ICM/Bubble Theory: Who Gets It? - High Stakes MTT Strategy

Two Plus Two Forums Archived Forums Strategy High Stakes MTT

ICM/Bubble Theory: Who Gets It?

...

10-31-2010 , 09:01 AM

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Hello all,

So I'll start by saying I'm mostly a cashgame player (1/2-2/4nl) who dabbles in MTTs/SNG on the side and have found them enjoyable. I was gonna type up a little bio so you all knew where I was coming from but realized this is gonna already border on tl;dr as it is.

So the issue I'm struggling with (and have been for some time), is obviously the title of the post. We all know what ICM is, how it affects our decisions (sorta), and for the post part, why.

What I don't know:
-Exactly how much of an effect does it have?
-When do we apply it? (more importantly, when do we not...or what rules do we bend?)
-How important is it to have a precise solution to ICM spots? (ie, what is the magnitide of any mistakes we could make?)

Important questions (with spoilered backstory if you have the time/inclination to read it):

Spoiler:

**From chat:**
-Q1: What flaws does a program like SNGWIZ have? Or more correctly, what inherent limitations does a mathematical approach to tournament equity have or impose upon us as players?
-Fairly simple +cEV, -$ or "Wiz-EV" shove spot. I don't think the hand is super important compared to the theory discussion that follows, but any thoughts would obv be welcome.

-Q2: Far-from-the-bubble, does ICM *NOT* apply? Or as my friend is phrasing it, does it apply but becomes negligible in the face of chip accumulation and its benefits?

-Qa: If ICM can be thrown out in spots, what about texts like Kill everyone that state Bubble factor throughout all MTTs is always greater than 1 (often >1.1)?

-Qb: If ICM is constant, does this mean there are tons of HSMTT/SNGers with marginal but expensive leaks?
-As an extreme example, should good players be routinely making a clearly -$EV call in ICM terms (the "22 v AK") purely to chip up because their skill edge justifies it?
Side note: credit to AMT in his classic DC series for explaining the math concisely...it could be from somewhere else but that's where I heard it first.

And Finally:

-Q: Do you prefer having "40bb when everyone else has 20" to "300bb when everyone else has 150bb"? Why (or where does your edge comes from that doesn't in the other scenario?)?
Assuming we're looking at raw chip edge of course...not just because in a typical structure we're closer to the money with 20bb stacks.

Again, I've talked to a couple people and read/watch a fair bit of material on the topic, to no satisfactory conclusion. Clearly though, it's a fundamental and important concept.
I feel like this problem has given me enough trouble to warrant a thread, but if this has been discussed/covered extensively, mods feel free to lock it and accept my sincerest apologies.

Thanks for taking the time to read/respond,
SemPeR

Last edited by SemPeR; 10-31-2010 at 09:04 AM. Reason: typos

10-31-2010 , 09:05 AM

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

The aformentioned chatlog from the spoiler that was very roughly paraphrased above:
(If your name is in the HH and you'd prefer it removed, let me know.)

Spoiler:

[1:23:04 AM] LSMTTF: Shoving ranges are just too easy to master, it's why SnGs are so reg infested in the first place
[1:24:17 AM] LSMTTF: like, I (and any good 180 or 90 grinder) undoubtedly have an edge over any top live MTT player at SnGs like Ivey or something.
[1:24:47 AM] LSMTTF: YOu're a smart guy, I'm sure you'll be able to master them in 3-4 hours of coaching
[1:26:14 AM] SemPeR: maybe over a couple weeks. I'd still take the over.
[1:26:32 AM] SemPeR: and yeah lol-live players
[1:27:03 AM] SemPeR: I mean I have wizard and am gonna be studying it a lot if I take this sne chase.
[1:27:20 AM] LSMTTF: Dude. Shovebotting is easy. You have 10x the mind for the game than I do, and it was not hard for me to master shovebotting at al.
[1:27:24 AM] SemPeR: so I just need someone who understands the climate of the games so I can plug in the right ranges(?) after I have the fundamentals down.
[1:27:35 AM] LSMTTF: Be careful with Wixard
[1:27:40 AM] LSMTTF: ya, bassically
[1:28:09 AM] SemPeR: what about wizard?
[1:28:10 AM] LSMTTF: once you learn standard ranges you just adjust for ante sizes and stuff like that, and you should be able to beat any SnG
[1:28:19 AM] LSMTTF: Wiz is gettin outdated
[1:28:25 AM] SemPeR: actually I have a hand i wasn't sure about the conclusion wizard came to.
[1:28:33 AM] LSMTTF: is can be super super wide in some spots
[1:28:45 AM] LSMTTF: you can send it if you like
[1:28:47 AM] SemPeR: iuno how something that just calcs he ev of a shove can become obsolute. structures aren't changing.
[1:28:52 AM] SemPeR: what do you mean super super wide
[1:29:15 AM] LSMTTF: Structures do change. 45s have different ante structures than 180s
[1:29:32 AM] LSMTTF: like, you have to adjust wizard very carefully to opponents calling ranges
[1:29:37 AM] SemPeR: I meant payouts.
[1:29:46 AM] LSMTTF: otherwise it can give you extremely incorrect feedback
[1:29:57 AM] SemPeR: ya ofc
[1:30:06 AM] LSMTTF: 45s also have different pay structures % wise than 180s
[1:30:26 AM] SemPeR: ***** Hand History for Game 51939324803 ***** (Poker Stars)
Tourney Hand NL Texas Hold'em - Sunday, October 31, 03:11:54 ET 2010
Table 326605906 1 (Real Money)
Seat 1 is the button
Seat 1: huihao ( $1120.00 USD )
Seat 2: PSCPRODIGY ( $2170.00 USD )
Seat 3: sook713 ( $1925.00 USD )
Seat 7: Dumzy ( $3600.00 USD )
Seat 9: HERO ( $4685.00 USD )
huihao posts ante of [$25.00 USD].
PSCPRODIGY posts ante of [$25.00 USD].
sook713 posts ante of [$25.00 USD].
Dumzy posts ante of [$25.00 USD].
HERO posts ante of [$25.00 USD].
PSCPRODIGY posts small blind [$200.00 USD].
sook713 posts big blind [$400.00 USD].
** Dealing down cards **
Dealt to HERO [ Jd 9d ]
Dumzy calls [$400.00 USD]

