Hi all. Hope the tables are treating you well.

The story of me beginning to grasp future game modeling is quite a long one. In it's earliest beginnings were the simple foundations of trying to understand hands that wiz could not calculate due to a 3bet of your raise and on from there in trying to understand multiway all-in pots. This was the foundation for the next steps for me. I have always been a nerd and very much into math, but I guess what I am trying to say here is that you don't really have to be a dork like me to get this stuff. But do your best to work on your foundation and next steps will come. You cannot build a house or a pyramid without foundation so you should always try to learn as much as you can. If you do, the opportunities to learn more in the future will come.

About 3 months ago I reread the Gigablock Theory and it finally made sense to me. I have trolled some threads in HSMTT where there was discussion as to whether or not it has a place in the end game of tourneys (MTTs in that context, but the larger principle is still there imo) and I would contend that it does. While the games have certainly changed since the time that Gigabet shoved the Q3 to everyone's dismay and amazement, the larger scheme of what is going on in the background is not simply stack equity (or ICM as we know it now). The things Gigabet described in a very insubstantial way stare at us every time when we play SNGs. I do not think I would have fully grasped the concept that Gigabet laid out with out some building blocks along the way. Situations like when to make a 'bad' call and why pot odds and steals are not always worth it were some of the underpinnings to this revelation of mine.

The problem is the blinds (both size and sequence), what the chips stacks are, and who possesses those chip stacks (most notably their tendencies), all effect the worth of our stack. In many situations certainly ICM represents the worth of our stack quite well. Additionally, if you put your opponents in situations where they misplay the situational, inevitably you gain equity that is not accurately represented by the current models.

So what is the solution? TBH, the short answer is there is no true solution. When the blinds go up, the imperfect informational nature of poker, and random occurrence all make a true solution impossible AFAIK. However there are ways that we can attempt to understand situations better, and use the tools we have analyze more accurately.

So, I present to you how to calculate a half-solution, Future Game Modeling. I suspect the computer tools we know and love (I mean hate, but they make us money so who is to argue) will eventually move in this direction, and it will be quite an interesting day if they do move this direction.


A hand was posted a little while ago (Xposted from the beginners forum, tee-hee) that I used as my example for this. It is a simple 18m situation:

Now if we run this through SNG wiz, this is what we get:



Without edge, it suggests 55+, A7+, A3s+, K9+, K7s+, QT+, Q9s+, JTs. But wait! If we fold, BB flips, and if he doubles up we will be in the BB, he will be AI in the SB, and we will still cover him! Wiz is telling us to shove K9o, but should we? If not, why? Lets look a few steps out, of if hero calls and if hero folds.





A short blurb about my notations, underlined text over lines is a decision by someone, bolded numbers are hero's estimated equity, lines that end with arrows are outcomes that involve more future situations, lines that end with circles are a finite solution since HU or V1 wins can both be easily and accurately calculated. After a couple steps out we have used ICM to calculate equity for 3 handed situations. I know what you are saying, "but CB79, isn't this whole post saying ICM is wrong?" Yes it kind of is, but once the % chance of something happening is down to a reasonably low number, we can use ICM to estimate the value of the chip set up within an acceptable margin of error. Obviously the best solution would be to calculate each future game situation out to infinity, but that ain't happening on my time today.

So now we have figured out what the value of each situation is... let's figure out what the chances of things happening in each...



So, if these calculations are correct, not only is K4o a fold, it is a +.6% fold.

Ok, so let's look at the K9o hand, which was the bottom of the range for a shove in the ICM shove/fold solution.



Still a fold, a .28% advantage to folding.

Now before you think I am just making up rogged #'s and that this will tell you AA is a fold, let's look at 88.



Shove!

Ok, before the SNG Wiz guy flies down here to burn my house down, and everyone who has studied the Wiz to death for the last several years pay for the ticket, I think we have figured out the solution fairly accurately so that we can run the #'s through Wiz. If we look at the BE ICM shove situation, we notice that K9o is a -.28% shove with future simulations. If we put a +.28 edge into Wizard, we get the following...



