the adjusted $400, which exceeds the ICM estimate of $383.93. For B, the ICM estimate of $327.50 exceeds the adjusted $320, which exceeds the raw chip chop of $300.

Select Deal Calculator in the tool to compare deals using the entered stacks and payouts. Any agreement still depends on tournament rules and player consent, as well as their views on skill and risk.

If the rules require a prize to be left for the winner, establish the amount actually available for the deal first. Recalculate the portion being divided; do not assume the tool automatically handles an arbitrary reserve.

How can you study ICM away from the table?

Study ICM off-table by changing one input at a time and comparing outcomes, rather than looking for a universal hand chart.

Start with the reproducible three-player example. Change only one player's stack, then investigate changes to the prize ladder separately. This makes the cause of each equity change easier to identify.

To review a decision, record all players' final stacks after folding, winning and losing. When someone busts and earns a prize, remove that player from the remaining calculation and account for the earned cash separately.

Use the equity calculator to study a hand against a range. The pot odds calculator and guide to pot odds and implied odds explain the chip-based baseline. They do not replace ICM valuation of the final stack vectors.

What can you practise in the trainer?

The trainer generates small-blind shove-or-fold spots against a fixed big-blind range; it does not solve arbitrary hands.

In the ICM tool, select Push/Fold Trainer, choose Easy, and press Start training. Wait for the generated spot's equity calculation, choose shove or fold, and review the payout-EV feedback. Use Next spot to continue.

Hand equity is approximated using Monte Carlo simulation, while ICM is calculated exactly within the specified model. Fixed ranges are training assumptions, not predictions of your opponents.

The trainer does not import arbitrary calculator examples or accept a 55% win probability for a custom calling decision. It is not a full GTO or Nash-equilibrium solver.

Where do ICM's limitations matter?

ICM can calculate its model precisely without describing every important feature of a real tournament.

It does not account for differences in skill, seating, blind rotation, rising blinds or future opportunities to apply pressure. Identical stacks and payouts therefore receive identical values even when the players' practical prospects differ.

FGS, or future game simulation, models future hands and can incorporate some of these effects. It is not implemented in this calculator.

Use ICM as a transparent baseline for comparing prizes and risk. Decisions driven by future blinds, complex ranges, postflop betting or bounties need richer analysis. Tournament equity remains a model estimate, not a certified payout or a promise of profit.