Challenging the basics behind ICM - Page 2

Freudian · #11 10-31-2005, 06:29 PM

I have no problem with ICM. But since the input is a guess, the output will also be a guess. The better you are at guessing, the better the guesses ICM makes will be.

I do think it is a useful tool for grasping situations where the solution is unintuitive.

#12 10-31-2005, 06:42 PM

[ QUOTE ]
you do read your opponents when using ICM by putting what their calling ranges are. Do you know what ICM is other than a statistical model?

[/ QUOTE ]It's not a statistical model, is it?

protoverus · #13 10-31-2005, 06:44 PM

[ QUOTE ]
The most important skill in poker (including SNGs) is to understand your opponents and how they probably view you. The examples supporting ICM make that point. You must estimate the likelihood of you opponents making a play (etc.).

[/ QUOTE ]

This is an arguable point. Sure, it's important, but THE MOST important in all of poker? What good its it to know your opponent if you can't play mathematically correct? EV still rules as the most important concept IMO.

[ QUOTE ]
We must learn to read our opponents more than relying on mathematical techniques.

[/ QUOTE ]

We must learn to read our opponents so that we can apply mathematical techniques more accurately.

PrayingMantis · #14 10-31-2005, 06:48 PM

[ QUOTE ]
We must learn to read our opponents more than relying on mathematical techniques.

[/ QUOTE ]

OK for the last time, using ICM and "reading our opponent" do not exclude each other at all. There's no meaning in saying that you must learn A more than B. Actually, with regard to SNGs, you can many times do fairly good with a very general read, while applying ICM thinking without being very specific/accurate with ranges. It won't be great, but will get you some results. The other way around (i.e, using reads, and no ICM at all) could work too in some sense, but will probably be less efficient, or even -EV especially if your strategies are very different from what an ICM-based strategy would suggest, specifically around bubble.

It doesn't seem like you understand the meaning of ICM at all. There's nothing wrong with criticizing ICM or suggesting new models, but in this case you are just throwing around words like "reads" and "intuition", that have nothing to do with how good or bad is ICM.

mlagoo · #15 10-31-2005, 06:50 PM

[ QUOTE ]
We must learn to read our opponents more than relying on mathematical techniques.

[/ QUOTE ]

This is wrong, I think. It's not a matter of doing one more than the other. It's a matter of using the two in collaboration.

If an opponent pushes allin for 1400 chips, you're in the BB with 1000 chips, and you have a stone read that he has AK, but have no understanding of math -- how is that read going to tell you what to do?

Conversely, if you know that versus a pushing range of AQ+ and 99+, you have x equity, but you have no idea of what villian's range is, you are also not going to make the correct decision (not consciously, anyway).

There is nothing wrong with ICM. It serves its purpose just as it was meant to. It's just that its purpose isn't what you apparently think it is. It works in conjunction with reads to help you make optimal decisions. This is a good thing.

FlyWf · #16 10-31-2005, 07:15 PM

ICM isn't a statistical model. I'm going to take you at your word about being a stat geek, but do you have any experience with ICM outside seeing people in this forum reference it as a reason for a given action? Because this whole thread doesn't make a tremendous amount of sense. Reads don't have anything to do with ICM. ICM only has one real underlying assumption(equal skill).

More bluntly, can you tell me what you think ICM is?

vinyard · #17 10-31-2005, 07:48 PM

[ QUOTE ]
There is nothing wrong with ICM.

[/ QUOTE ]ICM is great but ICM is not without its faults particularly around the edges. I suspect your know this and it was an oversight but lest a noob thinks ICM is universally applicable I figured I would issue a (gentle) clarification.

ChrisV · #18 10-31-2005, 08:16 PM

[ QUOTE ]
As a long time statistics geek, I always wonder about the underpinnings of any statistical model.

[/ QUOTE ]

As others have mentioned, ICM is not statistical.

[ QUOTE ]
ICM works perfectly if assumptions are met. But assumptions are NEVER met. I have never played at a table where everyone had the same skill level.

[/ QUOTE ]

If you consider yourself at a higher skill level than the rest of the table (which I hope you do) then the effect is usually to increase further the ICM's emphasis on not calling off your stack.

[ QUOTE ]
ICM does not take context into consideration. Has the villian been stealing the blinds at every opportunity? How loose are the other players? Who just had a bad beat? How do the other players assess my play (am I considered tight and solid or loosey goosey)?

[/ QUOTE ]

All these things are factors in the calling or raising standards of the players, which are parameters which are fed into the ICM. To complain about them not being included in the ICM is bizarre. It's like complaining that pot odds calculations don't include those things. The ICM is just a method of calculating equity.

[ QUOTE ]
4 handed, you are the small stack in the BB - 1400 chips after seeing the blind (200). The button goes all-in (4th time in 5 hands). He has 3300 chips. You have KQ unsuited - what do you do and why?

Can ICM give you the answer?

[/ QUOTE ]

No. What it does is tell me how far ahead I need to be on average before I should call. It's up to me to decide what range I put my opponent on, at which point I can work out whether I'm far enough ahead or not.

Sometimes ICM can give me the answer though, in that there are situations where I should fold no matter what range I put my opponent on. Suppose I have 3500 chips after posting the blind of 400. The other stacks are 800, 700 and 4600. It folds to the guy with 4600 in the SB who moves in. I have ATo. The ICM tells me here that I need to be a 74.4% favourite to call. Since I'm nowhere even close to that even versus a random hand, my opinion on what he has doesn't matter. (For that matter, I can give him the range "unpaired cards ten or less" and it's still a fold).

How do you plan to work the above out without the ICM? You can do it with intuition, but you can do everything with intuition if you're good enough. Why bother with pot odds, or anything mathematical at all?

tigerite · #19 10-31-2005, 08:19 PM

And on the flip side, there are situations where you can push with any two cards, and certainly with mid pairs and AK/AQ etc and there's no possible range your opponents can call with that loses you $EV. Same goes for calling, also, occasionally.

#20 10-31-2005, 08:24 PM

Your example is easily answered by ICM if you give us the other stacks.

ICM works amazingly well if you can put people on good calling ranges.

Admittedly, ICM doesn't tell you your real equity in a situation but rather what your equity would be if everyone were equally skilled. However, this is ridiculously powerful and people should shut up about how it's just a model because using it will make you play WAY better than not using it.