Final Table Strategy: ICM, Pay Jumps, and Making a Deal
Final table strategy is ICM strategy: every bust-out is a pay jump, so chips you can win are worth less than chips you can lose. Computed for a standard 9-handed $100k final table: the chip leader's 28.5% of chips is worth only 18.2% of the pool. Big stacks attack, medium stacks tighten, short stacks shove early — and before you agree to any deal, check the ICM split vs the chip chop: in our 3-handed example the chip chop overpays the leader by $1,607.
Reaching a final table converts a chip-accumulation contest into a dollar-preservation one. This is the spoke of our tournament strategy hub where the money math gets loudest — every number below is computed with the site's own ICM engine, the same one behind the ICM calculator.
What your stack is actually worth
Here is a full 9-handed final table — $100,000 pool, standard 30/20/14/10/7.5/6/4.5/4/4 payouts — with every stack converted to dollars by ICM (Malmuth-Harville):
| Stack | % of chips | ICM equity | % of pool |
|---|---|---|---|
| 4.5M | 28.5% | $18,216 | 18.2% |
| 3.2M | 20.3% | $15,749 | 15.7% |
| 2.4M | 15.2% | $13,819 | 13.8% |
| 1.8M | 11.4% | $12,068 | 12.1% |
| 1.4M | 8.9% | $10,705 | 10.7% |
| 1.0M | 6.3% | $9,135 | 9.1% |
| 0.7M | 4.4% | $7,786 | 7.8% |
| 0.5M | 3.2% | $6,793 | 6.8% |
| 0.3M | 1.9% | $5,728 | 5.7% |
Two things to internalize. The compression at the top: 28.5% of the chips buys only 18.2% of the money, because the leader can still only win one first prize. And the floor at the bottom: the 300k micro stack owns $5,728 — 5.7% of the pool — by existing. That floor is what laddering protects and what reckless calls burn. Enter your own final table in the ICM calculator to see your seat's real value, and read what is ICM if the model is new to you.
Playing each stack size
- ✓ Big stack: apply pressure where it hurts. Your targets are the medium stacks — they lose the most equity by busting, so they must fold hands that beat you. Attack their blinds and 3-bet their opens. Do not pick fights with other big stacks; that is the one clash that can cost you the laddering premium.
- ✓ Medium stack: the ICM vise. You can afford neither the call-offs of the big stack nor the desperation shoves of the short one. Tighten your calling ranges hard, keep 3-bet/folding to a minimum, and let the short stacks bust first. This is the seat where discipline is worth actual dollars.
- ✓ Short stack: shove while it still threatens. A 10bb shove folds out real hands; a 4bb shove gets called by everything. Use the push/fold chart as the baseline and tighten when the next pay jump is one bust away.
- ✓ Heads-up: ICM is over. With one pay jump left, the game reverts to pure chip EV — the model that made you fold earlier now says gamble. Many players fail to flip that switch and get ground down.
Ladder math: when folding is printing
Look back at the table: 9th pays $4,000 and 7th pays $4,500, but the equity gap between the 300k and 700k stacks is over $2,000. Every stack that busts before you lifts your worst-case finish. When two bigger stacks are in a raising war, the correct play with a marginal hand is almost always to get out of the way and let them collide — one of them busting can be worth more to you than winning a small pot. The mirror image also holds: if everyone at the table is visibly laddering, the player who isn't gets to steal at will. Notice which one is happening at your table.
Deals: ICM chop vs chip chop
Most final tables end in a deal, and the deal type decides who pays for it. Three players left, $100,000 still on the table (50/30/20), stacks 5M / 3M / 2M — both standard splits, computed:
| Stack | ICM chop | Chip chop | Chip-chop error |
|---|---|---|---|
| 5M (50%) | $38,393 | $40,000 | -$1,607 |
| 3M (30%) | $32,750 | $32,000 | +$750 |
| 2M (20%) | $28,857 | $28,000 | +$857 |
The chip chop (everyone locks up 3rd-place money, the rest split by chip count) pays the leader $40,000 — $1,607 more than the stack is worth — and takes that money from the two shorter stacks. This is why chip leaders propose chip chops and why you should counter with ICM numbers. The ICM calculator's deal mode computes both splits side by side for your exact stacks and payouts, ready to read out at the table. Legitimate reasons to deviate from ICM exist — a genuine skill edge, a huge blind level about to hit, personal variance tolerance — but know the fair number first.
Malmuth-Harville ICM ignores position, blind pressure, skill edges, and who is about to take the big blind. It is the standard for valuing stacks and deals, but at the table it is a compass, not a solver — we do not claim these numbers are GTO play, only that the equity math is computed exactly for the stated stacks.
Where these numbers come from
- • ICM equities: computed with the site's Malmuth-Harville implementation (app/lib/icm.js — the engine behind the ICM calculator) for the exact stacks and payout structures stated beside each table.
- • Chip chop: the standard casino formula — every remaining player locks the lowest remaining payout, and the rest of the pool splits proportionally to chip counts. Same formula as the calculator's deal mode.
- • Strategy guidance is exactly that — guidance. No postflop solver output is claimed anywhere on this page.
Common questions
What is the best strategy at a poker final table?
Play your stack size against the pay jumps. Big stacks should attack medium stacks (who lose the most dollar equity by busting), short stacks should shove before blinding into irrelevance, and medium stacks should tighten up and let the shorter stacks bust first. All of it comes from ICM: every elimination is a pay jump, so surviving one more bust-out has direct dollar value that a chip count doesn't show.
How does ICM change final table play?
It makes calling all-ins much more expensive than shoving. Computed with this site's ICM engine for a 9-handed $100k final table (standard 30/20/14/10/7.5/6/4.5/4/4 payouts): the chip leader holds 28.5% of the chips but only 18.2% of the prize pool, while a 1.9%-of-chips short stack already owns 5.7% of it. Chips you might win are worth less than the chips you'd lose, so marginal calls that are fine in a cash game torch money at a final table.
What is laddering at a final table?
Deliberately folding into pay jumps — letting shorter stacks bust before you so you move up the payout ladder. It is correct more often than aggressive players like: in our computed 9-max example the difference between 9th and 8th/7th is a real dollar jump, and a micro stack still holds several percent of the pool by simply existing. It becomes a leak when a big stack uses your obvious laddering to steal every pot you could have defended profitably.
Should I accept a final table deal (chop)?
Compare the two standard numbers before you answer. A chip chop pays everyone their chip percentage of the remaining money and systematically overpays the big stack: in our computed 3-handed example ($100k on the table, 50/30/20 payouts, stacks 5M/3M/2M) the chip chop hands the leader $40,000 while ICM values the same stack at $38,393 — the two shorter stacks give up $750-$857 each by agreeing. An ICM chop is the fairer default; deviate from it consciously (skill edge, fatigue, variance tolerance), not because someone at the table sounds confident. The ICM calculator's deal mode computes both splits for your exact stacks.
How do I play a short stack at the final table?
Shove or fold, and shove before you are desperate — the fold equity of a 10bb shove is the asset, and it evaporates below ~5bb. Pick spots against the stacks that can least afford to call you (medium stacks under ICM pressure), not the chip leader who calls light. Our push/fold chart gives the baseline heads-up Nash ranges by stack depth; tighten them when a pay jump is imminent.
Tools for the final table
Keep reading
New to poker? Take the free 15-minute crash course →