solved this myself

Originally Posted by yaqh
Originally Posted by yaqh
Originally Posted by Matthew Janda To Janda's quote, I have to say: "Getting good at poker is hard work. Hard work can be fun." Dunno how rude this post will be perceived as, but I think it needs to be said. The differences in ethos between Janda and Tipton are evident in their books, but concisely illustrated in the above quotes.

In my opinion, EHUNL stands head and shoulders above ANL in pretty much all relevant areas: writing style, depth, applicability for the target audience/customer, and raw content. I was joking around with a professional HU cash specialist friend of mine, and said the best thing ANL has going for it is the large 2+2 style playing cards that aid in speedreading the hand examples. At minimum, EHUNL will stand for years as a high quality book for improving one's HU game (in ways Moshman's book will be unable to, but Fritz's book might), and beyond that I think it will become indispensable primer for understanding the fundamentals of applying game theory to poker. Whether or not it is being marketed successfully and thus makes Tipton a lot of money is another matter, but I think I speak for a lot of grinders when that scared, small, shrill, defensive part of me says "I hope it bombs. Games don't need to be tougher."

Earlier in the thread someone said EHUNL will probably "force a rewrite of Janda's book", to which Mason responded by saying (paraphrasing) "no, Janda's book is about different stuff". Having read both books, it appears to me like Mason was just engaging in standard book promotion tactics, defending home turf on 2+2. The books cover similar concepts, albeit with extremely different approaches; I feel EHUNL does an infinitely better job, chiefly because of the author's different mindsets to the game, illustrated in the quotes at the top of this post.

For the record, I have nothing against Matthew: he seems like a great guy, and should be commended for writing his book. That sh* ain't easy. I've enjoyed a few of his videos on Cardrunners and sent him a PM expressing my excitement for his book before I received it (something I didn't do for EHUNL). I just figure since Tipton can't voice his opinion on these matters (I'm sure he knows how much better his book is), someone like me should probably do it.

Thanks again Will, for a great book. Here's to an even better, more successful Volume II. You've set the bar quite high.

^+1

?

He is referring to Mersenneary's HUSNG free ebook probably.

Yes, if SB raises the c-bet 9/(9+3.25) ~= 0.735 of the time, BB is indifferent between check-fold and cbet-fold on the flop.

What did you do exactly? What do you mean you plugged in some ranges, and how did you subsequently find the equilibrium? In the example in the book, 75s wasn't in the player's ranges at the end of preflop play.

Sounds like you figured this out yourself, but yea, we found that:

- BB is folding to the shove 28% of the time (i.e. he's calling 72%)
- BB is calling the shove with 19.23% of all hands

So, if his total flop betting range is X of all hands, then 0.72 * X = 19.23 and X = 26.7. (Some card removal related effects are neglected in this calculation.)

Ah, well we fixed the players' preflop ranges for the specific example we worked out, but here we're discussing the situation more generally. We're still working in the context of the decision tree in Fig. 4.3, but are allowing for the possibility that different preflop ranges and flops may end up making one player's range stronger or weaker than his opponent's. Sorry if that wan't clear.

Thanks for the kind words. Wish that post were the OP . TBH, though, I haven't read Matt's book, and I haven't been following its thread, so I don't know much about it.

Re-reading those quotes, though, another point worth making is that the application of game theory to poker, and it's development as something that's actually useful at the tables, is very much still in its infancy. There's tons of interesting, unanswered questions, much useful work left to be done, and there's lots of value to be gained by doing it. Most of what we know so far can probably be called basics, and it's hard to start doing new, advanced work on a topic without a very solid understanding of the basics.

yup, this

Btw, we updated the errata a couple days ago.

Ah, mers is called Fritz. Dunno, doesn't seem like a Fritz, but then, who does?

I found the examples at the end fascinating (despite the fact it force you to skip back and forward in the book ). For a while I was even reading one a night before I went to sleep. Sick life huh? If possible I'd love to see even more in the sequel, they were incredibly instructive.

Hey Will, quick Q.

In your equilibration exercise you assign a couple of ranges, and then the first strategy of the Villain.

