Independent Chip Model explained

basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
I have been reading about, and trying to understand, the independent chip model (ICM). However, all the online explanations I've come across so far only describe the intuition behind it; that you take the sum over money finishes multiplied by chance to win given your stack size. However, none of these explanations actually tell me how the chance to win given a stack size is calculated (rather, how it is calculated under the assumptions that ICM makes, clearly it depends on opponent types and your skill in reality). Does anyone have a good source on this?
 
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
It did, exactly what I was looking for! Thanks!
 
Michael Paler

Michael Paler

Legend
Silver Level
Joined
Apr 27, 2013
Total posts
1,203
Chips
0
Now, if you can figure out the value of the ICM to the average joe player.....
 
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
Are you guys implying that it's not useful in practice, or too hard to implement in practice? Because using it seems relatively straightforward to me, if you have PokerTracker or similar, you can calculate the ICM pot odds of your plays, and then use that along with equity to determine good play (good according to ICM, anyway).
 
K

kefir

Enthusiast
Silver Level
Joined
Sep 27, 2014
Total posts
84
Chips
0
Don't take every word on a poker forum for a good advice. Read some more and then decide who is telling something useful and who not. I agree with basse on ICM.
 
Everybodylovesdeuces

Everybodylovesdeuces

Rock Star
Silver Level
Joined
Sep 9, 2014
Total posts
148
Chips
0
I can't recommend highly enough the Kill Everyone book for ICM explanation. It's long and winded and took me 3 read throughs to understand but totally worth it.
[ame="http://www.amazon.com/Kill-Everyone-Strategies-Tournaments-Sit-n-Gos/dp/1935396307"]Kill Everyone: Advanced Strategies for No-Limit Hold 'Em Poker, Tournaments, and Sit-n-Gos: Revised and Expanded Edition: Lee Nelson, Tysen Streib, Steven Heston: 9781935396307: Amazon.com: Books[/ame]
 
W

WiZZiM

Legend
Silver Level
Joined
Oct 28, 2009
Total posts
5,008
Chips
0
Are you guys implying that it's not useful in practice, or too hard to implement in practice? Because using it seems relatively straightforward to me, if you have PokerTracker or similar, you can calculate the ICM pot odds of your plays, and then use that along with equity to determine good play (good according to ICM, anyway).

No i think it's important. However it really depends on your end goal, if you are looking to play super high stakes, then sure, nailing down ICM and NASH play is going to be a must. However if you plan to just get to mid-stakes, nailing down these things is not as important. It might be more important to play a less perfect game, simplify decisions and play more volume as the low to mid-stakes are relatively easy to beat.

Basically, ICM is great, but learning it is extremely time consuming and takes many hours of practise, not only on the tables but away from the tables too. I know this as i spent countless hours in SNG wizard studying ranges and trying to learn these things. However looking back, i could have done it in a much simpler way that is way less time consuming. If you would like me to elaborate on this let me know.

If there is an area where ICM is more important, it's the bubble. But again, keep it simple and study the stack sizes that come up most often.

For 6 max sng these stack sizes might be:-

. Big stack, two small stacks.
. Two medium stacks, one small stacks.
. equal stacks.

etc etc. What this does is quickly gives you a grasp of how the stack sizes on the bubble affect things, instead of just blindly reviewing games or looking at particular shoves you might have made. Model a situation for each of the stack sizes you are likely to encounter in game, then muck around with it from all angles to see how things change.

Sorry bit rushed, i can expand on this more if you like.
 
B

bnasp2

Visionary
Silver Level
Joined
Mar 17, 2013
Total posts
606
Chips
0
Well, ICM is very important for S&G tournaments and for final tables of MTTs. It tells you how much you are supposed to risk.
Because at those final tables, many decisions which are +EV in chips, are actually very -EV in terms of money.

There are couple of ICM calculators on the web. And good way to learn is to take some of your past decisions and try to check if they were good. Take your history (some tracker software helps a lot) find where you were thinking if to push or fold and check in calculator.
 
