ICM stands for Independent Chip Model. Essentially, this is a mathematical model that calculates a player’s equity in a tournament given their chip stack and the other stacks and payouts. It doesn’t come into play until we are near or in the money, and has its most pronounced effects when there are large pay jumps looming, like at the final table of tournaments.
In cash games, a $1 chip equals $1. If there is a 10 handed table, everyone buys in for $100 and is not permitted to quit or rebuy, when someone has all the chips, that person will have $1000 in real money. In a tournament that’s not the case however. In line with the 10 handed cash game above, let’s change it to a 1 table tournament, a sit and go. Now the person that wins all the chips does not have $1000 in real money, they actually only have $500 in real money (typical sit and go payout structure is 50%/30%/20%). Everyone gets this. What this means is though is when we are faced with ICM decisions, the chips we can gain from a play aren’t worth as much as the chips we may stand to lose from that play. An easy way to see this is to use a contrived extreme example. Let’s say it’s the bubble of that sit and go mentioned above, 4 players remain and it pays $500, $300, and $200 respectively for the top 3. Let’s further say the stacks are like this:
Hero: Big Blind 7,000
Player 2: Small Blind 7,998
Player 3: 1
Player 4: 1
For simplicity I’m ignoring the blinds and antes, so players 3 and 4 are not anted all in and both fold. The Small blind then raises all in, and the decision is on you. This is a clear ICM fold, regardless of what hole cards you have. Even AA is a fold. Do you see why? If you play this hand and win, you’ll have all but locked up the SNG. You’re not guaranteed the $500 yet because the small blind will still have some chips to potentially rally from, but your equity in the prize pool is now probably around $475 or even greater. If you call and lose the hand, you win $0. This is a pure ICM disaster, because by simply folding, you’ve essentially got 2nd place money locked up, and further once it’s heads up in the next few hands (or less), you’ll be very close in chips stacks, so you’re equity in the prize pool is probably around $440 right now by folding. So you can see if you call and win with the aces, you’ve only increased your equity in the prize pool nominally, while if you call and lose, you’ve forfeited all equity and walked away with nothing. The risk is enormous for a reward that’s extremely nominal. Thus the correct play, the one that maximizes your earn in the long run, is to fold all hands until players 3 and 4 bust their 1 chip and it’s heads up.
The effect of ICM leads to some very counterintuitive shoves or folds late in games. I recently had the extreme example of this occur to me, where I folded AA before the flop live on stream while the chat went wild. Here was the situation: It was a satellite with 4 tickets to the target event, on the direct bubble with 5 left. The blinds were 2000/4000 with a 400 ante. The stacks to start the hand were:
BU (Hero): 14,773
UTG raises all in for 25,652
CO re-raises all in for 59,616
In this scenario, where the stack in 3rd place is at immediate risk of busting out, it is the correct ICM play for us to fold all hands, including AA. If you run this through an ICM calculator, you’ll see this to be true:
This is, quite simply, because my equity in this satellite prize pool (my chance to get one of the 4 tickets to the target event), goes up if I fold AA vs. if I call with it in this situation.
Other interesting notes from this ICM run… the cut off, who re-jammed, is actually only supposed to do this with AA/KK. Why so tight? Intuitively, if the cut off loses this confrontation, they will have 8.5 bigs left and be in 3rd, and should I and the small blind win our next all ins, they will be the shortest. Also, they can be at risk right now to the big blind who still hasn’t acted yet. But again, whether or not this sounds too tight, his re-shove with QQ slightly decreases his chances of winning a ticket vs. if he simply folded, another counter-intuitive result of ICM. By the way if we edit the re-shove range of the cut off to include QQ and AK, we still have a very clear fold with the AA vs. the 2 all ins.
Another interesting thing to look at here… when the UTG player jammed, I said on stream “well, I have to call him obviously” just before the cut off re-jammed. Is this correct, and if so how wide should I be calling? Yes, it’s correct, but I should be calling his shove very tight myself had the cut off folded rather than re-shoved:
So had the cutoff folded, my AA is a snap call, but KK is only marginally profitable and anything else loses me money/equity and reduces my chance to win a ticket (because I will be at risk right now). Had you asked me on stream how wide I would have called the UTG shove if the cut off folded, I would have said JJ+/AK. As the ICM calculation illustrated above shows, calling off with JJ would be lighting money on fire.
The moral of this story is, ICM pressure is a real thing, and it’s often better to be the one shoving rather than calling off in the face of ICM pressure. These effects are more pronounced the bigger the pay jumps, like occur at the final table of tournaments, or the direct bubble of a satellite. Getting your head around these concepts, at the very least in a basic sense, will help improve your decision making deep in tournaments or satellites, leading to more long term profit while your less savvy opponents make huge ICM mistakes and blow up their equity.