mind the stacks.
It's a $114 I think. dumzy is a fish, 34/10.
pscprodigy is a mass tabling reg. all others unknown. wouldn't be surprised if they were also regs.
[1:30:30 AM] LSMTTF: but as far as EV of a shove goes, you need to adjust to ante size and ****
[1:30:58 AM] SemPeR: push or fold, and the bottom of your range
[1:31:13 AM] SemPeR: and yes, we'd prob minraise aces but nobody knows that ****.
[1:31:41 AM] SemPeR: (actually at <10 minraising aces is gonna look stronger and get a good player to fold like 77)
[1:31:52 AM] LSMTTF: if its a $114, players are going to see through 2x aces
[1:31:56 AM] LSMTTF: yeah
[1:32:11 AM] LSMTTF: I push
[1:32:22 AM] LSMTTF: bottom of my range is like T8s
[1:32:25 AM] LSMTTF: so it's close
[1:32:27 AM] SemPeR: but as a rule you think wiz is generally wider than normal
[1:33:14 AM] LSMTTF: ehhh, yeah, I would probably say so
[1:33:31 AM] LSMTTF: sometimes it is too tight, but i think usually it is too wide than too tight
[1:33:47 AM] SemPeR: okay. And now if you're in the bb, how wide are you calling
[1:34:11 AM] SemPeR: given i cover you and you think I'm pushing as wide as you are in that spot
[1:34:32 AM] LSMTTF: QJs would be bottom
[1:34:37 AM] LSMTTF: ehhh
[1:34:42 AM] LSMTTF: probably QJo
[1:34:58 AM] LSMTTF: but that would be rock bottom uncomfortable call
[1:35:07 AM] SemPeR: see I put this into wiz and gave people tighter than normal ranges. like A8o, A6s, KTs.
[1:35:15 AM] SemPeR: wiz still says it's a fold
[1:35:26 AM] LSMTTF: it doesnt want you to shove?
[1:36:29 AM] SemPeR: if the fish is limping 40% and calling 11. (like he'll fold 75% of the time, which is conservative but reasonable), and everyone else is calling <<<15% (like 11-14% avg),

our pushing range is 66+,ATs+,AJo+
[1:36:31 AM] LSMTTF: Is it telling you to ICM fold? Becausr that isn't a huge factor riht not
[1:36:47 AM] SemPeR: well that surprises me just based on the stuff I've read
[1:36:51 AM] SemPeR: Why is icm not a huge factor
[1:37:38 AM] LSMTTF: not on bubbe, we are chiplead, and 3 stacks behind us aren't too scary
[1:37:49 AM] LSMTTF: yeah, that doesn't seem right
[1:38:14 AM] LSMTTF: the limper in there is what widens our range from like JTs to like T8s
[1:38:25 AM] LSMTTF: J9s should always be a push IMO
[1:38:33 AM] SemPeR: I'm the last person to defend a real person's advice over something like wiz, but wiz's analysis makes sense to me.
[1:38:34 AM] LSMTTF: especially with dead money
[1:38:44 AM] LSMTTF: ok
[1:38:46 AM] SemPeR: Why is not being on the bubble / chiplead stuff
[1:38:58 AM] SemPeR: make us disregard them as much as you seem to be
[1:39:53 AM] SemPeR: -losing a flip vs them is still worth a lot
-we're still all <10bb, so maintaining FE pre is a concern
-we have very few chips if the fish calls and wins, which should tighten our range
[1:40:56 AM] LSMTTF: ICM is like a lottery ticket thing. ICM deff matters, but IMO (and I guess Bubuiis too since he coached me) it's not a big factor until bubble, and then again until FT, and then again until 4 left, cause top 3 are so heavy. In an STT, I dont think it's a large deal until bubble, but because ICM affects everyone else too, we can push wide with large stack, because everyone is calling so tght
[1:41:11 AM] SemPeR: EqPC: 5238
EqFC: 4660