Jacki Burkhart

Jacki Burkhart

long winded rambler...
Silver Level
Joined
Oct 12, 2013
Total posts
2,960
Awards
6
Chips
0
I often hear generalized statements like "ICM only really matters on the bubble, in sit n gos or on the final table". But I have to disagree. ICM is a factor in every single tournament situation (except for heads up play, obviously). Yes, it is definitely a much bigger factor on the bubble of a SnG than it is 15 minutes into the wsop main event...but it's still a factor.

I think the most important thing to remember about ICM is that chips you already have are worth MORE than chips you stand to win...so basically a measure of caution is prudent when looking at equity calculations in a tournament. So if a certain action is slightly +EV, then in a cash game you should basically always take that action, but in a tournament slight +EV might be a pass due to ICM.
 
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
I can't recommend highly enough the Kill Everyone book for ICM explanation. It's long and winded and took me 3 read throughs to understand but totally worth it.
Kill Everyone: Advanced Strategies for No-Limit Hold 'Em Poker, Tournaments, and Sit-n-Gos: Revised and Expanded Edition: Lee Nelson, Tysen Streib, Steven Heston: 9781935396307: Amazon.com: Books

Actually I got it from the link initially given. Thanks for the tip though, I might check this book out anyway. It seems to have a strong math focus, which is something that I enjoy. I am a little suspicious of its "optimal play" claims, since it's not clear what optimal play means in n>2 player games.
 
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
I often hear generalized statements like "ICM only really matters on the bubble, in sit n gos or on the final table". But I have to disagree. ICM is a factor in every single tournament situation (except for heads up play, obviously). Yes, it is definitely a much bigger factor on the bubble of a SnG than it is 15 minutes into the WSOP main event...but it's still a factor.

I think the most important thing to remember about ICM is that chips you already have are worth MORE than chips you stand to win...so basically a measure of caution is prudent when looking at equity calculations in a tournament. So if a certain action is slightly +EV, then in a cash game you should basically always take that action, but in a tournament slight +EV might be a pass due to ICM.

I wholly agree. It's not clear to me why it would only be relevant at the final table or on the bubble. Even if there are thousands of players left, one should still think about $EV, as opposed to chip EV.

So when people say that it only matters at top table/on the bubble etc. what they're really saying, I guess, is that the assumption of ICM does not hold earlier on in big tournaments. But what model of action EV are people then using instead? I think ICM is more likely to be correct early on than chip EV, for example.
 
Jacki Burkhart

Jacki Burkhart

long winded rambler...
Silver Level
Joined
Oct 12, 2013
Total posts
2,960
Awards
6
Chips
0
I wholly agree. It's not clear to me why it would only be relevant at the final table or on the bubble. Even if there are thousands of players left, one should still think about $EV, as opposed to chip EV.

So when people say that it only matters at top table/on the bubble etc. what they're really saying, I guess, is that the assumption of ICM does not hold earlier on in big tournaments. But what model of action EV are people then using instead? I think ICM is more likely to be correct early on than chip EV, for example.

I think what they mean is that when many tables remain it's so difficult to calculate and use table side that it isn't worth your time to calculate out each stack's equity...and with that I'd agree. But I think it is still worthwhile to understand the general principles of ICM at all tourney stages and how that affects your $EV and how that relates to your cEV.

So, my guess would be that very early in a MTT the cEV and the $EV would be closer than at any other time during the tourney (prior to heads up play). But they still will not be equal. The difference might even be negligible at certain times...but there will always be a difference.
 
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
No i think it's important. However it really depends on your end goal, if you are looking to play super high stakes, then sure, nailing down ICM and NASH play is going to be a must. However if you plan to just get to mid-stakes, nailing down these things is not as important. It might be more important to play a less perfect game, simplify decisions and play more volume as the low to mid-stakes are relatively easy to beat.

Basically, ICM is great, but learning it is extremely time consuming and takes many hours of practise, not only on the tables but away from the tables too. I know this as i spent countless hours in SNG wizard studying ranges and trying to learn these things. However looking back, i could have done it in a much simpler way that is way less time consuming. If you would like me to elaborate on this let me know.

