Free poker tool
ICM Poker Chart: Risk Premium by Situation
ICM is just the start
Multiway neural solves for tournaments — final table spots solved for your exact stacks, then drilled until they're instinct.
What a risk premium is
Chip EV asks whether a call wins chips. ICM asks whether it wins money. The two disagree because the payout ladder is not proportional to chips: you cannot win more than first prize, so the chips you stand to lose are worth more than the chips you stand to win.
The risk premium is the size of that disagreement, measured in pot equity. If a call breaks even in chips at 42% and the risk premium is 8 percentage points, the call breaks even in dollars at 50%. It is positive for anyone who can be eliminated, and it grows with the size of the next pay jump relative to the stacks in play.
Risk premium by stage
Illustrative ranges
| Stage | Extra equity needed | What it changes |
|---|---|---|
| Early levels, no payouts in sight | 0-2% | Essentially chip EV. Nothing to ladder into, so a flip for stacks is close to break-even in money. |
| Approaching the bubble, deep stacks | 2-5% | Marginal calls start to go. Opens and re-steals are unaffected; only the calls tighten. |
| The money bubble | 5-15% | The widest gap on the ladder relative to the chips. Medium stacks fold hands that are clearly ahead of the shoving range. |
| Just in the money, flat ladder | 2-5% | Pressure drops sharply the moment min-cash is locked. Short stacks correctly gamble again. |
| Final table, steep pay jumps | 5-20% | Depends heavily on stacks. A short stack next to two mid stacks faces the largest premium at the table. |
Illustrative only. The premium is a function of the exact stacks and the exact ladder, not of the stage label. Use these ranges to know roughly how hard to squeeze, then check the spot that is actually in front of you.
Run your own numbers
Enter the stacks and the ladder in front of you. The vs chip EV column is the premium in dollars, before you convert it into equity.
ICM is just the start
Multiway neural solves for tournaments — final table spots solved for your exact stacks, then drilled until they're instinct.
A three-player example, worked by hand
Three players remain. The prize pool is $1,000, paying $500 / $300 / $200 — a 50/30/20 ladder. The stacks are A 5,000, B 3,000 and C 2,000, for 10,000 chips in play.
Malmuth-Harville, the standard formulation, makes exactly one assumption: a player’s chance of finishing next among those still in is their share of the remaining chips. That gives first place straight away — 50%, 30% and 20%, the chip shares.
Second place needs conditioning on who won. Remove the winner, and recompute the shares:
- A wins (50%): B and C split 5,000 → B second 60%, C second 40%.
- B wins (30%): A and C split 7,000 → A second 5/7 = 71.4%, C second 2/7 = 28.6%.
- C wins (20%): A and B split 8,000 → A second 5/8 = 62.5%, B second 3/8 = 37.5%.
Weight those by how often each case happens. A finishes second 0.30 × 71.4% + 0.20 × 62.5% = 33.9% of the time. B finishes second 0.50 × 60% + 0.20 × 37.5% = 37.5%. C finishes second 0.50 × 40% + 0.30 × 28.6% = 28.6%. Third place is whatever is left over after first and second, and multiplying the three probabilities by the three prizes gives each seat in dollars.
Step 1 — the three seats in dollars
| Player | Stack | 1st | 2nd | 3rd | $ equity | Chip-EV share | Difference |
|---|---|---|---|---|---|---|---|
| Player A | 5,000 | 50.0% | 33.9% | 16.1% | $383.93 | $500.00 | -$116.07 |
| Player B | 3,000 | 30.0% | 37.5% | 32.5% | $327.50 | $300.00 | +$27.50 |
| Player C | 2,000 | 20.0% | 28.6% | 51.4% | $288.57 | $200.00 | +$88.57 |
The three equities sum to $1,000, and the three differences sum to zero. ICM never creates or destroys money — it moves it from the big stack to the short ones.
Player A holds half the chips and 38.4% of the money. Player C holds a fifth of the chips and 28.9% of the money. That $116.07 gap is the whole reason the big stack has to be careful.
