What Your Chips Are Actually Worth: ICM, Risk Premiums, and How Every Decision Changes as the Tournament Goes On

Here is the sentence that separates casual players from hosts who actually understand what’s happening at their table:

A chip is worth less than a chip.

At first that sounds like nonsense. You bought 10,000 chips for $20. Each chip is worth two tenths of a cent. Done.

It is not done. That accounting treats a tournament like a cash game, and tournament poker is not cash game poker. The mathematics of what a stack is actually worth in dollars shifts every time a player busts, every time blinds go up, every time the payout ladder tightens. If you do not understand how, you will — mathematically, literally — light money on fire at the table.

This post is a rigorous tour of the Independent Chip Model (ICM), the framework tournament pros use to translate chips into dollars at any point in a tournament. It covers the math, the intuition, and most importantly how it should change your play. It also covers the adjustment for mystery bounties, which tear holes in the model in interesting ways.

The sentence that breaks casino-brain

In a cash game, a $5 chip is worth $5. Stand up, walk to the cage, collect $5. Equity in the game is trivially linear: 1,000 chips equals $1,000.

Tournaments have a different accounting problem. You cannot cash out your stack mid-game. Your 10,000 chips are only worth money if you convert them into a finishing position on the payout ladder — and the payout ladder is not linear:

8-player, $20 buy-in, $160 pool, top 3 paid

Place	Payout	% of pool
1st	$88	55%
2nd	$48	30%
3rd	$24	15%

If you hold half the chips in play (40,000 out of 80,000), how much is your stack worth? Naive math says half the pool, so $80. That is wrong. You might bust fourth and win $0. You might win $88. Your expected payout sits between those extremes. ICM gives you the number.

ICM in one paragraph

ICM assigns each stack a dollar value equal to its probability-weighted share of the remaining prize pool, where the probability of finishing Nth is computed by treating relative chip stacks as lottery tickets. The only inputs are the current stacks and the payout schedule. The only output is each player’s dollar equity right now.

The intuition: a bigger stack is more likely to finish higher and less likely to bust next. You are not only “closer to the win” with chips — you are also buying insurance against busting before the money.

Worked example: three-player final, $1,000 pool

Three players remain. Payouts: 1st $600, 2nd $300, 3rd $100. Stacks:

Alice: 50,000
Bob: 30,000
Carol: 20,000

Naive (chip-proportional) math:

Player	Stack %	Naive $ value
Alice	50%	$500
Bob	30%	$300
Carol	20%	$200

Intuitive. Also wrong. Every number above is off.

ICM math. Treat each chip as a lottery ticket for first place:

P(Alice 1st) = 50/100 = 0.500
P(Bob 1st) = 30/100 = 0.300
P(Carol 1st) = 20/100 = 0.200

For second place, condition on who finished first and rerun the lottery over remaining stacks:

P(Alice 2nd) = P(Bob 1st) × 50/70 + P(Carol 1st) × 50/80 = 0.30 × 0.714 + 0.20 × 0.625 = 0.214 + 0.125 = 0.339
P(Alice 3rd) = 1 − 0.500 − 0.339 = 0.161

Do the same for each player. The full equity table:

Player	P(1st)	P(2nd)	P(3rd)	ICM $ equity
Alice	0.500	0.339	0.161	$417.80
Bob	0.300	0.375	0.325	$325.00
Carol	0.200	0.286	0.514	$257.20

Equity = P(1st) × $600 + P(2nd) × $300 + P(3rd) × $100. Totals to exactly $1,000.

Now look what happened:

Alice lost $82 to the naive model (she has 50% of chips but only 41.8% of the dollars).
Bob gained $25 (30% of chips, 32.5% of dollars).
Carol gained $57 (20% of chips, 25.7% of dollars).

This is the central ICM insight: chips are worth progressively less the more of them you have. The first chip doubles your equity (you are no longer out). The millionth chip is nearly worthless (you are guaranteed first anyway).

The corollary is the one players feel viscerally: short stacks are worth more dollars per chip than big stacks. Carol’s 20,000 chips are worth $257.20 (roughly 1.29 cents per chip). Alice’s 50,000 chips are worth $417.80 (roughly 0.84 cents per chip). The short stack’s chips are 54% more valuable per unit.

Why this happens: the concavity of equity

Dollar equity is a concave function of stack size. Double your chips, your equity increases by less than double.

