View Single Post
  #95  
Old 12-28-2005, 08:26 PM
ZeroPointMachine ZeroPointMachine is offline
Senior Member
 
Join Date: Jul 2005
Location: Las Vegas
Posts: 136
Default Re: $22: Push or wait with an uber short stack

[ QUOTE ]
[ QUOTE ]
I would argue that when your stack is really small this model over values it.

[/ QUOTE ]
Argue it then [img]/images/graemlins/smile.gif[/img]

To take a very extreme situation, you're HU with stacks of 7999/1. According to ICM, big stack has 50% equity and smallie has 30% equity. Overvalued? Of course not.

That's an extreme case, and you don't have to (and can't, in fact) argue that ICM overvalues the small stack. I am however interested to hear your argument as to what my math is missing here (and I really think there probably is something, I just don't know what it is).

[/ QUOTE ]

EDIT: All of the following is meant to apply to 5-8 handed situations such as in OP.

I think your calculations were correct (I can't really check them now), but I think the model breaks down here.

ICM assumes that everybody is of equal skill. But what does that really mean? It means every player will make the same average EV decisions.

When your stack gets this crippled you no longer have the opportunity due to lost FE and the limited number of hands you can play before being blinded out to pick your spots.

The rest of the table is now "out playing" you because they can make better +EV decisions than you can possibly make.

Your relative skill level at the table is now very low. ICM does not account for this.

Blind equity modeling actually got me thinking about another problem with short stack calculations.

If you turn on the blind equity modeling when you are UTG you get all kinds of push hands that were folds from UTG+1 because now you are calculating your fold EV based on the potential of losing your blind some percentage of the time next hand. Your position was better UTG+1, same stack, same cards, and the $EV of pushing was actually higher. But SNGPT says not to push because we're not UTG and it is not discounting the value of our stack based on paying the blind next hand and therefore calculates the value of folding as higher. The only thing you gain is the value of seeing one more hand. This is a real value, but it is small and hard to quantify.(If sombody can help here I would appreciate it)

If you can accept blind equity modeling affecting your EVfold from UTG(EVfold=EVutg-x%BB), wouldn't the next logical extention be that:

EVfold from UTG+1=EVutg-x%BB+(the value of one unseen hand)
EVfold from UTG+2=EVutg-x%BB+(the value of two unseen hands)

This is starting to ramble and I'm hungover at work. I'll let you chew on it for a little while and then see where we are at when I get home in a little while.
Reply With Quote