Why a break-even flip loses money
A opens, B moves all in for 3,000, and the hand runs exactly 50/50 — the purest chip-EV-neutral spot there is. Price both outcomes.
- A wins. B busts and banks $200. Two players remain, 8,000 against 2,000, playing for $500 and $300. A’s equity is 0.80 × $500 + 0.20 × $300 = $460.00.
- A loses. Stacks become 2,000 / 6,000 / 2,000, and running the same three-player calculation gives A $295.00.
A’s expected value from the flip is 0.50 × $460 + 0.50 × $295 = $377.50. A was holding $383.93 before the cards went in. A coinflip A would be delighted to take in a cash game costs $6.43. For it to break even, A needs ($383.93 − $295) ÷ ($460 − $295) = 53.9% — a risk premium of 3.9 percentage points.
B’s side of the same flip is far worse, because B is the one who can bust. Winning leaves B with 6,000 of 10,000 and an equity of $410; losing means third place and $200. That is an expectation of $305 against the $327.50 B was already holding. B’s break-even point is ($327.50 − $200) ÷ ($410 − $200) = 60.7%, a risk premium of 10.7 percentage points.
Note how much the premium varies inside a single hand: 3.9 points for the player who cannot bust, 10.7 for the player who can. That is why a stage label alone can never give you a number, and why the ranges in the chart above are ranges.
When to stop doing this by hand
Three players and three payouts is six possible finishing orders, which is arithmetic you can do on a napkin. Four players is 24, six is 720, and a nine-handed final table is 362,880. The hand calculation is a teaching exercise, not a method.
Use the calculator whenever any of the following is true:
- More than three players are left.
- The stacks are not round numbers, which is always.
- The ladder is not a clean 50/30/20 — real structures rarely are.
- You want the premium for one specific pair of players rather than a feel for the table.
The chart is for the table, where you have thirty seconds and need to know whether to shade a marginal call. The ICM calculator is for afterwards, where you can put the real stacks in and find out what the spot was actually worth.
What the chart cannot tell you
ICM treats every player as equally skilled and ignores position and blind level entirely, so it undervalues a strong player’s stack and overvalues a weak one’s. A good player should take marginal spots slightly less often than the raw model says, because the edge they give up by busting is real — but the adjustment is small next to the effect the model is capturing.
The premium also cuts both ways. Everything above prices calling. Because your opponents face the same arithmetic, they fold more, so your fold equity rises at exactly the moment your calling range narrows. Correct bubble play is tighter and more aggressive at once, which is the part most players get half right.
For the model behind the numbers rather than the numbers themselves, the Independent Chip Model explained covers where Malmuth-Harville comes from and what it assumes.
Frequently asked questions
What is an ICM risk premium?
The extra pot equity you need before calling an all-in is profitable in money rather than in chips. If a call breaks even in chips at 42% equity and the risk premium is 8 percentage points, you need 50% for the call to break even in dollars.
Is there a single ICM chart I can memorise?
No. The risk premium is a function of the exact stacks and the exact payout ladder, not of the stage label, so any chart is a calibration aid rather than a lookup table. Two bubbles with the same number of players produce different premiums if the stack distributions differ.
How much extra equity do I need to call on the bubble?
Typically 5-15 percentage points over the chip-EV break-even point, and more when a short stack is about to be blinded out. On a satellite bubble, where every seat pays the same, the premium can be large enough that no call is correct at all.
How do you calculate ICM by hand?
Use Malmuth-Harville: a player's chance of finishing next is their share of the remaining chips. Set first-place odds equal to chip share, then condition on each winner to get second place, and subtract to get third. It is workable for three players and impractical past four.
Does the risk premium apply when I am shoving?
No — it works in your favour. The same pressure that makes calls expensive for your opponents makes them fold more often, so shoving ranges widen at the same time as calling ranges tighten. ICM punishes the caller and pays the aggressor.
Related tools
- Equity CalculatorHand vs hand, hand vs range, range vs range.
- Variance CalculatorIs this downswing normal? Usually, yes.