An introduction to ICM

by MathrimC

In this article, I try to explain what ICM is and some of its implications for people unfamiliar
with the concept. Next to an introduction to the maths of ICM, I try to provide some practical
guidelines and link ICM implications to common poker intuitions. In this way, even if you
don’t understand or don’t care for the maths explained, you can still pick up some valuable
advice that might help you at the poker tables. In doing so, this article might oversimplify
some things or leave out some details. If you’re well familiar with the concept of ICM, this
article probably isn’t for you, but feel free to send any comments, suggestions, corrections or
questions to ​[email protected]​.

What does “ICM” stand for?

ICM stands for “Independent Chip Model”. It’s the name for the theory that states that in a
poker tournament, in contrast to a cash game, the monetary value of each chip is not fixed,
but dependent on the circumstances. These circumstances are the payout structure and the
chances you make it to each of those payout spots with your current chipstack.

How is the monetary value of the chips being calculated

The monetary value of a chipstack is calculated as follows: the chance that you finish 1st
multiplied by the payout for 1st place + the chance you finish 2nd multiplied by the payout for
2nd place + … Intuitively, you can see that this is the amount of money you would win on
average if you would continue to play the tournament an infinite amount of times from the
current situation. This concept is called “monetary expected value” or “$EV”. Or to put it
differently, it’s the amount of money your stack would be worth if you would stop playing right
now and sell your stack to someone else.

Example 1: there are 4 players left in a qualifier where the top 2 win €100. Each
player has exactly 100 chips. What is the monetary value of these chipstacks? Given
that everyone has the same stack, everyone has a 25% chance to finish 1st (under
the hypothesis they are all equally skilled players) and a 25% chance to finish 2nd.
The $EV each stack is €100x25% + €100x25% = €50.

Example 2: a little later in that same qualifier where the top two win €100, there are 3
players left: 1 player with 200 chips and 2 players with 100 chips each. The monetary
value of the stack with 200 chips is now €83.33. The monetary value of each of the
stacks with 100 chips is €58.33. See the attachment at the end of the article for an
example of how this is calculated in detail.
Notice that in example 2, the player that doubled his chipstack, didn’t double his $EV as
opposed to example 1. In other words: the 100 chips they won aren’t worth as much as the
100 chips they already had. You can grasp this intuitively without doing the maths: after
winning those 100 extra chips, the player still isn’t guaranteed to finish 1st or 2nd, and so not
yet guaranteed to win €100, so it would be impossible for their stack to be worth €100. Also
notice that the two players who have the same stack in example 2 as they had in example 1,
all of a sudden have a higher $EV. This is a confirmation of what most players also know
intuitively: if the bubble comes near, it’s good for you if other players bust, even if you don’t
win any chips yourself.

What’s the impact of ICM on how you should play?

Let’s take a closer look at the situation of the player who doubled. To keep it simple for now,
let’s leave blinds and antes out of consideration. By going all-in, the player risked 100 chips
to win 100 chips, but in $EV they risked €50 to win €33.33, or to get a total stack worth
€83.33. Given the $EV represents the real monetary value, they need a hand with 50/83,33
= 60% equity to be break-even in the long run in that all-in situation. This is different than in a
cashgame, where only 50% equity would be needed. In other words: you need a stronger
hand to call an all-in. This effect becomes stronger the closer you are to the bubble and if
there are other players with fewer chips not involved in the hand.

Example 3: in that same qualifier where the top 2 cashes €100, the player with 200
chips goes all-in. The first player with 100 chips folds, how much equity does the
second player with 100 chips need to call? We already calculated that a 100 chip
stack is worth €58.33 in this situation, so this is what they risk by calling. If they win,
they will become the player with 200 chips against two players with 100 chips, so the
player’s $EV will become €83.33. So they risk €58.33 to potentially end up with a
stack worth €83,33. They need $58.33/$83.33 = 70% equity to make a break-even
call!

For reference: AKs has 67% equity against a 100% range, 88 has 69% equity in that case.
So in example 3, only 99+ is a profitable call against a bigstack who goes all-in with 100% of
hands. If the bigstack goes all-in with only 50% of his hands, only JJ+ is a profitable call.
This situation where you can only profitably call with very few hands is called ​ICM-pressure​.