This has one enormous consequence: in almost every tournament spot, doubling up is worth less to you than busting is expensive.

Let us quantify. Back to Alice at 50,000 chips and $417.80 equity. Bob (30,000) shoves. Alice calls.

Alice wins the flip. Stacks become: Alice 80,000, Carol 20,000. Two players left, payouts reduce to 1st $600, 2nd $300 for Alice and Carol.

P(Alice 1st) = 80/100 = 0.80
Equity = 0.80 × 600 + 0.20 × 300 = $540

Alice gained $122.20.

Alice loses the flip. Stacks become Alice 20,000, Bob 80,000, Carol 20,000.

P(Alice 1st) = 20/120 = 0.167
P(Alice 2nd) = 0.667 × (20/40) + 0.167 × (20/100) = 0.367
P(Alice 3rd) = 0.467
Equity = 0.167 × 600 + 0.367 × 300 + 0.467 × 100 = $256.90

Alice lost $160.90.

Net: a 50/50 coinflip for Alice. Expected equity change = 0.50 × $122.20 + 0.50 × (−$160.90) = −$19.35.

Coinflipping with identical chip odds is not zero-EV in a tournament. In this spot, Alice loses nearly twenty dollars every time she takes a 50/50. She needs to be a 57% favorite in hand just to break even:

$p \cdot 540 + (1 - p) \cdot 256.90 = 417.80 \Rightarrow p = 0.568$

That extra 7% of required equity is called the risk premium. It is the tax ICM charges you for gambling with your tournament life.

The stages — and what your chips are actually worth at each

Chip value shifts through the tournament. Below is an illustrative table for an 8-player, $220 pool game (1st $121 / 2nd $66 / 3rd $33). Stacks in one consistent hypothetical path. The “discount” column is how much less the stack is worth in dollars vs. proportional-of-the-pool math.

Stage	Players in	Your stack	Naive $	ICM $	Discount
Start	8	10,000	$27.50	$27.50	0.0%
After lvl 4	7	12,000	$29.70	$28.55	−3.9%
Mid (rebuys done)	5	15,000	$36.67	$32.90	−10.3%
Pre-bubble (4 left)	4	25,000	$55.00	$48.60	−11.6%
Bubble (you medium, one stack short)	4	25,000	$55.00	$51.40	−6.5%
Just in the money	3	30,000	$66.00	$61.30	−7.1%
Heads-up	2	50,000	$93.50	$93.50	0.0%

A few things jump out:

Start of tournament: chip value equals dollar value per buy-in. Cash-game math works perfectly.
Middle of the tournament: the discount grows. Your chips are literally worth less than their chip-proportional share of the pool.
Bubble: the discount shrinks slightly if there is a stack short enough to bust below you. You are being paid a bubble premium for surviving.
In the money: every elimination resets the ladder. Your equity jumps up at each bust (you move one position up the payout ladder without winning a chip).
Heads-up: ICM evaporates. One payout difference between winning and losing. Chip EV and dollar EV reconverge.

Risk premiums as rules of thumb

Running ICM at the table is impractical. Internalize the rough risk premiums instead.

Situation	Required extra edge above 50%
Early (first third of tournament)	0–2%
Mid (middle third)	3–7%
Pre-bubble (one off the money)	8–15%
Bubble, you are covered (short stack alive)	15–25%
Bubble, you are the shortest	negative — be the gambler
Final table pay-jump (e.g., 4th→3rd)	5–12%
Heads-up	0%

“Risk premium” means: how much of a chip-EV edge you need to break even on dollar EV when your tournament life is on the line. At the bubble covered, you need to be a 60–65% favorite to justify calling an all-in that a cash player would snap-call at 51%.

How ICM should actually change your play

This is where the theory becomes action.

Stage 1 — Early: play like cash, mostly

ICM distortions are inside your noise floor. Play well, accumulate, do not agonize over a 2% risk premium. The biggest mistake here is playing too tight because you are “protecting your tournament life” — at this stage, your tournament life is cheap.

Stage 2 — Mid: stop coinflipping

At 3–7% risk premium, calling an all-in with AQ vs. a random shove is no longer trivially correct. You need real edge, not coinflip equity. Open your stealing ranges (chip-EV positive spots stay positive); tighten your calling ranges (chip-EV-marginal spots go negative).

