ICM/SNGPT rambling thoughts(long)

[BadMongo]

[ QUOTE ]
no u don't make .70...after you pushed the +.2, the next hand's value is +.1 or even -.4...but if u fold the +.2 then u get +.5

[/ QUOTE ]

How do you know this? Why are you assuming that every push you make will decrease your EV oppertunities on future hands?

It could just as easily increase them. Suppose you have pushed the past three hands, then you pick up aces and push again. Is your EV going to be higher if you had folded those previous three hands?

[AliasMrJones]

I think it's been acknowledged by everyone that a push now (or, more accurately a series of pushes) can influence the future calling range of opponents and it would be great if we could factor that into the model. There, now we can finish with that part of the discussion.

Saying ICM does not factor in future events is hogwash. It is meant to PREDICT THE FUTURE. Predicting the future is ALL ABOUT FUTURE EVENTS. It models those future events. The model may not be accurate, but to say ICM does not take into account future events is silly.

ICM DOES NOT SAY, "If the tournament ended after this hand and prize money distributed by chips..." That would be a tournament chip equity = prize money equity model WHICH IS NOT WHAT ICM IS. In fact, ICM is supposed to predict prize money equity AT THE END OF THE TOURNAMENT better than the tournament chip equity = prize money equity model does.

I think you need to do some more research about ICM before you make any more statements about what it does and does not do because you clearly don't understand it.

[Nicholasp27]

exactly...u don't know...and neither does ICM...which is why u have to use JUDGEMENT to decide

my example was saying that u would pass up a +.2 for a +.5 if the fold/push was more total ev than push/whatever is...

thus there are times when it's better to pass up the +.2 edge...some are saying if u put in the right inputs into ICM and the model is accurate, then you should not pass up +.2 because there are no other important factors

i'm saying there are other important factors...my example shows the outcome of how the factors made going against ICM's vacuum hand a good thing...now it's up to us to decide what those factors are and how to account for them and thus incorporate ICM into our arsenal without using it as gospel

[Nicholasp27]

ok, it's not 100% in a vacuum...it does predict how much you should expect to win in the long run given equal skill levels with X number of chips vs Y number of chips vs Z number of chips...then takes into account your hand and your opps' calling ranges and does probability of X, Y, Z occurring and then tells u which move has the outcome of giving you the most equity after this hand, thus which move gives you more money in the long run with ALL ELSE EQUAL...but it's not equal...

it's wrong to say that icm tells u your expected equity if skill is even...it's more than if skill is even...it also depends on table image as well as skill...

if you push 27o with +.1 ev and get called, it takes that into account, but it doesn't take into account that your table image is now shot and the range opens up widely...

it doesn't take into account that if u steal someone's blind, the next time you push into their blind they may open up their range...even if it was just once...

it just cannot predict the future on its own, so your statement that there are no other important factors cannot be correct...even with 100% accurate inputs, sngpt still only does one hand...


icm may say that pushing gives you .23 equity and folding gives you .22 equity...but that does not mean u take it as gospel and push it

[ZeroPointMachine]

[ QUOTE ]
[ QUOTE ]
Raptor is right that playing a certain hand right culminates in getting right the calling ranges of your opponents. But he is only almost there. You have to factor in what your decision will mean to the future calling ranges and how this will affect your EV on your future weighted average decisions. If opponents loosen up, your average EV will very often go down which means that if you stay with the threshhold you will have less situations where you can push for profit.

[/ QUOTE ]

Excellent, excellent summation and something that is very difficult to model, but would be an excellent addition to the current state-of-the-art ICM model (which includes more than just ICM, but all the extra stuff that eastbay has packed into SnGPT) if we could come up with something reasonable.

[/ QUOTE ]

This may be the best point to come from this discussion. It's not where my brain was headed when I started but I like this idea.

Key factors in an ICM calculation are the size of the blinds
and your FE vs. villian's range. Large +$EV situations almost always involve large blinds. Most pushes involve the same 1 or 2 villians on your left.

So, you've pushed from SB into BB the last two orbits. Here you are again with a marginal(+.6%) push against his current range with 100-200 blinds. If you have a workable stacksize and you feel that pushing this hand will open villian's range drastically on future hands, can you argue that the lost FE in the future higher blinds out weighs the current slightly positive situation?

