ICM vs Chip-EV Push/Fold: When to Tighten Your Shoves
Published 2026-10-02
Short answer
Chip-EV charts count every chip as equal while ICM prices your stack in prize money, so near the bubble the right shoving range is tighter than the chart — the chip-EV baseline on the 9-max button at 10BB is 38% of hands.
A Nash push/fold chart answers one question: which hands win chips when you shove? A tournament pays in money, and the two stop lining up once payouts are close. This guide covers what ICM changes on the bubble and at the final table, how much a Button shove at 10BB tightens in a published solver example, and the spots where a chip-EV chart is still the right tool.
What is the difference between chip EV and ICM?
Chip EV scores a play by the chips it wins on average. Every chip counts the same, whether it is your first or your last. That is the model behind every chart on RangeMyHand: each player shoves or folds, the players behind call or fold, and the payoffs are in chips (see the methodology).
ICM, the Independent Chip Model, scores a play by prize money. It turns every stack at the table into a share of the remaining prize pool, using each stack's size to estimate how often that player finishes first, second, third, and so on. The model is also called the Malmuth–Harville method: David Harville published it in 1973 for horse racing, and Mason Malmuth independently rediscovered it for poker in 1987 (Independent Chip Model, Wikipedia).
The two models disagree for one reason: in a tournament you cannot win more than first prize, but you can lose everything. Doubling your stack never doubles your share of the prize pool.
Why a chip you lose costs more than a chip you win
Take the simplest bubble there is: 4 players left, 3 places paid 50/30/20 percent of the prize pool, everyone on 10BB. The Malmuth–Harville model prices each stack like this:
| Your stack | Share of the chips | Share of the prize pool |
|---|---|---|
| 10BB (now) | 25.0% | 25.0% |
| 20BB (you double up, one player busts) | 50.0% | 38.3% |
| 0 (you bust on the bubble) | 0.0% | 0.0% |
Winning the all-in adds 13.3 points of prize pool. Losing it costs 25.0. You are risking 1.87 times what you stand to gain, a ratio players call the bubble factor. Ignoring the blinds, an all-in that gets called needs 65.2% equity to break even here, against 50.0% in chip EV. That extra 15.2 points is the risk premium.
A chip-EV chart assumes a bubble factor of exactly one. Every ICM adjustment is a way of paying that premium: shove fewer hands that get called and lose, and call shoves far tighter still.
What ICM changes on the bubble
On the money bubble the pressure is heaviest on the short stacks. Busting pays nothing, surviving one more elimination pays a min-cash, and doubling a short stack barely moves your chance of reaching the big prizes. So short stacks tighten the most.
Barry Carter's solver comparison for GTO Wizard shows the size of it. On the bubble, with a 8bb stack in the Hijack and a 1bb ante, the chip-EV strategy shoves 30% of hands and the bubble strategy shoves 12%. He puts that stack's risk premium at 18.9% (Playing Under 10bb – Part 2: ICM). What survives the cut is weighted toward suited aces, which block the hands that call and keep some equity when they are called.
The stack that covers the table gets the opposite deal. Players who cannot afford to bust must fold more against it, so a big stack can often shove wider than the chip-EV chart on the bubble.
What ICM changes at the final table
At a final table everyone is already paid, and every elimination is a pay jump. The pressure moves from the short stacks to the medium stacks. A medium stack that busts before the short stacks gives up several pay jumps it was likely to collect by folding.
The shortest stack has the least to protect. In the same GTO Wizard article, a 2bb stack under the gun at a final table carries a risk premium as low as 1.6%, which the author calls practically chip EV. The shorter you are compared with the rest of a final table, the closer the chip-EV chart is to right. If you are a medium stack with shorter stacks still in, tighten the most.
Worked example: the Button at 10BB
Chip-EV baseline (our chart)
On the Button with 10BB in a 9-max game, the RangeMyHand chart shoves 38% of hands. The closest decisions are 98o, K9o, and J9o, which the solver mixes. This is a pure shove-or-fold model with no ante. See the full 9-max Button 10BB chart.