If there is an area where ICM is more important, it's the bubble. But again, keep it simple and study the stack sizes that come up most often.

For 6 max sng these stack sizes might be:-

. Big stack, two small stacks.
. Two medium stacks, one small stacks.
. equal stacks.

etc etc. What this does is quickly gives you a grasp of how the stack sizes on the bubble affect things, instead of just blindly reviewing games or looking at particular shoves you might have made. Model a situation for each of the stack sizes you are likely to encounter in game, then muck around with it from all angles to see how things change.

Sorry bit rushed, i can expand on this more if you like.

Thanks for the long explanation! I see what you're saying about simplified strategies being easier to play at high volume. At that point, I guess it's also a question of what my goal is, though. I think I would rather play really well at low volume, than make more money at higher volume. At least initially. I can always move to simpler strategies and higher volume later if I play well.

When you say that learning ICM takes a long time, do you mean calculating the ICM odds yourself during play, or learning to understand what ICM does?
If the former, can't you just use an ICM calculator? I do not intend to actually sit and practice mental ICM calculation. I think the important part is understanding what the model does, and its limitations. After that, I'm more than happy to let an online calculator do it for me.

Unless you're referring to live play, in which case I guess you'd have to learn heuristic ways to do it fast in your head.
 
W

WiZZiM

Legend
Silver Level
Joined
Oct 28, 2009
Total posts
5,008
Chips
0
Thanks for the long explanation! I see what you're saying about simplified strategies being easier to play at high volume. At that point, I guess it's also a question of what my goal is, though. I think I would rather play really well at low volume, than make more money at higher volume. At least initially. I can always move to simpler strategies and higher volume later if I play well.Yep that's a fine way to go imo :)

When you say that learning ICM takes a long time, do you mean calculating the ICM odds yourself during play, or learning to understand what ICM does?Learning to understand what it does is relatively easy, i think you are most of the way there already since you seem quite intelligent. What can take a long time is countless review sessions and inefficient training methods. Like, reviewing your past play can be beneficial, but not for learning basic ICM ranges.
If the former, can't you just use an ICM calculator? I do not intend to actually sit and practice mental ICM calculation. I think the import an`t part is understanding what the model does, and its limitations. After that, I'm more than happy to let an online calculator do it for me.Yes for sure! it's incredibly important to know the ins and outs and limitations of ICM. It's also important to know how to counter some of these limitations. The main thing I am talking about is inefficient use of the ICM calculators, which i'll cover in a little bit.

Unless you're referring to live play, in which case I guess you'd have to learn heuristic ways to do it fast in your head.

So just to make this clear, i am talking about the use of ICM calculators and inefficient study methods or learning methods.

Most people when they use these calculators, will just review a few tournaments or trouble spots, go over the hands, "good shove, bad shove" etc.

this is not a very good way to study efficiently. This was also the way i went about it, but knowing what i know now, i would do things differently. I'll outline a few ways i would do things now if i were to start learning ICM ranges.

1. Build a SNG shove/fold chart. This is really handy for learning the basic ranges needed. The most important part is that you develop it for your games, as every level has a slightly different population tendency. I'll try and link my own shove/fold chart below, but i'm not on my own computer.

2. Setup a study plan. To do this, i would start at the bubble, as that is where ICM is most extreme. This of course depends on the games you play, however i'd try and keep it as simple as possible. The things i would look at are differing stack sizes, and how the ranges change when you are in differing stack setups. So for example, i play 6 max SNG so i would look at the following to study. These stack setups are just the most common, but the key is that it gives you a great starting ground, then once the stack setup changes slightly, you kind of get a feel for how much you need to adjust your range. It also gives you insight into how important it is to have the big stack and how you can use a slight chip advantage to totally dominate good players who abide strictly to ICM theory.

Equal stacks
Big stack, two small stacks
Two equal stacks one small stack
one big stack, one medium stack, one small stack

If you are playing 9 max sng, this will obviously differ. So when you study a particular stack setup, it's important to switch yourself between these stacks to see what you can shove at differing blind levels. This will take some time obviously, but you will learn far quicker studying these basic setups.