From there you devise the most profitable strategy for the Hero given villain's strategy (and the board and the ranges you assigned each player ofc).

How did you do that part (calculating the most profitable strategy)? Using some software, or..?

Completely agree. The examples are very interesting. I am still trying to wrap my head around the concepts and put them to work in other situations. Here is a spot for which I'd like to hear your opinion.

River equity distribution looks like this



Hero is oop with only 40% equity overall. However, the very top of the ranges are close to symmetric and the equity advantage of villain comes mostly from weaker showdown value hands.
My guess for equilibrium river play is that hero probably still wants to bet the top of the range and bluffs accordingly. Probably slow play some of the stronger hands to somehow protect the checking range since he still has to check a lot and some of the hands in the checking range want to go to showdwon.

Cliffs: Hero weaker range overall, but more polarized. Is he better off having a leading range or checking his whole range?

Looking at the question in bold on page 113.

I started with this:

(x)(21) + (1-x)(18) = 19.5
x = 0.5

This frequency makes sense given the chart on page 95. I ran it in CREV and got a BB pushing range of 48.1%. I know there are some "close enoughs" going on in the CREV algorithm, so I'm guessing that's the reason for the 2% difference (and I'm assuming the correct frequency is exactly 50%). Not sure if other CREV users can verify I'm doing this correctly. I'm just getting familiar with the scripting feature.

Thanks.

I see this was previously discussed:

http://forumserver.twoplustwo.com/sh...&postcount=222

Hook me up with an advance copy of Volume II, and I'll have a detailed review (and whatever else you want) prepared to post in tandem with the launch date.

...just sayin'.

Glad to hear you enjoyed the examples. Yea, there were some trade-offs regarding their placement in the text. I wanted to give examples of river starting distributions at the beginning to motivate the discussion. But I didn't want to give full discussion of the computational solutions until after we'd done all the work to be able to make sense of them. So that's why the examples were split up between the beginning and the end of the chapter.

There will definitely be more computational examples in Vol2, and I've gone with the more usual placement this time, placing them more evenly throughout the text.

Yea, the theory behind those calculations is just the same as described in Section 2.2.1 (The Most Profitable Way to Play a Hand). See, for example, the 73o example on pg. 42. But of course doing that manually for every possible hand combo would be pretty tedious, so it was automated. The sofware I used isn't publicly available, but I know people have followed along with that section (and others) using CardrunnersEV.

Yea, generally degree of polarization very important for deciding whether or not the player will have a betting range, while overall average equity isn't very important. The thing that makes me think he might not have a leading range is that he actually doesn't have too many strong value hands -- the top of his distn drops off pretty quickly. Of course, effective stacks and such matter here too. Post the actual hand/ranges/sizes here, if you want, and I'll take a look at the soln.

Yup, your math is correct, and the post you quoted discusses some card removal effects which are responsible for the BB shoving range being a bit less than 50% of all starting hand combos. The actual number is something like 0.488 of starting hands, so it does look like your CREV result is still a bit off tho.

If I remember correctly CREV rounds to the nearest hand in its 'almost but not spot on GTO tools'
meaning going up/down with 12 combo's for offsuit
going up/down with 6 combo's for pocket pair
going up/down with 4 combo's for suited hand

If I still remember correctly Scylla made a vid how you can go on from this 'almost ... ' result to exact GTO result. Never did this as this accuracy is too much to remember & to use at the tables.

Cool thanks, here is the hand:

PokerStars Zoom No-Limit Hold'em, $2.00 BB (6 handed) - PokerStars Converter Tool from http://www.flopturnriver.com

Button ($107.70)
SB ($201)
Hero (BB) ($200)
UTG ($235.99)
MP ($239.24)
CO ($241.26)

Preflop: Hero is BB with 9, Q
3 folds, Button bets $5, 1 fold, Hero calls $3

Flop: ($11) Q, 3, 10 (2 players)
Hero checks, Button checks

Turn: ($11) 7 (2 players)
Hero bets $7, Button calls $7

River: ($25) 9 (2 players)
Hero ??