[BadMongo]

[ QUOTE ]
it's wrong to say that icm tells u your expected equity if skill is even...

[/ QUOTE ]

O rly?? [img]/images/graemlins/confused.gif[/img]

[ QUOTE ]
if you push 27o with +.1 ev and get called, it takes that into account, but it doesn't take into account that your table image is now shot and the range opens up widely...

[/ QUOTE ]

[ QUOTE ]
it doesn't take into account that if u steal someone's blind, the next time you push into their blind they may open up their range...even if it was just once...

[/ QUOTE ]

Yes it does. Have you seen those hand range input boxes SGA has? you should use them.

[BadMongo]

This is an interesting point, but I can't imagine how you could quantify something like that.

[AliasMrJones]

[ QUOTE ]
it just cannot predict the future on its own, so your statement that there are no other important factors cannot be correct...even with 100% accurate inputs, sngpt still only does one hand...


icm may say that pushing gives you .23 equity and folding gives you .22 equity...but that does not mean u take it as gospel and push it

[/ QUOTE ]

On a side note -- How long is the bubble going to last? 10 hands? 15? 20? There just isn't much effect possible to your opponents on average in that period of time. Some effect? Maybe, but enough to justify much deviation from ICM? I don't think so.


OK, back to the main discussion...

Please, please, please read about ICM so you actually understand it before you post anything more. ICM PREDICTS THE FUTURE. IT USES A MODEL TO DO SO. THAT MODEL MAY NOT BE 100% ACCURATE. THAT DOESN'T MEAN IT ISN'T TAKING FUTURE ACTIONS INTO ACCOUNT.

I think we can agree that being dealt AA is the highest +EV situation one can get preflop. What is the chance that we'll be dealt AA on the next hand? Exactly the same as our opponents' chances. Extrapolate -- what is the chance we'll get a +$EV opportunity on the next hand -- exactly the same as our opponents'. What is the probability that we'll have more +$EV opportunites than our opponents' from now until the end of the tourney? Exactly the same as the chance they'll have more +$EV opportunities than us. (Assuming equal skill -- again a current weakness of ICM model as implemented in SnGPT.)

So, you see, ICM DOES take into account future possibilities. It does so by using the fact that on a long-term basis and on a short-term probability basis, everyone will be dealt the same hands and thus have the same opportunities. It then uses a chip=lottery ticket model to determine chances to win prize money, which also is taking into account future actions (i.e. 1 chip does not equal 1/X share of prize money where X is total chips in play).

Have I ever passed up a +$EV opportunity because of table image. Probably, yes. Is it often correct to do so? Even on average once per tournament? I doubt it.

[Nicholasp27]

yes, rly, imo...if everything is equal, then you all have .1 at the beginning of tourney...but that's not just if skill is equal...it's also if table image is equal...

say u sit down next to a 2+2er of exactly equal skill...your ev is the same on hand 1 as theirs...but if u know who they are and they don't know who u are, then you have higher ev than them if u use that info to your advantage...thus their table image has changed for u...

well in the bubble, if someone catches u pushing 27o, then your table image changes...likewise if u fold the hand, then you are seen as a lil tighter than if u push, so ur table image changes...this affects your future ev and ICM can't model that currently

if u agree that knowing the person and them not knowing u raises your ev, then you agree that table-image is a factor along with skill in determining your prize equity...again, icm is static and just gives you your expectation based on equal skill, equal table image, based on your input of stacks and ranges and cards that hand



and icm does NOT take into account how your actions this hand affect your future ev

it takes it into account THAT hand...based on your inputs

there is currently no model that states that "if you push this hand it will give you +x ev and will change the villian's range to y which will then lower the expected value of your next hand's ev by z"

[ QUOTE ]
[ QUOTE ]
Don't play above your bankroll, and squeeze as much EV out of your play as possible.

[/ QUOTE ]Isn't it possible that you can expect to be given better EV situations later?

You and I play heads up. First hand I show you AQ of clubs and go all-in. You hold 6d6s. Do you call? Calling has positive expectation, but probably not nearly as much as merely getting me to sit down at the table with you.

Anyway, I think there is an opportunity cost that is not factored into EV calculations.

[/ QUOTE ]

I can always play HU with you again. If not, I'd fold (assuming you're actually worse than me). However, in the case of SnGs when you're not playing above your BR, you can play as many as you want and therefore this should not come into play.