ICM Calculator: What Your Tournament Chips Are Really Worth

In a tournament, chips are not money. You cannot cash them out, and the prizes are fixed by finishing position: whoever busts sixth gets sixth-place money no matter how many chips they held an hour earlier. Typically only the top 10 to 15 percent of the field is paid at all, with the money weighted heavily toward the top. That creates a problem no cash-game player ever faces - doubling your chips does not double your money. (If tournaments are new to you, start with what an MTT is.)

The Independent Chip Model solves that problem. It takes the current stacks and the remaining payouts and answers one question: if the tournament were paid out fairly right now, what would each stack be worth in dollars? That number is your ICM equity. It is what deal-makers use to split prize pools at final tables, and it is the yardstick that tells you whether a chip gamble is also a money gamble - near the payouts, the two diverge far more than you expect.

Note what ICM does not answer: it never tells you what to do with a hand. It tells you what your stack is worth. Strategy is what you build on top of that number.

The model behind the calculator is called Malmuth-Harville, and the whole thing rests on one assumption plus one loop.

The assumption: your chance of finishing first equals your share of the chips. Hold 5000 of the 10000 chips in play and the model gives you a 50% chance of winning the tournament. That is the entire modeling step - no cards, no positions, no reads.

The loop handles every other finishing position. To find your chance of finishing second, the model walks through each opponent in turn and says: suppose this player wins, with probability equal to their chip share. Remove the winner from the table. Your chance of finishing second is now your share of the chips that remain. Add that up across every possible winner and you have your total probability of second place. Then remove each possible second-place finisher and repeat for third, and so on down the ladder until every paid position is covered.

The last step is just accounting. Multiply your probability of each finish by the prize for that finish, add them together, and the sum is your equity in dollars. The calculator above runs this recursion exactly for every player at once - it enumerates every finishing order rather than sampling, so the numbers never wobble from run to run.

Run the recursion on a simple spot and something strange falls out. Three players, $100 prize pool paying $50 / $30 / $20, stacks of 5000, 3000, and 2000 - so the players hold 50%, 30%, and 20% of the chips.

• 5000 chips (50% of chips) is worth $38.39 - only 38.4% of the money
• 3000 chips (30% of chips) is worth $32.75
• 2000 chips (20% of chips) is worth $28.86 - 28.9% of the money

The equities compress toward the middle, and the recursion shows you exactly why. The short stack is protected from below: even the 2000-chip player has real chances of climbing into second-place and first-place money, and can never be worth less than a shot at the $20 for third. The chip leader is capped from above: no matter how many chips they pile up, they can never win more than the $50 for first. Half the chips cannot buy half the money because first place is a ceiling.

That is the core lesson of ICM: each additional chip buys less money than the one before it. Which leads to the classic line, and it is worth memorizing: the chips you win are worth less than the chips you lose. Doubling your stack never doubles your equity, but busting always zeroes it. Rebuild this exact spot in the calculator above and change the stacks - the compression never goes away, it only changes shape.

The ICM tax is the gap between your share of the chips and your share of the money. The calculator shows it as its own column, and the default demo is built to make it visible: five players, $280 prize pool paying $100 / $70 / $50 / $35 / $25.

• 9000 chips = 30.0% of the chips, worth $66.11 = 23.6% of the pool - the leader pays 6.4 points of ICM tax
• 7500 chips = 25.0%, worth $61.98 = 22.1% - pays 2.9 points
• 6000 chips = 20.0%, worth $57.07 = 20.4% - collects 0.4 points
• 4500 chips = 15.0%, worth $51.12 = 18.3% - collects 3.3 points
• 3000 chips = 10.0%, worth $43.72 = 15.6% - collects 5.6 points

Read the table top to bottom and you can watch the money flow: the two big stacks pay 9.3 points of tax, and the three smaller stacks collect the same 9.3 points as a subsidy. It is not a punishment for playing well - it is the ceiling of first place pushing value downhill. The bigger your stack relative to the field, the more of your chip share the ceiling shaves off.

Hold on to that 6.4-point number. It is the exact amount a chip leader stands to overcharge you in a deal, and we will come back to it below.

The bubble is the moment when one more elimination puts everyone else in the money. This is where ICM stops being an accounting curiosity and starts deciding hands. Take four players with only the top three paid $50 / $30 / $20, stacks 8000 / 6000 / 4000 / 2000. The calculator prices them at $33.60, $29.49, $23.59, and $13.32.

Now the hand that decides tournaments. The chip leader shoves, and the 6000 stack holds a hand that is a pure coin flip - 50/50 in chips. Call or fold?

In dollars it is nowhere close to 50/50. Win, and the 12000-chip stack is worth $40.45, a gain of $10.96. Lose, and the stack is worth $0, a loss of $29.49. Risking $29.49 to win $10.96 means the call needs about 73% equity just to break even. A coin flip in chips is a disaster in dollars on the bubble.

That one asymmetry explains the whole texture of bubble play:

• Big stacks bully. The leader can shove relentlessly, because a caller in a spot like the one above needs roughly 73% equity, and almost no hand is 73% against a shoving range. The big stack is not bluffing with cards - it is bluffing with the payout structure.
• Medium stacks go quiet. They have the most to protect: real dollar equity that one lost confrontation erases. Hands they would call in a heartbeat in a cash game become folds.
• Short stacks pick their spot. The shortest stack has the least equity left to protect, so it is the one player who can justify taking a stand - though even a short stack profits from folding while two bigger stacks collide.

The same force operates at every pay jump - each step up in prize money as another player busts - and it is strongest wherever the jumps are steepest, which is usually the final table. If you have studied our push/fold chart, note that it is computed in chip EV - chips won and lost, ignoring payouts. This page is the reason you tighten those ranges as the money approaches: the chart is the baseline, ICM is the correction. Part Ten drills ICM near pay jumps hand by hand.