Here are the ranges that I used for the river distribution:

Hero: 2c2h,2d2h,2d2c,2s2h,2s2c,2s2d,3c3h,3d3h,3d3c,4h3h, 4c3c,4d3d,5h4h,5c4c,5d4d,5s4s,6h4h,6h5h,6c4c,6c5c, 6d4d,6d5d,6s4s,6s5s,7c7h,7s5s,7s6s,7s7h,7s7c,8h6h, 8c6c,8d6d,8s6s,8s7s,9h8h,9h8c,9h8d,9h8s,9c8h,9c8c, 9c8d,9c8s,9s7s,9s8h,9s8c,9s8d,9s8s,Th7h,Tc7c,Ts7s, Ts8s,Ts9s,Jh8h,Jh9h,Jh9c,Jh9s,Jc8c,Jc9h,Jc9c,Jc9s, Jd8d,Jd9h,Jd9c,Jd9s,Js7s,Js8s,Js9h,Js9c,Js9s,Qh5h, Qh6h,Qh7h,Qh8h,Qh9h,Qh9c,Qh9s,QhTc,QhTs,QhJc,QhJd, QhJs,Qc5c,Qc6c,Qc7c,Qc8c,Qc9h,Qc9c,Qc9s,QcTh,QcTs, QcJh,QcJd,QcJs,Qd5d,Qd6d,Qd8d,Qd9h,Qd9c,Qd9s,QdTh, QdTc,QdTs,QdJh,QdJc,QdJs,Kh2h,Kh3h,Kh4h,Kh5h,Kh6h, Kh8h,Kh9h,Kh9c,Kh9s,KhTc,KhTs,KhJc,KhJd,KhJs,Kc2c, Kc3c,Kc4c,Kc5c,Kc6c,Kc8c,Kc9h,Kc9c,Kc9s,KcTh,KcTs, KcJh,KcJd,KcJs,Kd2d,Kd3d,Kd4d,Kd5d,Kd6d,Kd8d,Kd9h, Kd9c,Kd9s,KdTh,KdTc,KdTs,KdJh,KdJc,KdJs,Ks2s,Ks4s, Ks5s,Ks6s,Ks7s,Ks8s,Ks9h,Ks9c,Ks9s,KsTh,KsTc,KsJh, KsJc,KsJd,Ah2h,Ah3h,Ah4h,Ah5h,Ah5c,Ah5d,Ah5s,Ah6h, Ah6c,Ah6d,Ah6s,Ah8h,Ah8c,Ah8d,Ah8s,Ah9c,Ah9s,AhTc, AhTs,Ac2c,Ac3c,Ac4c,Ac5h,Ac5c,Ac5d,Ac5s,Ac6h,Ac6c, Ac6d,Ac6s,Ac8h,Ac8c,Ac8d,Ac8s,Ac9h,Ac9s,AcTh,AcTs, Ad2d,Ad3d,Ad4d,Ad5h,Ad5c,Ad5d,Ad5s,Ad6h,Ad6c,Ad6d, Ad6s,Ad8h,Ad8c,Ad8d,Ad8s,Ad9h,Ad9c,Ad9s,AdTh,AdTc, AdTs,As2s,As4s,As5h,As5c,As5d,As5s,As6h,As6c,As6d, As6s,As7s,As8h,As8c,As8d,As8s,As9h,As9c,AsTh,AsTc