I also would have liked to have looked out another event or two on the existing flowcharting, but this was long enough as it, and hopefully you have learned something. There were a few other situations I wanted to look at here but did not have the willpower and time to run like this... A couple good candidates IMO, are these:

http://forumserver.twoplustwo.com/36...l-offs-675783/

Either of these hands...
http://forumserver.twoplustwo.com/36...stions-720939/

And I am sure you can use your imagination for the rest.

I would also like to take this opportunity to thank those who have posted on these forums to keep the game progressing. If the games were not hard, we would not make any money. So hats off to posters like Little John, IFoldPktOnes, BeSerious, and Gigabet who have helped make this post possible. I'd also like to thank simpledude for helping me improve my game to the next level, Badr for msging me a endless stream of beats to remind me that it isn't just me who is getting sucked out on, Reasons for turning me into a winning player, GregDC and Yardle for generally shooting the **** and talking pocerz. There are too many good posters on the forum for me to mention with regards to the innumerable spots that I misplay, so thank you all too.

Hopefully there were not too many errors in here and it wasn't too hard to read. Now if you will excuse me, time to reply to some threads that were tilting the **** out of me while I was working on this. Hope you enjoyed and let me know your feedback please.

OMG OMG FISTR !!!

tl;dr

gonna read now ( : ! contratz on 3.5k

2nd, been looking forward to it. too long but will def read. Gratz on 3.5k buddy!!!!

Very well thought out, calculated, and presented. Well worth the time to read...and probably read multiple times until it sinks in and can be implemented to supplement one's end game strategic thinking.

A+ post as always, Castle. Cheers.

Congrats on 3.5K Mike!

wp'd sir. looking forward to rereading this many times, like greg suggests, until i get it

This is sick. Well played sir.

Grats on 3.5k man.

Sick post, been writing code all day so i'll have to read it again, head assplode.

-Hiway

nasty post

I assume you deliberately excluded the ante as it was enough of a headspin already...? And it makes no difference presumably.

If I've understood correctly, which I probably haven't, you've got SB calling 50% in your calculations if you shove. Is this right? [If you shove the big stack has to call here 95% of the time readless (100% is presumably optimal)]. But You also say that if hero shoves, assumed V1 calls 100% and there is a possiblity of V2 folding...? And if hero folds then V1 can fold?...Did you just ignore these brances of the tree for your calculations as they are impossible. I'm missing something.

I salute your calculations, but it's hard enough to approximate ICM in game, let alone this!

wow looks interresting as hell will read it later cause this is really tl;dr.

Congratz to 3.5k.

Wasnt it just the other day you reached 3k I think the fact you reached 3.5k so quickly is testament to your contribution to this forum.

This post is another exception contribution and really shows someone like me, who feels they do a lot of studying, just how much more examination and thought needs to be put into spots like this to maximise our edge.

Grats on the 3.5k posts of support and advice to many people across the forum. Now hurry up and make another 499 posts so we can see what you've got lined up next for 4k

I read it twice, and although it's pretty good, I really want to commend you on the visuals. The diagrams are nice. Looking forward to snacking on this in more depth.

Ante's are included.

Yeah I assumed he isn't folding. In the original hand, he is getting call of 300 to pot of 2000.

To clarify, I assume V1 is calling 50% in two future game spots where he is on the BTN and H/V2 are in the blinds. He should shove something like 90% assuming nash/icm are correct... which is an unfounded assumption.

V2 is AI in the BB in the first hand.

Yeah this is obviously close to impossible to figure out in game. If you told me there was a magic calculator that estimated your tEq and told you which ranges to shove 2 years ago, I would have thought you were crazy then too probably. I assure you that some of the principles becomes applicable over time. (I still suck at pushfold imo fwiw. Probably less than a lot of the regs I play with, but anywho...)

Well done Castle. Tackling such a difficult consept as future game modeling should definitley be commended. Getting a better understanding for what I always thought could only be estimated through common sense and experience can only be a help. While I'm light years away from being able to implement this into my own game I think that the general principle of what your approximating is definitely useful. Cheers man + congrats.