(from when I switch to chip equity mode).
But I mean we both know that. It's obvious it's a fistpump vaccum shove.
[1:41:25 AM] LSMTTF: right
[1:42:15 AM] SemPeR: It sounds like you're going to play it like a cashgame then
[1:42:20 AM] SemPeR: up until the direct bubble
[1:42:36 AM] SemPeR: which is final 4 in a stt.
last 5-10 in a mtt.

and nowhere else
[1:42:52 AM] SemPeR: (disregarding stuff about preserving stack for FE and stuff. I just mean the shoving ranges)
[1:43:58 AM] LSMTTF: Bassically yeah. And that might be a leak in an STT, but in MTT SnGs I was at like 50% ROI, so I don't think its a leak there. I'm trying to transfer over skillset as best I can to apply to STT
[1:44:32 AM] LSMTTF: In MTT SnG you always play to win. Decisions aren't affected too much unless there is a hyper shorty, and even then that can make you push wider sometimes
[1:44:49 AM] LSMTTF: I would deff imagine its same in STT
[1:47:11 AM] SemPeR: that just seems incorrect to me.

keep in mind I don't really have an opinion on this. I feel kinda clueless and just want to figure out the right path, so to speak. So all of my disagreement is in the interest of just playing better poker.

-basically all the stuff I've learned about icm/bubble is that while it is definitely most noticeable right at the money jumps, the effect/adjustments u need to make are constantly present throughout the MTT/SNG.

std example is the bubble factor in the sunday milly or wsop is gonna be something between 1.05 and 1.15.
[1:48:23 AM] SemPeR: Which means in every single decision, from a pot odds call to a multibarrel bluff, is effected by a "for every chip you lose it's worth 1.xx more than a chip you win" factor, or 'bubble factor'
[1:48:51 AM] SemPeR: so playing it "basically" like a cashgame is gonna be a fundamental mistake.
[1:48:57 AM] LSMTTF: I mean, for example, on the bubble, everyone elses call ranges are likely to tighten up because next to act after you is so short if you shove, and they dont want to bust before him, which is why you can shove wider. I remember us talking about bubbe factor in Vancouver, but I still don't know any MTTers who apply that theory.
[1:49:33 AM] LSMTTF: The MTTers I know, do play MTTs like cash games that are 30bbs or less deep, and that they are very willing to gamble in
[1:49:58 AM] SemPeR: yeah thats the impression I got from you and talking to some others
[1:50:05 AM] SemPeR: iuno this whole thing keeps confusing me
[1:50:11 AM] LSMTTF: No matter how many sessions ive played with other MTTers, or topics weve discusses, ive never heard bubble factor brought up
[1:50:27 AM] LSMTTF: ok, so im not the only one you hear this from?
[1:50:35 AM] SemPeR: So does the standard in MTTc basically just mean that all these random texts and math-y simulations are wrong?
[1:50:56 AM] SemPeR: Ofc, because clearly you're hearing it from similar players.
[1:51:11 AM] LSMTTF: I mean, the concept deff makes sense to me in theory, but in practice I think it's hard to apply sometimes
[1:51:43 AM] SemPeR: But I'm not finding a single decent argument in there that refutes the idea of a bubble factor in an mtt.