When the last few players agree to split the remaining prize pool, there are two common ways to do it:

• A chip chop splits the money in proportion to chip counts.
• An ICM chop splits it according to each player's ICM equity - the numbers the calculator above produces.

You now know why the difference matters. In the default demo, the chip leader holds 30.0% of the chips but a stack worth only 23.6% of the pool. A chip chop hands that player 30.0% of the money - overpaying them by exactly the ICM tax, the 6.4 points from the last section - and shorts everyone else, with the shortest stacks giving up the most. When the big stack pushes hard for a chip chop, this is why.

An ICM chop is the standard fair split, and the model on this page is the same exact computation those deals are priced with. Before you agree to anything at a final table, enter the real stacks and payouts into the calculator and see what your stack is actually worth. Two minutes of typing can be worth a pay jump.

Three situations, in order of how much money is on the line.

• Deals. Enter the stacks and payouts before agreeing to any final-table split, and compare the ICM numbers against whatever was proposed. You get the fair split in seconds - and a number to negotiate from.
• Bubble and pay-jump decisions. Before you talk yourself into a marginal call near the money, price it. Run the spot three times: current stacks, the stacks if you win the all-in, the stacks if you lose. The gain and the loss in dollars tell you the equity you needed - the same arithmetic as the 73% flip above.
• Study. Recreate spots from tournaments you played, move chips between stacks, and watch how equity responds. The intuitions - compression, the tax, bubble asymmetry - only become automatic after you have watched the numbers move a few dozen times.

If push/fold play is new to you, Part Nine covers the M-ratio - how many rounds of blinds and antes your stack can pay - and when your stack enters shove-or-fold territory, and the lessons index gives you structured spots to work through.

ICM is a model, and you should know exactly what it leaves out before you lean on it.

• It assumes equal skill. The recursion knows chip counts and nothing else. If you are the best player at the table, your true equity is higher than the model says; if you are outmatched, lower.
• It ignores the blinds about to hit you. A 2000-chip stack with the big blind arriving next hand and one that just paid it get the same equity. In reality they are very different stacks.
• It ignores position and who acts next. The model prices the stacks, not the hand in progress or who holds the initiative.
• It is a snapshot, not a strategy engine. It values the current position; it does not play the game forward, and rising blinds and antes are invisible to it.

Serious study software layers future-game simulation on top of ICM to patch some of these gaps. But the model in the calculator above - exact Malmuth-Harville - is not an approximation of those tools. It is the standard the poker world actually uses when real money gets split, and the right first lens on every spot near the money. Know what it measures, know what it misses, and it will stop you from paying the tax by accident.

ICM stands for Independent Chip Model. It is the standard method for converting tournament chip stacks into real-money equity, based on the remaining payouts and every player's chip count.

The standard method is Malmuth-Harville. It assumes each player's chance of finishing first equals their share of the chips. For second place, it considers each possible winner in turn, removes them, and gives each remaining player a chance equal to their share of the remaining chips - then repeats for third and so on down the ladder. Each finishing probability is multiplied by its prize and summed. The calculator on this page runs that computation exactly, with no simulation.

The ICM tax is the gap between your share of the chips and your share of the prize money. When prizes are spread across several finishing positions, big stacks pay it and short stacks collect it, because no stack can be worth more than first-place money. In the calculator's default demo, a stack holding 30.0% of the chips is worth only 23.6% of the pool - 6.4 points of tax.

An ICM chop is a deal where the remaining players split the prize pool according to their ICM equities instead of their raw chip counts. It is the standard fair method for final-table deals. Enter the stacks and payouts in the calculator above to get the exact split.

A chip chop splits the remaining money in proportion to chips; an ICM chop splits it according to model equity. Because equities compress toward the middle, a chip chop systematically overpays the chip leader by exactly the ICM tax and shorts everyone else. If you are not the chip leader, insist on ICM numbers.

No. ICM applies whenever payouts are fixed by finishing position and stacks differ - most sharply on the money bubble and at every pay jump. Final tables are simply where it gets computed most often, because that is where deals happen. Deep in a tournament with the money far away, chip value and dollar value are close enough that chip EV is a fine guide.

No. In a cash game a chip is worth its face value at all times - you can stand up and cash out, so 100 chips are worth exactly twice 50 chips. ICM only exists because tournament prizes are fixed by finishing position, which caps what a big stack can win and protects what a short stack can lose.

Because losses cost more equity than equivalent wins gain. In the bubble example on this page, a 50/50 all-in risks $29.49 of equity to win $10.96, so it needs about 73% equity in the hand just to break even. Hands that are clear chip-EV plays become clear money-losing plays.

Because first place is a ceiling. Extra chips can only improve your chances up to the point of winning the top prize, so each additional chip buys less money than the one before it. With half the chips three-handed in the example on this page, the leader holds only 38.4% of the money.

ICM pressure is the strategic squeeze the payout structure puts on medium stacks near pay jumps: calling off risks far more equity than winning gains, so they must fold wide while big stacks can shove wide. The bubble section above shows the exact numbers behind it.

Exact. It computes the full Malmuth-Harville recursion over every finishing order for 2 to 12 players, so the same inputs always produce the same equities to the cent. Nothing is sampled or approximated.

No. The model sees only chip counts and payouts. It assumes all players are equally skilled and ignores position, who acts next, and the blinds about to hit each stack. It is a snapshot of stack value, not a strategy engine - serious study tools layer future-game simulation on top, but exact ICM remains the standard for pricing stacks and making deals.