Conversely, this results in a situation where the bigstack can go all-in with a very wide range,
much wider that the usual push/fold range, because they’ll almost always successfully steal
the blinds. Here it is important to note that this only applies if the other players are
“ICM-aware”​ and call tight enough. If the other players are too loose and call too many all-ins
(a mistake that is known as “​ICM-suicide”)​ , it might become unprofitable for the bigstack to
shove very wide, because the bigstack also risks more $EV by losing an all-in than they can
gain by winning one. However, the effect on the bigstack is smaller because they still have
chips left if they loses the all-in. In this case, ​if​ the shortstack calls, the bigstack needs 60%
equity to break-even. This is already a lot less than the 70% that the shortstack needs to
make a break-even call.
Which factors influence ICM-adjusted ranges

Payout structure
The qualifier example above is quite extreme because of the flat payouts. In general, the rule
is the flatter the payouts, the bigger the ICM-pressure. Let’s have another look at example 3,
with the same players, the same stacks, the same total prizepool, but a different payout
structure:

Example 4: in a SNG with 3 players left where 1st place wins €133 and 2nd place
wins €67, a player with 200 chips goes all-in. A second player with 100 chips folds.
Which equity does a third player with 100 chips need to call? The current $EV of the
player with 100 chips is €55,58. If they win the all-in, their $EV becomes €88,83. So
they risk €55,58 to potentially end up with a stack worth €88,83. For this they need
55,58/88,63 = 62,5% equity.

62,5% is less than the 70% that was needed in a similar situation with a flat payout structure.
This too is relatively easy to grasp intuitively: given that, after an elimination, the remaining
two players will still have to continue playing for first place, which pays double compared to
second place, it’s an advantage to begin that heads-up battle with a bigger stack. Therefore,
taking risks to win some extra chips on a SNG bubble has more advantages than in a flat
payout structure.

Note that the absolute amounts of the payouts change nothing to the ICM-pressure, it’s the
relative percentages of the payouts compared to one another that determine the
ICM-pressure. For example, if we change the payouts in example 4 to €30 and €15, the
equity calculation will give the exact same result.

Shortstacks
A factor that increases the ICM-pressure is the presence of shortstacks that are not in the
hand. Let’s take another look at example 4, with the only difference that the bigstack has an
additional 50 chips while one of the shortstacks has 50 chips less.

Example 5: in a SNG with 3 players left where 1st place wins €133 and 2nd place
wins €67, a player with 250 chips goes all-in. A second player with 50 chips folds.
Which equity does a third player with 100 chips need to call? The current $EV of the
player with 100 chips is €63,56. If they win the all-in, their $EV becomes €91,39.
They need 63,56/69,55 = 69,55% equity to make a break-even call.

69,55% is 7% more than in example 4 where the other shortstack had 100 chips. This too is
easy to grasp intuitively: if a player is about to be eliminated on the bubble, you need to take
less risks yourself and wait for the player to bust.

Note that the impact of the shortstack becomes very different if he’s the one going all-in:
 Example 6: in a SNG with 3 players left where 1st place wins €133 and 2nd place
wins €67, a player with 50 chips goes all-in. A second player with 250 chips folds.
Which equity does a third player with 100 chips need to call? The current $EV of the
player with 100 chips is €63,56. If they win the all-in, their $EV becomes €91,75. But
especially: if they lose, their $EV doesn’t become 0. If they lose, they still have 50
chips, good for €33,38 in $EV. So they only risk €63,56 - €33,38 = €30,18 by calling
an all-in, and need only 30,18/(91,78-33,38) = 51,71% equity to make a break-even
call.

The 51,7% equity that’s needed here is a lot less than in our prior examples. This is easy to
understand intuitively as well, because you’re not risking your entire stack. But even if we
return to the situation with 1 player with 200 chips and 2 players with 100 chips, and a player
with 100 chips goes all-in, we’ll see that the ICM-pressure on the other player with 100 chips
is still reduced as compared to when the player with 200 chips shoves:

Example 7: in a SNG with 3 players left where 1st place wins €133 and 2nd place
wins €67, a player with 100 chips goes all-in. A second player with 200 chips folds.
Which equity does a third player with 100 chips need to call? The current $EV of the
player with 100 chips is €55,58. If they win the all-in, their $EV becomes €100. So
they risk €55,58 to potentially end up with a stack worth €100. For this they need
55,58/100 = 55,58% equity.

This 55,58% equity is 7% less that they needed in example 4, the only difference being that
it was the bigstack going all-in in example 4. This can be graped intuitively as well: the
situation that arises in this example after you win the all-in is more advantageous as
compared to example 4: here you’re guaranteed a 2nd place.