All of them say basically what you say. Everyone else does it this way, so that's the theory.
[1:52:51 AM] LSMTTF: let me cook up an example for you real quick where bubble factor is impractical
[1:52:55 AM] SemPeR: again im not trying to defend the book or say the math is always right, but when fairly static stuff is worked out to a near-solution (like bubble factors are; i mean all we're talking about is how to approach an mtt in general. nothing amorphus), it kinda demands to be disproved
[1:53:08 AM] SemPeR: okay give it a shot
[1:53:27 AM] LSMTTF: I understand where you're coming from
[1:53:56 AM] LSMTTF: ok, i just realized that I'm not going to be able to come up with an example that you couldnt disprove mathematically
[1:54:16 AM] LSMTTF: did I ever talk to you about why -EV shoves are ok to make in early stages?
[1:54:24 AM] SemPeR: yup
[1:54:27 AM] SemPeR: chipping up
[1:54:32 AM] LSMTTF: exactly
[1:54:35 AM] SemPeR: gigabet's block theory (note for anyone reading this: google Gigabet Dilemma if it sounds new)
[1:54:44 AM] LSMTTF: sure, if thats what its called
[1:54:48 AM] SemPeR: nah that's a side thing
[1:54:51 AM] SemPeR: but go on
[1:54:54 AM] LSMTTF: haha
[1:54:55 AM] LSMTTF: ok
[1:55:23 AM] LSMTTF: I imagine the reason that bubble facotr is ignored is largely the same reason
[1:56:03 AM] SemPeR: If that's the only reason then I still don't know. Should probably post in hsmtt about it with a couple examples.
[1:56:06 AM] LSMTTF: because regs can use chips they gain to more advantage than their opponents can, so the skill factor would negate the bubble factor
[1:56:10 AM] LSMTTF: or soemthing to that effect
[1:56:47 AM] SemPeR: And I imagine in the 10bb scenario you were talking about the risk of losing some chips is not enough to compensate for having a 15bb stack that can continue to shove wider and pick up more chips.
[1:57:03 AM] SemPeR: (just so we both think I understand precisely what you're getting at)
[1:58:04 AM] LSMTTF: I mean, I think the hand from before it just largely +EV, bubble factor or not, because everyone behind next to act is calling so tight once the guy with 2.5bbs folds
[2:00:37 AM] SemPeR: this is what I'm seeing:

all mtters subscribe to the "if you have 22 in the bb you should in theory fold it at 150bb if someone ships AKo and shows you before the action gets to you"

and somewhere along the way that gets lost in translation and we're suddenly lags again. (and i guess from what I understand of what you're saying, lots of hsmtts would call the 22 online b/c they can use the doubleup to chip up; conversely they're prob not unhappy to be holding the AK too)
[2:01:23 AM] LSMTTF: I would agree with all of that.
[2:01:45 AM] LSMTTF: I am happy on both ends of the spectrum early in an MTT really
[2:02:12 AM] LSMTTF: I mean, 150 is a bit deep to be taking flips
[2:02:22 AM] SemPeR: Like with the theory of bubble factor in an mtt, it's not supposed to be 0 ever (right?).

So I guess I'd use this conversation we had to lower tha bubble factor of those kinds of decisions.

For example I could put an edge of -.1% (or more).
[2:02:32 AM] SemPeR: ..in wizard
[2:03:04 AM] SemPeR: to do my shoves from now on (will have to ask better players abt that since it's kinda out there. obv the goal is to widen ranges...assuming we can get semicorrect ranges to beginwith for our villains)
[2:03:22 AM] LSMTTF: yeah, theoretically you can deff do that
[2:03:23 AM] SemPeR: ok, I'm a little confused about this:
[2:02 AM] LSMTTF:

<<< I mean, 150 is a bit deep to be taking flips
[2:03:33 AM] SemPeR: wouldn't you want to be taking flips as deep as possible
[2:03:41 AM] SemPeR: so the value of having a large stack is multiplied
[2:03:58 AM] LSMTTF: but you should deff talk to some HSSTT players about call/shoving ranges in STTs and such
[2:04:02 AM] SemPeR: Or are you saying having a large stack at 20bb is worth more than having a large stack at 300bb
[2:04:09 AM] LSMTTF: i know STTs dont have antes usually
[2:04:29 AM] SemPeR: well we're just talking about theory here and your opinion is all.
[2:04:34 AM] SemPeR: lots of guys to ask about that stuff.
[2:04:48 AM] LSMTTF: I would say having 40bbb when everyone else has 20 is worth more than having 300 when everyone else has 150
[2:04:57 AM] SemPeR: okay. that's new.
[2:04:59 AM] LSMTTF: right, in my opinion
[2:05:04 AM] SemPeR: ofc ofc
[2:05:25 AM] LSMTTF: and keep in mind, there are obv plenty of better MTTers than me, haha
[2:05:47 AM] LSMTTF: but I DO think a large percentage would agree with that statement

-How much of what my friend is saying makes sense? How much of it does not? Multiple opinions on this would be ideal.

Again, thanks.

10-31-2010 , 09:33 AM

skater3598

Pooh-Bah

Join Date: Jun 2009 Posts: 4,597

cool thread

i'm super interested in responses to Q1 especially when the blinds are high relative to your stack, and Q2-QB which i think are minor alterations to the same question about chasing small edges away from the money

i think you should rephrase the last question a little bit for clarity, but i think what you're getting at in that question is more of an issue related to stack sizes and people are just going to tell you "the more BBs effective you have the bigger your edge is", as the question stands

i don't think the chatlog is really necessary

Last edited by skater3598; 10-31-2010 at 09:48 AM. Reason: .

