There's two schools of thought on multi-way all-in EV calculations. One is that you calculate winning percentages at the point those are known, giving you the long run EV of the exact spot you find yourself in. This was used by HM2. The other is that where a short stack is all-in, that information distorts the ranges / actions of the remaining players making the all-in EV calculation incorrect and so should not be calculated. This was used by PT4. HM3 went for the second, PT4 approach.
The second one is absolutely and utterly wrong but is the one HM3 adopted. They have promised to address it if enough people ask for the change but so few people understand \ care that they can't get the interest going to justify the work. Because, obviously, thinking it through for themselves and dealing with the logic is beyond them. (Sorry but I'm pretty irritated by this one as well as the absence of standard deviation info which were in HM2).
There's meant to be a thread somewhere on here where you can pressure them to include the correct way of working this out as an option.
HM3 uses the same equity code as PT4 now which is more accurate and less biased than the code we used for equity in HM2.Originally Posted by huaba
In a hand where a player is all-in but other players are not, net adjusted results are not calculated. When one player is all-in preflop and other players are not, then they usually have the option to act postflop and this means that sometimes one of them will fold. Having the option to fold* after another street is dealt means that in the long run there will be more strong hands which get to showdown and less weak hands. If we were to calculate net adjusted results in hands like this it would introduce a systematic bias (because there will be more known strong hands) so we exclude all hands like this from net adjusted calculations. You can read more about this here.
*Note: Attempting to calculate adjusted results when unknown but folded cards already called the all-in also introduce a bias into the results so HM3 does not make these calculations in any situation where a player calls the all-in and later folds (even if it's on the same street).
That thread is here - https://forums.holdemmanager.com/sho...d.php?t=542485
yeah that hm3 is calculating it like pt4 if we have unkown information is not new to me (also dont like this method). but in this spot as pointed out in my post we are all 4 preflop allin no action on later streets anymore. so there is no chance for anyone to fold postflop and yet still its shown as -12000 chips for ev value event tho we had 57% equity PREFLOP.
Last edited by huaba; 12-06-2020 at 11:59 AM.
It's cos of different stack sizes. If you were all 100bb, or whatever, then it'd calculate it but multi-way with different stack sizes and nope.
well if im not wrong it works perfectly fine 3 handed and it should not be too hard to calculate this with side pots since all information of hands is given
No bias is introduced because the adjusted results are simply what hero should expect his results to be if he won at the % given by his all-in equity at the point he was all-in and cards are known to be able to calculate that result. The presence of more, or less, strong hands is irrelevant. If this exact situation gets repeated ad infinitum, NOT 'this situation but with the possibility of unknown other hands in villain(s) range(s) which might lead to some systematic bias', but this exact situation is repeated through to the long run, what is the amount of money hero should expect to win - which can be compared to the money he actually won.
Introducing the idea of ranges into whether the all-in EV statistic has value is confusing the perspective of what that stat represents.
No, it's not to do with being 4-handed vs 3-handed. They know how to do it, they did it in HM2, it's a choice arising from being confused by a poker 'authority' who wrote some contextually flawed nonsense.
*edit: what he wrote wasn't nonsense - it's the application of this to what this stat represents that's flawed because it's taken out of context.
Last edited by Knowlestoo; 12-06-2020 at 02:58 PM.
i´m with you on what you saying but this has not really that much to do with my problem.
it does make a difference how many players are involved or it has some other reason.
cause in this spot in the picture below we are only 3 handed also with different stack sizes and therefore the EV values are calculated. as you can also see the AI Equity is correctly and i think its farely easy calculated with side pots or however.
but in my spot from the opening post we are 4handed all in also preflop like in this spot but there are no EV values calculated.
I said to myself: "I wonder if there was action after the short stack was all-in because if not then the all-in calculation will work." From the looks of the hand example you gave, that's the case. The all-in calculation works in the 3-handed example because you shoved, the short stack called then the other player called. This means there was no action that causes the supposed bias issue.
Posting Permissions
Reply With Quote