Villain: 3d2d,4d2d,4d3d,5d3d,5d4d,6d4d,6d5d,7h5h,7h5c,7h5d, 7h5s,7h6h,7h6c,7h6d,7h6s,7c5h,7c5c,7c5d,7c5s,7c6h, 7c6c,7c6d,7c6s,7c7h,7s5h,7s5c,7s5d,7s6h,7s6c,7s6d, 7s7h,7s7c,8h7h,8h7c,8h7s,8c7h,8c7c,8c7s,8c8h,8d5d, 8d6d,8d7h,8d7c,8d7s,8d8h,8d8c,8s7h,8s7c,8s8h,8s8c, 8s8d,9h7h,9h7c,9h7s,9c7h,9c7c,9c7s,9c9h,9s7h,9s7c, 9s9h,9s9c,Th5h,Th6h,Th7h,Th8h,Th8c,Th8d,Th8s,Th9h, Th9c,Th9s,Tc5c,Tc6c,Tc7c,Tc8h,Tc8c,Tc8d,Tc8s,Tc9h, Tc9c,Tc9s,Ts8h,Ts8c,Ts8d,Ts9h,Ts9c,Jh7h,JhTh,JhTc, JhTs,Jc7c,JcTh,JcTc,JcTs,JcJh,Jd2d,Jd3d,Jd4d,Jd5d, Jd6d,JdTh,JdTc,JdTs,JdJh,JdJc,JsTh,JsTc,JsJh,JsJc, JsJd,Qh2h,Qh4h,Qh5h,Qh6h,Qh7h,Qh7c,Qh7s,Qh8h,Qh8c, Qh8d,Qh8s,Qh9h,Qh9c,Qh9s,QhJh,QhJc,QhJd,QhJs,Qc2c, Qc4c,Qc5c,Qc6c,Qc7h,Qc7c,Qc7s,Qc8h,Qc8c,Qc8d,Qc8s, Qc9h,Qc9c,Qc9s,QcJh,QcJc,QcJd,QcJs,Qd2d,Qd4d,Qd5d, Qd6d,Qd7h,Qd7c,Qd7s,Qd8h,Qd8c,Qd8d,Qd8s,Qd9h,Qd9c, Qd9s,QdJh,QdJc,QdJd,QdJs,Kh7h,Kh7c,Kh7s,KhTh,KhTc, KhTs,Kc7h,Kc7c,Kc7s,KcTh,KcTc,KcTs,Kd2d,Kd3d,Kd4d, Kd5d,Kd6d,Kd7h,Kd7c,Kd7s,Kd8d,KdTh,KdTc,KdTs,Ks7h, Ks7c,KsTh,KsTc,AhTh,AhTc,AhTs,AcTh,AcTc,AcTs,Ad2d, Ad3d,Ad4d,Ad5d,Ad6d,Ad8d,AdTh,AdTc,AdTs,As7h,As7c, AsTh,AsTc,AsQh,AsQc,AsQd,AsAh,AsAc,AsAd

Sorry for spelling out the ranges like this, but there is no way around it in this example. The ranges are just a pretty rough estimate and very pessimistic for hero. Checking back ranges on flops like that are very villain dependent, but normally I do not think they check back most A- and K-high backdoor flushdraws like I let him here. Nevertheless, it is just an example and a solution will certainly be instructive.

Btw, I did some quick CREV calculations and the EV between having a leading range and checking everything seemed close. I did not give villain the opportunity to make large overbets though...

Random thought:
Capped high vs capped low --- i think the way most people understand capped low, is what EHUNL explains as "bounded below"


Some Errata:

Pg 298
"Making the 12bb betting option to the BB does not however make an especially large difference in value of the game for the SB - about half a BB per 100 hands."
-Instance of BB is probably SB...because using BB would imply leading...


__________________
Pg200:
sentence that starts with "on a board like this one, however..." is worded strangely.
Probably some punctuation between the words "off" and "betting"

_______________
Pg 322:
diagram on 322 is mislabelled in the description. Ad should read 3h.




Question 1:


In the discussion on blockbetting, you mention that blockbetting hands are often trying to decide between
c/f, block-fold
c/c, block-call, bet/fold

I assume a full solution (with appropriately non-polar distributions) would include hand segments that represent all of these things, but I can't see an order anywhere in the book.

Pg 257 has a solution structure for the bet-or-check game.
Pg 290 illustrates the structures SB holds when facing blocks, checks, and bets.

I'm trying to construct good blocking ranges in HUNL and realized there wasn't a clear "right" way to order the segments. Is there a clear order for these cutoff hands in a GTO strategy that includes all possible actions?

Question 2:

I looked at the solution to the SB bet or check game here. I understand everything but how you represented the equity distributions. It looks like a long vector of equities.