10-31-2010 , 09:47 AM

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Ty.

To go ahead and rephrase that last Q before it gets chased down a page or 2:

I'll admit it's a little tricky and you can probably tell my wording is iffy with that Q:
but in essence I'm asking

Early or Late? At what stage in the tournament would you be happier to take a marginal situation and risk your stack (or large parts of it) to chip up?

At least that makes an attempt at hinting at exploitability/potential without centering it around stacksizes. I'm not sure if I was successful. Honestly I sat here for >2mins trying to rephrase without mucking it up. Perhaps I'm asking the wrong question..

I'd make a poll but I feel like there's enough stuff in this thread. Maybe midway if there's actually a debate.

10-31-2010 , 09:53 AM

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Quote:

Originally Posted by skater3598

i don't think the chatlog is really necessary

hehe. read the 1st edit before your 2nd. ;P

Feel free to put it back. I think it contributes.
It's definitely not the meat of the thread but hearing what parts of his responses sounded off/dumb/plain bad was the goal.

So purely self-interested post there, just to see how he's thinking compared to the avg poster here. I'd take it down or find a way to condense if I could edit, but as it is I think the spoiler is a reasonable middleground.

10-31-2010 , 09:56 AM

skater3598

Pooh-Bah

Join Date: Jun 2009 Posts: 4,597

i'd be happier early on because of screen space equity plus getting more BBs allows me to remain in the tournament to the point where there are antes more often and have a lot of chips at that point... it gets more favorable to have a lot of chips the later you go (more chips = more you can push your edge) all the way to the FT, but there's probably a ceiling to this though where i don't really want to chase 5% edges on 1000 BB pots early on when the average stack in the tournament is only 100 BB because you bust often and the other guy auto-final tables often which doesn't seem optimal for us... maybe not though. TBH i wouldn't be surprised at all if it was revealed to me that chasing a 5% (arbitrary #) edge at any point in a tournament isn't optimal if you risk your tourney life.

fwiw i think the factors you should consider are what % of the chips in the tournament you have and how close you are to the money, not how many BBs.

re-edited in: he's pretty off with a few things, for starters "shovebotter" is not a word, sng wiz can't be outdated as you picked up on, you can't learn how to pushbot in 3 hours, sngs are reg infested... didn't read the whole log

Last edited by skater3598; 10-31-2010 at 10:15 AM. Reason: ninjas gonna edit

11-01-2010 , 08:10 AM

skater3598

Pooh-Bah

Join Date: Jun 2009 Posts: 4,597

Quote:

Originally Posted by SemPeR

What I don't know:
-Exactly how much of an effect does it have?
-When do we apply it? (more importantly, when do we not...or what rules do we bend?)
-How important is it to have a precise solution to ICM spots? (ie, what is the magnitide of any mistakes we could make?)

**From chat:**
-Q1: What flaws does a program like SNGWIZ have? Or more correctly, what inherent limitations does a mathematical approach to tournament equity have or impose upon us as players?

-Q2: Far-from-the-bubble, does ICM *NOT* apply? Or as my friend is phrasing it, does it apply but becomes negligible in the face of chip accumulation and its benefits?

-Qb: If ICM is constant, does this mean there are tons of HSMTT/SNGers with marginal but expensive leaks?

edited the OP... had cool questions but was a little tl;dr

can any of you ICM nits please comment on these?

11-01-2010 , 10:47 AM

Shai Hulud

old hand

Join Date: Feb 2010 Posts: 1,919

Quote:

Originally Posted by skater3598

edited the OP... had cool questions but was a little tl;dr

can any of you ICM nits please comment on these?

In response to Q1, there are two flaws with ICM. It calculates tournament equity ($EV) based on the current stack setup and the payout structure, nothing else. If player A and player B have the same stack size, then it is assumed they will cash the same amount of time. So the first flaw is neglect of skill as a determining factor when calculating various players' equity. The second flaw is that equity is determined based on the current stack setup in one hand only. ICM is a static model. There is no way to account for future +-EV situations. For instance, it does not account for the future value of having a large stack on the bubble. Nor does it account for the future loss of equity from blinds reducing your stack. ICM is also positionally naive. For instance your equity can be very different based on whether you have a LAG on your left or on your right, or, if you're short, depending on whether you or another shorty will hit the blinds first.