The same seat in an ICM solver (published figures)
Andrew Brokos ran this spot both ways for GTO Wizard (Short-Stacked Play in MTTs). His solver also allows a min-raise, so its baseline differs from a pure push/fold chart.
| Button, 10bb | Shoves | Min-raises |
|---|---|---|
| Chip-EV model, all stacks 10bb | 33.9% | 5.5% |
| ICM model, 25% of the field left (SB 21bb, BB 53bb) | 30.2% | 9.7% |
The shoving range shrinks and the min-raise range grows. The hands that move are the weakest shoves: Brokos names J7s, Q6s, and Q9o as hands the chip-EV model mostly shoves and the ICM model prefers to min-raise and fold to a re-shove.
For comparison, on the RangeMyHand 9-max Button chart at 10BB, which has no min-raise option, J7s shoves 90% of the time, Q6s shoves 91% of the time, and Q9o shoves 19% of the time.
Two things to take from it. First, that example has 25% of the field left, which is before the bubble, and the range has already started to tighten. Second, the cut gets deeper as the bubble gets closer: the 8bb bubble example above falls from 30% to 12%. We have not found a public solver figure for a Button shove at exactly 10BB on the stone bubble, so we are not going to print one. The direction is certain: start from the chart, drop the mixed hands first, then the weakest pure shoves.
When chip-EV charts are the right baseline
- Far from the money. A GTO Wizard simulation study found ICM starts to change results in a statistically significant way with between 50% and 37% of the field left, for a structure that pays 15% of entrants (When does ICM become significant in MTTs?). Before the halfway mark, the measured gain from switching to ICM was within the margin of error, so chip EV is a sound model there.
- Heads-up and winner-take-all. With one prize left to play for, your share of it is proportional to your chips, so ICM and chip EV give the same answer.
- The shortest stack at a final table. As above, its risk premium is close to zero.
- Study. An ICM range depends on the payouts and on every stack at the table, so it cannot be memorized as one chart. The chip-EV range can. Learn it first, then learn which direction to move.
How to adjust a chip-EV chart for ICM
- Cut from the bottom. The mixed hands on a chart are break-even in chips, so they are the first to lose money under ICM. On the 9-max Button at 10BB that means 98o, K9o, and J9o.
- Check who can bust you. A shove into stacks that cover you risks your tournament. A shove into shorter stacks risks only part of your stack, and they are the ones under pressure.
- Tighten calls more than shoves. A shove can win without a showdown. A call always goes to showdown, so it pays the full risk premium.
- Do not fold your way to the blinds. ICM tightens ranges, it does not cancel them. A stack that is about to be blinded out has little left to protect.
For the chip-EV side of all this, start with the guide to how Nash push/fold charts are built, or look up the short definitions in the push/fold glossary.
FAQ
What is the difference between chip EV and ICM?
Chip EV measures a play by the chips it wins on average and treats every chip as equal. ICM (the Independent Chip Model) converts each stack into a share of the prize pool, and in that currency the chips you lose cost more than the chips you win are worth. Away from the money the two agree closely; near the bubble and at final tables ICM ranges are tighter.
When does push/fold strategy end?
Going up in stack size, push/fold stops being the whole strategy at roughly 15 big blinds: deeper than that, raising small with part of your range does better than shoving it. RangeMyHand still solves the shove-or-fold game up to 25BB as a study baseline. Going through a tournament, pure chip-EV push/fold ends where payout pressure begins, and from there the chart is a ceiling to tighten from.
Does RangeMyHand have ICM push/fold charts?
No. Every RangeMyHand chart is a chip-EV Nash range with no ante, because an ICM range depends on the payout structure and on every stack at the table, so no single chart can cover it. Use the chip-EV chart as the baseline and tighten from it near the money. For an exact spot, run it in an ICM calculator such as HoldemResources Calculator or ICMIZER.
Sources
Every ICM solver figure on this page is quoted from the articles below, checked on 2026-10-02. RangeMyHand's own numbers come from its chip-EV range data.
- Andrew Brokos, Short-Stacked Play in MTTs, GTO Wizard.
- Barry Carter, Playing Under 10bb – Part 2: ICM, GTO Wizard.
- Tombos21, When does ICM become significant in MTTs?, GTO Wizard.
- Wikipedia contributors, Independent Chip Model, Wikipedia.
Learn the chip-EV baseline first — free, no account needed.
Get push/fold answers in your Discord. Add the Discord bot