How do I reproduce something like this, for custom distributions?

What I'd like to do is combine what was done in the "SB bet-or-check game" (with asymmetric distributions), and the solution to the "big river game" (with symmetric distributions), to start solving real spots.

Thank you,
Semp

Two questions for Will. Both about Hand distributions. One suggestion for Will, and one for everyone else.

Question 1) Page 135, hand distributions, and your question about when each 3betting range might be appropriate. I would like your input on the nuances and a few "whys" which I am unsure about.
Specifically, choosing a) over b) over c) over e) - and what makes us adjust in these ways. (Especially b over e) )

a) A wide 3betting range, containing broadways only. Villain is opening 100% and we 3bet that because we expect him to flat a lot of our 3bets. We want value.

b) We 3bet a wider range, but only half the time (netting about the same frequency as a).
This one puzzles me. Villain is opening again 100% (we're 3betting wide again). His folding frequency, or postflop plays in SRPs, or 4betting frequency are factors which I cannot seem to place anywhere in this example. Getting tricky?

c) A standard polarized range. This makes sense from a GTO point of view. Our value hands are balanced with our bluffs making it tough for villain to call correctly or fold correctly without either paying us off or folding a better hand. If he 4bets a condensed range, we can 5bet what's good and fold the rest (and 5betbluff occasionally, again keeping a balanced range).
This one I think is best out of all the listed ranges specifically if Villain likes to 4bet a lot (and calls 3bets sometimes).

d) Our villain is opening very wide, 100% or close to that. He seems to be folding every time we 3bet him. We don't want to waste premiums to see him fold; we think we'll get more money if we flat them.

e) We 3bet the same range as in a), but without the premiums.
Things that might make me want to do that -
Villain is opening 100% (our 3betting range is quite wide); he is calling a lot and he is not 4betting much. So we like to get value.
He also tends to get crazy in SRPs, trying to barrell us off of anything. That would be a reason to include premiums in our preflop calling range.

Am I on the right track? What would you add, or modify in the reasonings above? And what's with that b) thing?


Question 2) Page 137, percentiles of an ATC range on 4 different flops. I ran these in Flopzilla and some of them did not match the places where you placed them.
Flop 222 - J8 is perc. 56
Flop J53 - K9 is perc 56 (A high would be perc 50 there)
Flop KJ7 - deuces are top40% (again, A high would be perc 50)

Where does the discrepancy come from?



A suggestion - is there a place where all the blurbs (as you called them) are answered? I have never seen a workbook without solutions available in some way.
Do you know of gruops of players working on these things?
Do you have the solutions written up and uploaded somewhere?
I have to bug you here about these if they aren't anywhere...




To everyone else - if you study poker in order to improve, if you like Tipton's and Janda's work and would like to bounce ideas (like the ones I've typed above about the 3bet ranges) off of some other players, ship me a PM and we can help each other out. I am playing HU cash, small stakes. I am beating those, I want to improve, and I intend to move up.



Eagle

Hey, guys. I tried the pre-flop shove/fold equation and I would like to know if I understood it right. Can someone please tell me if it´s right or wrong?

EFFECTIVE STACK - 16BB
OPPONENT´S CALLING RANGE - 25%
OUR EQUITY vs OPPONENT´S CALLING RANGE - 36,27%
TOTAL POT (AFTER VILLAIN´S CALL) - 32BB

EV = % VILLAIN FOLDS * (EFF. STACK + 1 BB) + % VILLAIN CALLS * (TOTAL POT AFTER VILLAIN´S CALL * OUR EQUITY)

EV = 0,75 * 17 + 0,25 * (32 * 0,3627)

EV = 0,75 * 17 + 0,25 * 11,6064

EV = 12,75 + 2,9016

EV = 15,6516

So, in this case, as 15,65 > 15,5 BB, shoving is 0,15 BB +EV, is that it?

Thanks

^ That all looks right, up to card removal effects. So, jamming is better than open-folding, but your other options might be even better, ofc.

Sorry I'm behind responding to the other questions. I'll get to them.

Thank you for your help.