A more dynamic model taking into account skill factors and future equity would obviously be better but it would be tremendously complicated, and calculating whether a move were +EV would probably take a lot of computational power. Until such a model is developed, though, I think ICM is the best we have, and it's certainly useful if one is aware of its limitations.

11-01-2010 , 10:58 AM

skater3598

Pooh-Bah

Join Date: Jun 2009 Posts: 4,597

great points
ty sir

11-01-2010 , 09:06 PM

#10

iFloat

banned

Join Date: Sep 2010 Posts: 332

Quote:

Originally Posted by Shai Hulud

Having said that, I was wondering if you might answer the following questions from the OP? This is a subject I am very interested in as well, and you seem a good deal more knowledgeable on it than I.

-Q2: Far-from-the-bubble, does ICM *NOT* apply? Or as my friend is phrasing it, does it apply but becomes negligible in the face of chip accumulation and its benefits?

-Q: Do you prefer having "40bb when everyone else has 20" to "300bb when everyone else has 150bb"? Why (or where does your edge comes from that doesn't in the other scenario?)?
Assuming we're looking at raw chip edge of course...not just because in a typical structure we're closer to the money with 20bb stacks.

11-01-2010 , 11:14 PM

#11

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Good response.

It sounds like with the kinds of programs out right now, we won't be able to have/use a more dynamic model as you put it, anytime soon. Honestly this isn't a bad thing, and I have my doubts on exactly how it'd help, since a lot of the variables you mention will be hard to quantify.

I think the route to go is coming up with ways to remedy the weaknesses of ICM.
Getting an idea of how much a large stack is really worth, quantifying the -value of having the blinds hit you in 1,2, or 3 hands. Because obviously the danger we run into when we just ballpark-account for them is vastly over/under-adjusting (which would be quite easy to do at various blind levels even at high stakes).

11-02-2010 , 03:49 AM

#12

Wizard-50

Pooh-Bah

Join Date: Apr 2008 Posts: 5,783

First of all, listen to your friend. I'm guessing he's a significant MTT winner.

-Q1: What flaws does a program like SNGWIZ have? Or more correctly, what inherent limitations does a mathematical approach to tournament equity have or impose upon us as players?
It can't calculate(realistically) the implications outside of that particular hand. For example if you call off in a slightly -ev spot, you won't get shoved on as often, therefore having higher ev on your future opens. It can't calculate the value of your stack size if you win. etc.

-Q2: Far-from-the-bubble, does ICM *NOT* apply? Or as my friend is phrasing it, does it apply but becomes negligible in the face of chip accumulation and its benefits?
Agree w/ your friend for the most part.

-Qa: If ICM can be thrown out in spots, what about texts like Kill everyone that state Bubble factor throughout all MTTs is always greater than 1 (often >1.1)?
The ICM factor is what is driving my(and your friend's) desire to have a bigger stack to use effectively. In other words, it will motivate other players to play sub-optimally against me in many cases.

-Qb: If ICM is constant, does this mean there are tons of HSMTT/SNGers with marginal but expensive leaks?
See Qa.

And Finally:

-Q: Do you prefer having "40bb when everyone else has 20" to "300bb when everyone else has 150bb"? Why (or where does your edge comes from that doesn't in the other scenario?)?
40/20 > 300/150 for the same reasons. ICM is a greater factor with 40/20 than with 300/150 which makes many opponents exploitable. Ideally I'd prefer closer to 50/25 because it will lower opponents 3b frequencies, thus making for more and higher +ev opens.

edit: I'm starting to teeter on this one. 300/150 might have a higher edge for me than 40/20 considering the ammount of play. Make it 40/20 vs 80/40 and I'll choose 40/20 all day.

I'm leaving for Venetian tomorrow so i might not check back for a while. I'll send you my email/aim SemPer.

Last edited by Wizard-50; 11-02-2010 at 04:11 AM.

11-02-2010 , 06:14 AM

#13

generalts0

grinder

Join Date: Feb 2010 Posts: 601

Quote:

Originally Posted by skater3598

i'd be happier early on because of screen space equity plus getting more BBs allows me to remain in the tournament to the point where there are antes more often and have a lot of chips at that point... it gets more favorable to have a lot of chips the later you go (more chips = more you can push your edge) all the way to the FT, but there's probably a ceiling to this though where i don't really want to chase 5% edges on 1000 BB pots early on when the average stack in the tournament is only 100 BB because you bust often and the other guy auto-final tables often which doesn't seem optimal for us... maybe not though. TBH i wouldn't be surprised at all if it was revealed to me that chasing a 5% (arbitrary #) edge at any point in a tournament isn't optimal if you risk your tourney life.

fwiw i think the factors you should consider are what % of the chips in the tournament you have and how close you are to the money, not how many BBs.

re-edited in: he's pretty off with a few things, for starters "shovebotter" is not a word, sng wiz can't be outdated as you picked up on, you can't learn how to pushbot in 3 hours, sngs are reg infested... didn't read the whole log

11-02-2010 , 06:27 AM

#14

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Quote:

Originally Posted by skater3598

i wouldn't be surprised at all if it was revealed to me that chasing a 5% (arbitrary #) edge at any point in a tournament isn't optimal if you risk your tourney life.

