ICM explained: why a chip isn’t a dollar.
The Independent Chip Model prices what your stack is actually worth in money rather than in chips. It is the reason correct tournament play folds hands a cash game would call, and the reason a big stack can bully a table without ever showing a hand.
Tournament chips cannot be cashed out — they buy a share of a fixed, top-heavy prize pool, so your stack is always worth less than its face value. That raises the equity a call needs above 50%, hands you leverage over anyone who cannot afford to call, and switches off completely the moment nobody can bust below you.
Why a chip is not a dollar.
Four reasons, and none of them is complicated. Together they are the whole model — everything the calculator does is arithmetic layered on top of these.
You cannot cash chips out
In a cash game a chip is a dollar because you can stand up and take it to the window. Tournament chips have no cash value at all — they are only a claim on a share of a fixed prize pool, and that claim is not linear.
The prize pool is top-heavy
First place typically takes a multiple of what second does. Because the money is concentrated at the top, the chips that carry you toward first buy less and less additional equity the more of them you already hold.
Busting has a floor and a ceiling
Once you are in the money, elimination pays something — and that something is guaranteed the moment you fold. Doubling up, by contrast, is never guaranteed and never doubles your equity. The two outcomes are not mirror images.
Other people's stacks are part of your equity
A short stack at your table who is about to bust is worth money to you without your playing a hand. That is why the same holding can be a call at one table and a fold at another with identical stacks in front of you.
What your stack is really worth.
One table, five stacks, priced two ways. The middle column is what a chip count says you own. The one beside it is what you would actually be paid. Every row moves in the same direction, and the direction depends entirely on stack size.
Five players, a $1,000 pool paid 50% / 30% / 20% to the top three. The chip leader is the only player who loses — 45% of the chips buys under 35% of the money, because the two thirds of the pool below first place are already within reach of everyone else. Every short stack gains, which is precisely why they can call you far wider than you would like.
The curve, and why it bends.
Plot chip share against real-money share and the line does not run straight. It rises steeply at the bottom, where a few chips buy real survival, and flattens at the top, where you already have most of what the ladder can pay. The gap between that curve and the diagonal is what ICM charges.
Double your stack and your real-money equity climbs far less than double. That curve — diminishing returns on every chip — is the entire reason ICM forces folds a cash game never would.
Four ways to use it, not fear it.
The half of ICM that most players skip. It is not only a brake on your own calls — it is a lever on everyone whose tournament can end, and knowing exactly whose it can is where the money is.
Attack the stacks that can fold
ICM pressure is aimed, not sprayed. The medium stack with a pay jump to protect has the highest folding frequency at the table, because calling risks a real, calculable amount of money. That is who you raise into, every orbit.
The player with something to lose is the player who will let you have it.
Never pressure the committed
A stack short enough that folding also kills them has almost no ICM tax to pay — their equity is already close to the floor. They will call correctly and wide. Jamming into them converts a weapon into a coinflip you did not need.
Nobody folds when folding costs the same as calling.
Know exactly who covers whom
Only a player who covers you can end your tournament, which means only they can charge you the full ICM tax. Against anyone you cover, the risk is capped at their stack. Stack sizes decide who can apply pressure to whom, before cards are dealt.
Leverage is not about your hand. It is about whose tournament can end.
Widen the instant the tax disappears
The moment nobody can bust below you, the whole calculation collapses back to chip EV and the correct range roughly triples. Players who keep folding at this point routinely reach heads-up and rarely win it.
The discipline that got you here is the thing that now costs you the title.
What a call actually needs.
If you remember one thing from this page, make it this. Calling off your stack is break-even at 50% equity in chips — always, everywhere. In money it is not, and the gap between the two is what ICM charges you at each stage of a tournament.
Read it as a tax on calling. On the bubble you need roughly 64% to justify putting your tournament at risk, so a hand that is a clear call in a cash game is a fold — and the identical hand becomes a call again heads-up, where the tax falls to nothing. These are computed for even stacks at each stage; your own spot moves with who covers whom.
A chip-chop is not a fair deal.
Sooner or later someone at a final table suggests splitting the money by chip count. It sounds even-handed and it is not, because it ignores the prize every remaining player has already locked up. Here is the same three-handed table divided both ways.
Both columns add up to the same $10,000. The difference is who gets it. A chip-chop pays the leader for chips they have not converted yet and ignores that the short stack has already banked third-place money by surviving to this point. Here it moves $1,876 from the short and middle stacks to the leader.
If you are the short stack, ask for ICM — it is the standard every tournament floor uses and nobody can reasonably object. If you are the leader, understand that a chip-chop is you asking for a premium, and expect to be told no. Either way, run the actual numbers first with our ICM calculator rather than agreeing to a split at the table.
Satellites: where chips stop mattering.
Every seat pays the same, so the prize list is flat rather than top-heavy. That turns the usual diminishing returns into a cliff: once you have enough chips to qualify, additional chips buy you almost nothing at all.
6 players, 3 seats, each worth $1,000. Look at the last two rows: going from a third of the chips to half of them adds barely anything, because a seat is a seat and you cannot win two. That is why satellite play looks so strange from the outside — correct strategy is to fold hands that would be automatic calls anywhere else, and why a big stack should almost never take a flip once qualification is in sight.
Where the model stops being true.
ICM is the best tool available for pricing a tournament decision and it is still a simplification. Knowing exactly where it is wrong is what separates using it from obeying it.
It assumes everyone plays identically
The model distributes finishing positions purely by stack size. A strong player is worth more than it says and a weak one less, which is why good players correctly deviate toward taking marginally more risk than a raw ICM read suggests.
It cannot see position or blinds
Being one hand from the big blind, or sitting to the left of the only aggressive stack, changes your real equity and does not appear anywhere in the calculation. ICM prices the ladder, not the seat.
It is a snapshot, not a plan
Every figure it returns describes this exact moment. It has no model of the next orbit, so it will happily recommend a fold that leaves you unplayably short two hands later.
Harville is an approximation
The standard model assigns later places by renormalising the remaining stacks, which slightly misprices the middle positions. Alternatives exist and disagree at the margins. For decisions at a table the difference is far smaller than the difference between using ICM and not.
The ICM cheat sheet.
Everything above, compressed into what the model tells you to do and what it does not.
- Raise into medium stacks who have a pay jump to protect — they cannot call.
- Demand ICM numbers, not a chip-chop, when a final table discusses a deal.
- Check who covers whom before you decide anything about a marginal hand.
- Open your range dramatically the moment the last elimination is behind you.
- Don't read ICM as an instruction to fold; half of it is an instruction to attack.
- Don't apply pressure to a stack short enough that folding kills them anyway.
- Don't fold into oblivion — the model cannot see the blinds coming for you.
- Don't carry any of it into heads-up, where the ladder no longer exists.
A claim, a tax, and an off switch.
A chip is a claim, not a coin.
Tournament chips cannot be cashed out; they buy a share of a fixed, top-heavy pool. That is the entire model, and everything else on this page is arithmetic that follows from it.
It moves the bar, in both directions.
On the bubble a call for your tournament life can need 64% equity instead of 50%. The same maths makes everyone else fold too often — which is the half of ICM that makes money rather than saves it.
And then it switches off.
Heads-up there is no ladder, so chips are worth face value again and the correct range widens enormously. Knowing when the model stops applying matters as much as knowing what it says.