In general, we assume the following rules:

● a shortstack that isn’t involved in the hand increases the ICM-pressure. The less
chips the shortstack has compared to you, the bigger the pressure becomes and the
more equity you need to profitably call an all-in. Conversely, the more chips the
players have that aren’t in the hand, the lower the ICM-pressure, and the less equity
you need to call an all-in.
● the less chips the all-in player has, the lower the ICM-pressure, so the less equity you
need to make a profitable call. Conversely, the more chips the all-in player has, the
bigger the ICM-pressure, and the more equity you need to call the all-in.

Pot-odds
The more chips are in the pot already in comparison to the amount of chips you need to call,
the looser you can call. Let’s look at example 4 again, but this time with blinds and antes.
Let’s also adapt the stacksizes a bit so that after paying the blinds and antes, the players
have the same amount of chips as in example 4, so that the additional chips in the pot are
the only difference in the calculation.
 Example 8: in a SNG with 3 players left where 1st place wins €133 and 2nd place
wins €67, the blinds are 10/20 with a 3 chip ante. After paying the blinds and antes, a
player with 200 chips goes all-in. A second player with 100 chips folds. Which equity
does a third player with 100 chips need to call? If they fold, the $EV of the player with
100 chips will be €53,04. If they call and win the all-in, their $EV becomes €98,94. So
they risk €53,04 to potentially end up with a stack worth €98,94. For this they need
53,04/98,94 = 53,6% equity.

Because of the additional blinds and antes in the pot, we need 9% less equity to make a
profitable call as compared to example 4. Note that this is a general rule in any situation,
regardless of ICM.

How to translate this to real ranges

All factors mentioned above and their complex interactions make it impossible to compose
fixed ranges. Also, calculating the needed equity percentages is too complex to do while
playing. The best thing you can do is practice to get a good intuition of how these factors
influence your ranges, and save screenshots or hand histories of difficult decisions to verify
them in an ICM-calculator. This way, you’ll get better in correctly estimating your ranges
while playing.

Famous examples ICM-calculators are ICMizer and Holdem Resources Calculator. On

icmizer.com, you can make a free account which allows you to make three free ICM
calculations each day. You enter all the relevant information about the situation (payout
structure, stacksizes, bets) and the calculator makes the same calculations as we did in the
examples above. It can handle much more complex situations which would be nearly
impossible to do by hand. The application also goes one step further: based on the equity
you need to call, it also calculates the nash-equilibrium ranges of hands with which you can
call and your opponent can go all-in.

An important note here is that it calculates an equilibrium: it calculates a situation in which

both the player shoving and the player calling use a range that is optimally tailored towards
each other. In a real poker situation, this is rarely the case. When one of both players
deviates from his optimal range, the other player needs to adjust their range as well if they
want to keep playing optimal. For example, if the player shoving is more selective and
shoves a stronger range, then the player calling also needs a better hand to make a
profitable call. Conversely, if a player calls too loose, the player that goes all-in needs to
shove with a stronger range than usual, because they’ll be able to steal the blinds less often
and so they’ll more often need the equity of their hand to win the pot.

Despite all this, the calculated equilibrium ranges are still a very good starting point. As
we’ve seen above in the explanations of the influencing factors, many of these factors are in
line with common player intuitions. Many recreational players will already intuitively tighten
up in places where ICM requires this, although often not enough. It’s important to pay
attention near the bubble of a tournament to see which players are ICM-aware and which
aren’t, and to adapt your ranges based on that knowledge.

If you want to do ICM calculations without ICMizer, you can use ​this​ spreadsheet. You’ll
have to make a copy to your own google drive to use it (click “File” -> “Make a copy…”). The
sheet works with a script, which you might have to give access permissions to run. If you
don’t wish to do that, ​here​’s a more limited version where everything is done in formulas.
However, this last sheet can only handle 2 payouts and 5 players.

As you’ll see below, using the sheet is a little more work than calculating ranges using
ICMizer, but because it shows you some of the numbers behind the calculation, it will give
you more practice with how certain situations impact equity and get you more familiar with
the equity of certain hands against certain ranges.

Here’s an example how to use the sheet to calculate a calling range.