Honestly, to me this is the point of the entire thread. It's that grey area that could potentially be super expensive at the later stages of a highbuyin, and I really want to figure it out.

Wizard, do you think you could address this specifically? Or does your previous reply stand as is?

Like on one end of the spectrum, there are gonna be textbook spots pulled right out of an SNG where you have to make that fold with QQ or whatever because there's a shorty involved or about to hit the blinds.

On the other end you have someone really good (elky's pca run a few years back vs david pham comes to mind), making a clearly -cEV and $EV call to knock out a strong player and dominate the table.

Somewhere in the middle there are people making mistakes, and given how bubble factors work, these mistakes aren't that small in magnitude (definitely not in frequency).

There's a question I ask in the chat, where I say something along the lines of
"would you play it like a cash game?". I've been wondering about this a lot.

Basically it seems crazy to just throw out bubble factor (ie, "yes, we should play it like a cash game or possibly even looser!).

11-02-2010 , 06:30 AM

#15

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

...not that it'd really be against it.
Keep in mind I come from a hu / 6m cash background. ;p

I just want to play the best poker I can. Either I'm being exploited by all these regs overvaluing cEV, or they're making fundamental mistakes that I'll be able to profit off of by picking the correct cutoffs.

If someone like shaun deeb or timex just came out and said "yeah fk bubble factor", I'd feel right at home and start lagging it up again.

11-02-2010 , 07:58 AM

#16

Shai Hulud

old hand

Join Date: Feb 2010 Posts: 1,919

According to ICM there is always a bubble factor at any point in any tournament (except winner-take-all). The chips you gain are never worth quite as much as the chips you lose. This is because X amount of chips does not always equal X amount of chips. If you are down to your last 500 chips, those chips are worth a LOT more in terms of tournament equity than the 500 chips you gain from a steal when deep stacked. And this discrepancy is true the very first hand of a tournament. Doubling your stack does not double your equity. This discrepancy, called "the bubble factor" or "ICM tax," is always there, but it is not always that noticeable. Its importance depends on how "far" you are from the bubble, the percentage of the field cashing, and the relative flatness of the payout structure. The earlier in the tournament it is, the lesser percentage cashing, and the more top-heavy the payout structure, the less important the bubble factor. In double or nothings the effect is huge. In a winner-take-all tournament there is no bubble factor at all.

However, this doesn't mean it's a good idea to pass up certain +CEV spots just because they're -$EV. Due to rising blinds one cannot wait around forever for +$EV spots so chip accumulation is an end in itself. As I said before, ICM is a static model that places only a limited value in the future equity gained by chip accumulation. I say limited because the increase in equity assigned by ICM is based solely on the stack setup. But having a workable stack affords a good player with many more +$EV opportunities in future hands.

For example, say it's 200/400 and you have 2000 chips and get it in in some situation that is +CEV but slightly -$EV. It is worth it so long as you anticipate being able to make up the loss in $EV by the increase in +$EV opportunities afforded to you by having a larger stack. And this would generally be the case since with a 10BB stack you gain fold equity and hence can shove wider and can hence exploit opponents more effectively.

It is also worth it to gamble in -$EV spots if you anticipate that not gambling for all your chips will lead to even worse -$EV spots in the future. A common situation is being short-stacked UTG where taking the blinds would destroy your fold equity. Here ICM overestimates our tournament equity because it doesn't consider that our tournament equity will drop drastically over the next two hands if we pass up on slightly -$EV spots. This is why we should shove much wider in this spot than programs such as SNGWiz would suggest, because SNGWiz calculations use ICM and don't consider the gain or loss of equity in future hands.

11-02-2010 , 01:08 PM

#17

Wizard-50

Pooh-Bah

Join Date: Apr 2008 Posts: 5,783

Quote:

Originally Posted by Shai Hulud

good stuff here.

Quote:

Basically it seems crazy to just throw out bubble factor (ie, "yes, we should play it like a cash game or possibly even looser!).

I'm not suggesting at all that we should throw out the bubble factor; I'm just saying you need to be aware of it and how it affects your opponents play. Certainly if you have a murders row of players on your left I don't think you should be opening much at all if you're not prepared to put in some light action because the bubble is more of a factor for you than your opponent now. Hence you are now playing sub-optimally against them.