3. Once you have mastered the stack setups above and understand how they relate, you can study spots that confuse you. It could be 3bet shoving. To study this spot effectively, first go into your tracker, filter for these specific situations for certain blind levels. This will give you a bunch of hands, run these hands through an ICM calculator and study 3-5 of them in a row. Once you have a grasp for that spot, you may want to tinker with the blind levels or opponent ranges to see what you can get away with shoving profitably.


That's it really. Those are the 3 things i would focus on in that order to nail down basic ranges that you can use in your head on the fly. No matter what the end goal is to be able to use ranges effectively in game, and not be confused in game or think it's the end of the world if you shove a little bit loose in a particular hand, in reality it doesn't really matter too much.
 
Last edited:
basse

basse

Enthusiast
Silver Level
Joined
Sep 24, 2014
Total posts
68
Chips
0
So just to make this clear, i am talking about the use of ICM calculators and inefficient study methods or learning methods.

Most people when they use these calculators, will just review a few tournaments or trouble spots, go over the hands, "good shove, bad shove" etc.

this is not a very good way to study efficiently. This was also the way i went about it, but knowing what i know now, i would do things differently. I'll outline a few ways i would do things now if i were to start learning ICM ranges.

1. Build a SNG shove/fold chart. This is really handy for learning the basic ranges needed. The most important part is that you develop it for your games, as every level has a slightly different population tendency. I'll try and link my own shove/fold chart below, but i'm not on my own computer.

2. Setup a study plan. To do this, i would start at the bubble, as that is where ICM is most extreme. This of course depends on the games you play, however i'd try and keep it as simple as possible. The things i would look at are differing stack sizes, and how the ranges change when you are in differing stack setups. So for example, i play 6 max SNG so i would look at the following to study. These stack setups are just the most common, but the key is that it gives you a great starting ground, then once the stack setup changes slightly, you kind of get a feel for how much you need to adjust your range. It also gives you insight into how important it is to have the big stack and how you can use a slight chip advantage to totally dominate good players who abide strictly to ICM theory.

Equal stacks
Big stack, two small stacks
Two equal stacks one small stack
one big stack, one medium stack, one small stack

If you are playing 9 max sng, this will obviously differ. So when you study a particular stack setup, it's important to switch yourself between these stacks to see what you can shove at differing blind levels. This will take some time obviously, but you will learn far quicker studying these basic setups.


3. Once you have mastered the stack setups above and understand how they relate, you can study spots that confuse you. It could be 3bet shoving. To study this spot effectively, first go into your tracker, filter for these specific situations for certain blind levels. This will give you a bunch of hands, run these hands through an ICM calculator and study 3-5 of them in a row. Once you have a grasp for that spot, you may want to tinker with the blind levels or opponent ranges to see what you can get away with shoving profitably.


That's it really. Those are the 3 things i would focus on in that order to nail down basic ranges that you can use in your head on the fly. No matter what the end goal is to be able to use ranges effectively in game, and not be confused in game or think it's the end of the world if you shove a little bit loose in a particular hand, in reality it doesn't really matter too much.

Thanks for all the advice! I like your idea of building a fold/shove chart, I had been thinking that I should come up with something to write up as I learn, to make me remember it better. This is a perfect solution for that. I think I would like to make it applicable to MTTs, since I will be playing those initially. In that case, I should probably have different fold/shove charts for different stages in the game, right? Also, the shove/fold charts are usually for preflop play, right? I think my biggest weakness is going to be post-flop, as it's not nearly as easy to quickly learn heuristics for how to play there.

The differing stack sizes is also a good idea, as the combinatorial explosion in possible stack sizes and hands is quite vast, obviously, and thus hard to study more specifically.

Your third point will probably take me a while to get to :)
 
W

WiZZiM

Legend
Silver Level
Joined
Oct 28, 2009
Total posts
5,008
Chips
0
Yes it's a slow process, but lucky for you, you can learn from my past mistakes if you choose too :) Making it a much quicker process.
 
Top