Hello, could you clarify this spot that I'm trying to understand?
This is a $15 18p tournament turbo, 4 prize pools. There's 3 peole, two short stacks and I am the CL. I'm in BB with 38bbs and villain is in BT with 2.6bb. He shoves, and I have to pay 1.6bb for a total pot of 5.8bbs (26% pot odds). I'm always calling any2 here, but I decided to see what ICIMIZER would say.
1) When I run with no FGS, it gives villain 12% and hero 91% calling range.
2) When I run FGS1, it gives villain 31% and hero 100% calling range (this is, for me, the natural and correct response for both)
3) See what happens when I run with FGS5:


This is insane. 29% shove for villain and 21% calling range for hero.

This seems extremely wrong for me. I thought that ICM FGS would always be the best approach to study ICM spots. Could anyone clarify?

Thanks!!

What's the 3rd to 2nd prize jump (particularly relative to the 2nd to 1st jump)? If it's more linear than usual, could be FGS0 and FGS1 take into account the likelihood of said high relative prize jump when busting BTN balanced vs the relatively low risk to your ICM if calling and losing, but FGS5 requires both of them to post blinds twice and there are more scenarios where the "work is done for you" (or alternatively more hands in which you can abuse their ICM scenarios from the button). Not sure, though. It is odd. What are FGS2-FGS4? Is the big drop off when hitting FGS2 or FGS4?

If you do FGS1 and FGS4 you will notice the results are very similar, but other FGS numbers give very different results. Why is this? Because the position setup that the simulation ends on will have a huge impact on ranges. Ending the tournament and getting your ICM chop after you have played your button is much more profitable than doing so after you have played your big blind.

For example on FGS1 and FGS4 the SB will call the shove insanely tight. Why? Because these FGS numbers have the tournament ending after the SB has played it's button. So the SB only calls the shove 12% of the time. However change the FGS so that the SB doesn't finish on such a favorable situation and suddenly it will call much wider.

In order to avoid these distortions caused by each stack knowing on what position it will end the tournament, you need to keep it "random" by making sure the FGS plays until everyone goes back to their starting position. Thus all stacks are future position agnostic. In this case you want to use FGS3 which will have BU pushing 38%, SB calling 36% and BB calling 54%

Also:

If you fold now, look at the set-up next hand, it will be massively profitable for you. By prolonging the bubble, assuming everyone plays perfectly, youll keep on grinding your opponents down. Thats why the more FGS you put in the less calling% you should get.

Thanks for your answer! The prize jump is very linear. $25 for 4th, $50 for 3rd, $75 for 2nd and $100 for 1st
Then optimum response would be:
FGS0: 12% shove for villain, 91% call for hero
FGS1: 31% shove for villain, 100% call for hero
FGS2: 33% shove for villain, 44% call for hero
FGS3: 42% shove for villain, 37% call for hero
FGS4: 21% shove for villain, 55% call for hero
FGS5: 29% shove for villain, 21% call for hero

As we can see, FGS2~FGS4 keeps my calling range between 37~55%. And FGS5 goes down massively.

Thanks a lot for your answer. I understand and agree what you are saying. But I posted above all the scenarios and FGS1 differs significantly from FGS4. FGS1 makes me call with 100% of my range and FGS4, even if I adjust the range of villain to 31% shove, that would give me 73% calling range. I made another test, all the scenarios I adjusted the shoving range of button to 30% of range, SB calling range to 18%, and here we have the results for the BB:
FGS0: hero calls 100%
FGS1: hero calls 100%
FGS2: hero calls 42%
FGS3: hero calls 27%
FGS4: hero calls 72%
FGS5: hero calls 21%

Definitely makes sense to think that every 3 rounds we have the same position again, but even though as the others said, for every future game that we go, the calling range goes down. I run a deep scenario where I considered the blind increasing the next round and FGS4 (maximum possible). This would give BT pushing 5.4%, SB calling 3.5%, and BT calling 31%. Crazy stuff. But if I adjust the same shoving range to 30%, sb calling 18%, would give me again 100% calling range.


Thanks for your answer. I agree with you that there is this bubble effect that I can pressure with my massive stack, but in this case, folding would give BTN 2540 chips for free, when calling and losing will give him 1000 more chips (3530). For me, this 1000 chips would not make any difference in terms of my bubble pressure. The next round would be the same effect. The difference is that I'm giving him a chance to comeback easily to the tournament by folding. That's my sentiment but it is not what ICMIZER is showing me.

I'm very confused with which approach to use. Probably I am thinking too much in a simple spot. In general, I will keep calling any2 with every strong opponent that I play against and maybe fold some hands to a fish if the other player is a reg.

Thanks again everyone for answering!

I don't use ICMizer and also you will get different results depending on what ICM algorithm your program uses. It just goes to show that there isn't much point in agonizing over 0.1% decisions when the models themselves are far from that accurate. IMO in this spot it's more important to understand the general concept that the big stack has huge future ev and does not want to take slight chipev calls. I do believe the population in general is too splashy with big stacks and not splashy enough with medium and small stacks. Anything between 50% and 80% seems reasonable to me. Calling 100% gets very close to chipev but at the end of the day it's a very minor difference that isn't going to make or break your winrate.

Nice to hear it. Thanks for your answer!
Btw, do you plat sit go? What is your nickname?

Cheers! Have a nice day!