Quote:
Originally Posted by buffyslayer1
Hi
I am not sure if this is a bug or i doing something wrong or using icmizer in the wrong fashion?
bvb ran nash ranges for a spot see below
then when i click on calculate to show the individual hands in this range and the EV of each hands etc I get this
Which is a 2% difference to the original one (which I believe is the correct nash range)
The same happens if I calculate nash then look at charts - result. The range changes from 68% to 70%?
what is occurring here?
Hi
First I need to explain the difference in the two buttons.
1) Calculate Nash equilibrium finds an optimal nash ranges
approximation for all players. ICMIZER is using pure strategies for its Nash ranges approximations. This means that individual hands are either present or absent in the resulting ranges of Nash equilibrium approximation.
This also means that in certain "difficult" spots its hard to find stable solution using pure strategies (the easiest scenarios that come to mind are deep stacked 30+BBs situations).
2) Calculate fixes all opponent ranges and calculates
single precise range for specified hero position and opponent actions (on your screenshot: all fold to hero, we calculate open push range).
Since Nash ranges are an approximation, and a precise Nash solution pretty much never exists almost any range will make a small jump when tested with Calculate button.
I explain this effect in more detail in this article:
http://www.icmpoker.com/en/blog/anal...sh-fold-spots/
Other products that I am aware of are using mixed strategies, which allows to find closer approximations by playing some hands in the range only a certain % of the time. This leads to more stable Nash approximation, but has a feature of being very theoretical, since real world players do not follow complex mixed strategy on every position (because they simply cannot calculate this mixed strategy on the fly, and its impossible to remember).
The good thing is that hands in question (the jumping hands) usually are closest to 0% EV out of entire 169 basic hands. So in most cases it isn't a mistake to push or fold these hands, the mistake is to push hands which have significantly -EV or to fold hands which have significant +EV.
I hope it helps with this situation. It is indeed very complex, I get a lot of questions regarding this. The difference between mixed Nash equilibrium and pure nash equilibrium is hard to explain, and consequences from both approaches are not transparent.
Quote:
Originally Posted by buffyslayer1
Also have another question
Are there any plans to add weighting to hand combos? I think this is really important feature that's missing. When looking at or using the program to construct say 3b/calling ranges most people are playing a mixed strat now with some hands.
HRC has this feature and PIO solver pure preflop calculator does also which i am finding myself using more and more for more complex spots even though set up is a little more laborious.
thanks in advance
Hi. Right now our
release schedule plan has following big ICMIZER related features:
1) Support for Progressive Knockouts calculations using our advanced IKCM model. Coming later this week.
2) Nash equilibrium calculations for preflop raise, reraise, etc. spots. A major feature which is coming in the upcoming 1-2 months.
2) Support for weighted ranges, coming in the upcoming 2-3 months. We will also probably add support for mixed strategies Nash equilibrium approximations with this addition.
We also have a mobile edition of SNG Coach launching within a month, which will allow to use our SNG Coach on smartphones and tablets.