I came across wiarygodny's last month post which I thought was awfully interesting and since this is a topic I have also encountered recently, I decided to create a follow up post to add my 2 cents and allow people to keep developing the subject.
Surprisingly, this topic came up in one of the SnG coachings 2 weeks ago in a very interesting situation:
Blinds 25/50
BU = 75 chips
SB = 600 chips
BB = 600 chips (Hero)
So BU folds, and SB shoves into us. Assuming SB and BB are regs and playing nash, the equilibrium for this situation gives the following calling range:
Then the coach said, that he would call wider than nash in this situation, and for instance make a -ev call with 88. This came as a bit of a shock, at least for me and some other good regs present in the coaching. Since the BU has only 1.5BB left and is getting blinded out in the next hand, why in hell would we call wider than nash in this spot?
However I asked the coach if the reason was due to Future Game, and the fact that in we win we would almost always get 1st place, but his answer was no, that those were not the reasons, at least the main ones. However due to the limited time of the coachings and the other handful of questions that were going on, the topic was left up in the air.
Now, I am inclined to believe the coach statements, since he has way more experience than we do, however we were not able to find a proper justification for this call. On top of that, if we run a FGS on the same situation, 88 is still not even close to a call (-0.77).
So to put it simple, the questions are:
- In this situation, Call or Fold, and your reasons (all of them :))
- Does really a FGS make sense in this situation? I think in general this algorithms are presented to us in a 'black box' format and we just use it as a magic trick without knowing the mathematical limitations, i.e. in this case 3 of the variables are the 3 stack sizes, and one is of a lesser order of magnitude than the other 2. In mathematics this discrepancy can sometimes lead to spurious/meaningless results, since the algorithm is not as efficient as it is intended to be in this extreme cases. I am not saying this is the case, is just to give an example to question the reliability of this simulations. Maybe someone has some more in depth info on the FGS.
I think I am forgetting something, but for now I think it's enough to start the discussion. Thanks guys!