Stage 3 — Pre-bubble: punish the scared

The short stacks are desperate. The medium stacks are terrified of busting. The big stack (you, hopefully) has free license to open every orbit. Every fold from a medium stack transfers chips to you at full chip-EV value but at discounted ICM cost to them. This is bubble abuse — structurally the highest-EV spot in tournament poker.

Simple example: if you are the chip leader and a medium stack folds a hand where they had 55% equity preflop, they are losing about 8 cents per dollar of stack they folded, and you are gaining the blinds at 100% EV. That asymmetry compounds.

Stage 4 — Bubble, you are covered: fold premium

If you hold 20 BB and a short stack at the table has 4 BB, tighten to premium pairs and AK, maybe AQ. Doubling up is worth less than busting costs. Wait for the short stack to bust. Your equity grows just by surviving — you can gain dollars without playing a hand.

Stage 5 — Bubble, you are shortest: gamble

Flip the last strategy. When you are the short stack about to bust the bubble, other players will tighten up against you. Your shove-fold range widens substantially. Any two cards 40–50% of the time, depending on stack depths — because if you fold to death, you finish fourth for $0. The upside of winning chips is full chip-EV (you are not protecting anything).

Stage 6 — Final table pay jumps: ladder first, attack second

Goal is not “win the tournament.” Goal is maximize expected payout. For most home-game pay curves, that means playing tight against equally-stacked opponents and aggressive against short stacks that are about to ladder you up.

Stage 7 — Heads-up: cash-game brain

ICM is dead. First pays 2× second. Every chip you win equals its chip-proportional share. If you carry mid-tournament habits into heads-up you will play far too passive. Shove wide, call wide, take variance. This is where tournaments get won.

Mystery bounties: the ICM wildcard

Mystery bounties tear visible holes in ICM. They add per-elimination cash that bypasses the payout ladder. Busting the short stack now has cash value independent of how far you’ve laddered.

Quantifying the adjustment: an expected bounty value of $B means that each elimination is worth $B more than chip-proportional EV would suggest. At the bubble, that cash offsets some of your ICM risk premium.

Worked example. You are the big stack. Short stack shoves. Bounty pool average is $15. Main prize pool is $220, which means $15 is 6.8% of the pool size.

Without bounties, you might need to be a 62% favorite in hand to call. With bounties, you get $15 in cash if you win, entirely off-ladder. That is worth roughly 6.8% of pool equity. Your required edge drops from 62% to about 55%. You are now calling much wider.

The practical shift: mystery bounties make the chip leader the most aggressive player at the final table, because they cover everyone and collect cash on every knockout. This is precisely why the format is so entertaining to watch. The pros playing mystery bounty formats at the WSOP shove wildly wider at the final table than they would in a flat format.

For designing the bounty pool and distribution so this effect lands correctly, see the mystery bounty guide.

Using the ICM tool that ships with Poker Timer

Poker Timer ships a companion CLI tool (apps/icm/) that computes ICM equity for any stack configuration and payout schedule. Typical use:

icm --stacks 50000,30000,20000 --payouts 600,300,100

Output:

Stack     Equity    % of pool
50,000    $417.80   41.78%
30,000    $325.00   32.50%
20,000    $257.20   25.72%

The tool also plots the equity curve across all possible single-stack sizes, which is useful for sanity-checking deal proposals. A common situation: three players at the end of a home game propose a chop “split it by chip counts.” That is the naive model. It systematically overpays the chip leader and underpays the short stacks. An ICM chop is the fair split. Run the tool on your actual stacks and show the table the numbers; arguments evaporate.

The cheat sheet

A chip is worth less than its face value in dollars. The discount grows through mid-stage and peaks around the bubble.
Doubling up is worth less than busting is expensive. Every coinflip has a hidden cost that is invisible in cash-game math.
Play your risk premium. Tight on the bubble when covered, loose on the bubble when short, wide heads-up, measured in the middle.
Big stacks print money on the bubble. Abusing medium-stack ICM fear is the single highest-EV structural spot in tournament poker. Be the big stack. Abuse.
Bounties loosen ICM’s grip. Expected bounty value directly offsets risk premium, which widens calling ranges — especially for the player who covers the field.
An ICM chop is the fair chop. Chip-proportional deals systematically misprice stacks. Always run the numbers.

Run a tournament on Poker Timer this weekend and watch for the moments. Someone will fold a hand they shouldn’t. Someone will call a shove they shouldn’t. Point at the chips and say: “A chip is worth less than a chip.” They will laugh. Then they will lose.

Run your next tournament →