Put the payouts in cells A2 to A8, your own stacksize (after blinds and antes) in cell C2, a
potential bet or blind that’s already in front of you in cell D2, the bet of the all-in opponent in
cell D3 and the chipstacks of the other players at the table in cells C4 to C11. The all-in
opponent has to be on the third row of the sheet, regardless of his position at the table. The
order of the other players that aren’t in the hand doesn’t matter, as long as they are listed.
Possible chips that are already in the middle should be put in cell E4.

If you click on “Calculate equity”, the $EV of your stack will be calculated after you fold
(€51,21 in the example above), after you call and win (€69,68 in the example above) and
after you call and lose (€34,31 in the example above). Based on those three values, the
sheet calculates the equity your hand needs against the opponents range to make calling
more profitable than folding (47,78% in the example above).

Using Equilab (a free application available ​here​), we can convert this percentage into a
range of hands. To do this, we first need to estimate the range of hands with which our
opponent goes all-in. A good default for this, is the nash-range for an opponent in this
situation if there were no ICM-pressure (referred to as the “standard nash range” below).
The opponent shoves for 8bb from the small blind. This is the standard nash range for that
situation:
If you enter this range in Equilab and go to “Tools” -> “Hand range calculator”, you can
calculate the range of hands that have at least 47,78% equity against this shoving range.
An important attention point in this calculation is the range you assume for the all-in player.
The example above assumes a standard nash-range which doesn’t take ICM impact into
account. This means that in reality, the optimal shoving range for the player can differ from
this. In the above example, the optimal shoving range of the opponent will be smaller than
the standard nash-range because the other shortstack is not in the hand. If the bigstack calls
their all-in, they need a lot more than 50% equity to break even. If we enter the example
above in ICMizer, we indeed see a slightly smaller optimal all-in range:
This results in the following optimal calling range:
Also, it will often happen that recreational players that are unfamiliar with these ranges use
different ranges. For example, when the optimal all-in range is very big, like in the example
above, recreational players often won’t shove many of the weaker suited hands in the range.
For example, suppose in the example above a recreational player goes all-in with the
following range instead:

Then this will become our optimal calling range, the range of hands that have at least
47,78% equity:
Now, before you feel overwhelmed or confused by all these different scenarios, remember
that in many cases the standard nash shoving ranges will be a very good default hypothesis
to calculate calling ranges. Many players use these ranges, and even you suspect your
opponents to use different ranges, it’s useful to know what to call against the standard
nash-range and use that as a basis to derive from depending on the specific reads you have.
Also, as you can notice in the example above, the optimal calling range doesn’t change that
much for the different shoving ranges. Remember that you don’t need to make perfect
decisions to be a profitable player, all that matters is that your decisions are better than
those of your opponents. With practice, your ICM intuitions and estimations will become
better than those of the average player at your tables.
Annex: detailed ICM calculation
The ICM value of a stack is calculated as follows:
The chance of finishing 1st multiplied by the 1st place payout
+ the chance of finishing 2nd multiplied by the 2nd place payout
+ … (for every payout spot)

The chance for a player to finish 1st is calculated by dividing his amount of chips by the total
amount of chips in play.

The chance for a player to finish 2nd is calculated by the sum of the chances for every other
player to win multiplied by the chance for the player to finish 2nd. The chance for player A to
finish 2nd if player B wins is calculated by dividing the chipstack of player A by the total
amount of chips in play substraced with the chipstack of player B. This has to be done for
every possible winner, and the sum of all the chances gives you the chance player A
finished second.

Let’s apply this to example 2 above: player 1 has 200 chips while players 2 and 3 have 100
chips each. There are 2 payouts of €100 each. The ICM-value of player 1 can be calculated
as follows:

(The chance player 1 wins) x (1st place payout)

+ ( (the chance player 2 wins) x (the chance player 1 finishes 2nd if player 2 wins)
+ (the chance player 3 wins) x (the chance player 1 finishes 2nd if player 3 wins) )
x (2nd place payout)

That’s:
(chipstack player 1) / (total amount of chips in play) x (1st place payout)
+ ( (chipstack player 2) / (total amount of chips) x (chipstack player 1) / (total amount of
chips - chipstack player 2)
+ (chipstack player 2) / (total amount of chips) x (chipstack player 1) / (total amount of
chips - chipstack player 3) )
x (2nd place payout)

And if we fill in the numbers:

200 / 400 x €100
+ ( 100 / 400 x 200 / (400 - 100)
+ 100 / 400 x 200 / (400 - 100) )
x €100

That’s €50 + €33,33 = €83,33.