Where ICM is lacking?

[taipan168]

[ QUOTE ]
[ QUOTE ]
Slim Pickens wrote a really good post on the shortcomings of ICM:

http://forumserver.twoplustwo.com/sh...Number=9305406

[/ QUOTE ]

Thank you. And he has, indeed, brought out more shortcoming of the model. I think ill try last night's post again tonight since everyone seems fairly fired up about all this.

[/ QUOTE ]
The fact that ICM has shortcomings doesn't mean that it is wrong. It's still the best thing that we have to pwn SNGs. The key is to be able to make adjustments to ICM when we know the shortcomings are significant (which is very rarely IMO).

[IFoldPktOnes]

Interesting discussion, thanks for clarifying a few things for me pineapple.

One concept I have trouble applying is when to take slight -EV pushes, you alluded to this above:
[ QUOTE ]

Part of the skill of STTs is understanding how important each of those five factors become in various situations, to set your minimum edge correctly. This min edge can range from quite large to potentially slightly negative (on very tough tables where the sizable blinds are about to hit you).


[/ QUOTE ]
[ QUOTE ]

ICM undervalues big stacks slightly (steal equity) and overvalues micro-stacks slightly (loss of fold equity).

[/ QUOTE ]
So we should make slightly negative EV pushes to prevent losing our fold equity when large blinds are about to cripple our stack? An example of when this is correct would really help me out. Sometimes I find myself doing this intuitively and later look back at my HH and wonder if it was the right play.


Btw ChipLeader look closely at how you worded your example. If you meant to say the 1st toss is a "fair" bet and you have $2 after winning, then the whole example is pointless since a fair bet doesn't change expectation it only increases variance (this might be what you are getting at im not sure).

[pineapple888]

[ QUOTE ]
Interesting discussion, thanks for clarifying a few things for me pineapple.

One concept I have trouble applying is when to take slight -EV pushes, you alluded to this above:
[ QUOTE ]

Part of the skill of STTs is understanding how important each of those five factors become in various situations, to set your minimum edge correctly. This min edge can range from quite large to potentially slightly negative (on very tough tables where the sizable blinds are about to hit you).


[/ QUOTE ]
[ QUOTE ]

ICM undervalues big stacks slightly (steal equity) and overvalues micro-stacks slightly (loss of fold equity).

[/ QUOTE ]
So we should make slightly negative EV pushes to prevent losing our fold equity when large blinds are about to cripple our stack? An example of when this is correct would really help me out. Sometimes I find myself doing this intuitively and later look back at my HH and wonder if it was the right play.


[/ QUOTE ]

And everyone else at the table is clueful so they aren't about to bust each other out. It really doesn't come up all that often. I could construct an example I suppose but don't have one at my fingertips.

[fluorescenthippo]

i didnt read any posts but B is wrong. icm should factor in other players actions. a program like sngpt wont factor in other variables once utg+1 calls utgs push. but icm does.


something else i didnt see in the OP was the value of a shortstack. i thought this was the most obvious misunderstanding of ICM. ICM overvalues short stacks, esp on the bubble.

[tshort]

[ QUOTE ]
i didnt read any posts but B is wrong. icm should factor in other players actions. a program like sngpt wont factor in other variables once utg+1 calls utgs push. but icm does.

[/ QUOTE ]

Semantics using the term "ICM" are throwing me off in so many of these posts.

ICM - "Independent Chip Model"

A model / algorithm for determining real dollar equity of chip stacks in a tournament based off a the prize structure.


Now, you say "icm should factor in other players actions"?

Are you referring to a program (a program which happens to use the ICM to assign real dollar equity to chip stacks)? I can see how a program should factor in other players actions.

[fluorescenthippo]

[ QUOTE ]
[ QUOTE ]
i didnt read any posts but B is wrong. icm should factor in other players actions. a program like sngpt wont factor in other variables once utg+1 calls utgs push. but icm does.

[/ QUOTE ]

Semantics using the term "ICM" are throwing me off in so many of these posts.

ICM - "Independent Chip Model"

A model / algorithm for determining real dollar equity of chip stacks in a tournament based off a the prize structure.


Now, you say "icm should factor in other players actions"?

Are you referring to a program (a program which happens to use the ICM to assign real dollar equity to chip stacks)? I can see how a program should factor in other players actions.

[/ QUOTE ]

first im not really sure what the OP is saying. B and C are the same to me. B sounds like you dont want to go all in even if its +EV which makes no sense. i assumed he meant: that what seemed +EV might not be because of other players left to act.

in this sense ICM is still correct. you just need to analyze the possible pushing/calling ranges after you and see how that affects your EV.

[K䲰䮥n]

ICM and this whole forum was just a big hoax to get me donating my money to Dikshit's bots.

[ChipLeader]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
i didnt read any posts but B is wrong. icm should factor in other players actions. a program like sngpt wont factor in other variables once utg+1 calls utgs push. but icm does.

[/ QUOTE ]

Semantics using the term "ICM" are throwing me off in so many of these posts.

ICM - "Independent Chip Model"

A model / algorithm for determining real dollar equity of chip stacks in a tournament based off a the prize structure.


Now, you say "icm should factor in other players actions"?

Are you referring to a program (a program which happens to use the ICM to assign real dollar equity to chip stacks)? I can see how a program should factor in other players actions.

[/ QUOTE ]

first im not really sure what the OP is saying. B and C are the same to me. B sounds like you dont want to go all in even if its +EV which makes no sense. i assumed he meant: that what seemed +EV might not be because of other players left to act.

in this sense ICM is still correct. you just need to analyze the possible pushing/calling ranges after you and see how that affects your EV.

[/ QUOTE ]

Please read other posts, as most of this has been addressed. Regarding the semantics, we are indeed talking about programs such as SnGPT that determine push or fold based on ICM in many cases, and were also discussing where the ICM falls short such as overvaluing short stacks and undervaluing bigstacks on the bubble. I will not go into more detail here as all of your questions, including those about B and C have been discussed thoroughly.

[tshort]

[ QUOTE ]
Regarding the semantics, we are indeed talking about programs such as SnGPT that determine push or fold based on ICM

[/ QUOTE ]

It would be a good idea to specify that you are addressing SNGPT, or SNGWIZ, or program xxxx rather than referring to "ICM".

Also, the shortcomings of both ICM and those programs have been discussed so many times you should be searching for plenty of good info in past posts.

[ymu]

I think there are two ways the OP is going wrong:

1. Assuming that those occasions where other considerations override or significantly modify ICM are much more common than they actually are.

2. Thinking that if he has a skill edge post-flop he can make a standard raise with 10BB and still have enough chips to outplay his opponent on the flop. He's not talking about folding some +EV pushes (although he's undoubtedly doing that too) - he's talking about raising and playing poker instead.

I'm interested to know what the OP thinks is causing his recent lack of profit in SNGs (the subject of his last thread) if he rejects all the advice he got in that thread (ie that he is failing to apply ICM appropriately). Passing up a lot of +EV edges seems to be a pretty good way to neutralise an otherwise +EV playing style.