The problem is that playing cEV optimal makes it a poo-flinging contest that could be -$ev in a static example with ICM considerations. But the future implications could be +ev. For example if one player is giving you lots of problems(like the elky example) there is a value to knocking him out that is hard to quantify.

The same can be applied to the implications of certain actions you can take when it might be close to neutral cEV. 4 betting will(should) lower their 3b frequency, as well as calling a 3b in some cases. calling a shove when you have a big enough stack and there are multiple shoving stacks at the table can lower their jamming frequency.

Watch the best MTT players in the world. They leverage ICM for their benefit. I'm not saying ICM is not to be considered, just that you should consider it differently.

11-02-2010 , 09:17 PM

#18

unrealzeal

banned

Join Date: Feb 2008 Posts: 13,990

I can answer question 1 I think!

the thing that sngwiz doesn't take into account is opportunity cost for the blinds moving up shortly...I can't find the hand history for this game, but a hand that I played recently comes to mind

STT 6max 3 players left so directly on the bubble...the button is an experienced reg with 5bb's left at 125/250 ante 50, he shoves what I consider to be roughly ATC...I find KTo in the SB and have him covered almost 4.5:1

while KTo is def far enough ahead to call, I elect to fold because the blinds are moving up shortly and he will be left with 4bbs, and will be forced to call my raises much lighter, so what wiz doesn't factor in is what i consider the opportunity cost of losing...should I call with KTo and lose he will have enough blinds to be selective with his calling range...if i call and win i do make the money, but i think starving the low stack for future blind raises is a crucial aspect of ICM that wiz misses

incidently he shoved KQo into my AKo on the next orbit and won

-- so much for my theory!

11-04-2010 , 02:14 AM

#19

MUD

adept

Join Date: May 2005 Posts: 817

Quote:

Originally Posted by SemPeR

-Q1: What flaws does a program like SNGWIZ have? Or more correctly, what inherent limitations does a mathematical approach to tournament equity have or impose upon us as players?

The biggest flaw w/ SNGWIZ is the that the ranges are only in top X%. For example you cant include 54s w/o also including K6o.

But the limitation of ICM (interchangable w/ math approach) is based at very high blinds. I mean <4 Big blinds. Everywhere else I consider it close to perfect, given correct ranges and EQ% edges needed given future advantage (which I would say is overestimated by good players)

Quote:

-Fairly simple +cEV, -$ or "Wiz-EV" shove spot. I don't think the hand is super important compared to the theory discussion that follows, but any thoughts would obv be welcome.

-Q2: Far-from-the-bubble, does ICM *NOT* apply? Or as my friend is phrasing it, does it apply but becomes negligible in the face of chip accumulation and its benefits?

It does apply but often its so small and your not talking about big $EV pots that it doesn't really matter. But ya if you want to divide your direct and implied odds by 1.05 early on in tourneys you'll be a better player.

Quote:

-Qa: If ICM can be thrown out in spots, what about texts like Kill everyone that state Bubble factor throughout all MTTs is always greater than 1 (often >1.1)?

The books are right and most everyone else is wrong

Quote:

-Qb: If ICM is constant, does this mean there are tons of HSMTT/SNGers with marginal but expensive leaks?
-As an extreme example, should good players be routinely making a clearly -$EV call in ICM terms (the "22 v AK") purely to chip up because their skill edge justifies it?
Side note: credit to AMT in his classic DC series for explaining the math concisely...it could be from somewhere else but that's where I heard it first.

Yes, I would say ICM leaks esp. given todays flatter payouts are most Mid-HSMTT biggest leak.

No

Quote:

And Finally:

-Q: Do you prefer having "40bb when everyone else has 20" to "300bb when everyone else has 150bb"? Why (or where does your edge comes from that doesn't in the other scenario?)?
Assuming we're looking at raw chip edge of course...not just because in a typical structure we're closer to the money with 20bb stacks.

Think about it this way. If every chip is worth less than the last, then anytime you get to immediately threaten a bigger percentage of a chip stack the better. Say your bubble factor is 1.2 for a call, it is probably ~1.15 for the first chip and ~1.25 for the last, avg 1.2. But it works in reverse as well, as you will have higher bubble factors against them.

11-04-2010 , 07:43 AM

#20

skater3598

Pooh-Bah

Join Date: Jun 2009 Posts: 4,597

^gets it

11-04-2010 , 01:51 PM

#21

SemPeR

centurion

Join Date: Jul 2008 Posts: 171

Quote:

Originally Posted by skater3598

^gets it

truuu.

Excellent post MUD. Thanks for taking the time.

02-01-2011 , 12:11 PM

#22

sufur

Carpal \'Tunnel

Join Date: Sep 2008 Posts: 6,364

I'm surprised this didn't get more discussion =/

...