AK offsuit under the gun

Against tough players, AK must be worth less than half of the total of the blinds. This is obvious once we notice two things:


A. The EV IS the blinds when everyone folds.


B. Given tough opponents (who are putting you on a hand that is on average about what you have) almost every call or reraise reduces your EV from that of a blind steal. In other words if a tough opponnent offered to fold without looking, you should jump at the offer.


When you are played with (about 75% of the time), any player behind you, given his positional advantage, should on average at least split the blinds with you. Even a call from a blind cuts into your EV given he is a good palyer calling your UTG raise. Throw in those times AK runs into a big pair (which cuts AK's EV down to zero or much worse) and you can see that AK makes only a few bucks against tough players when it doesn't steal the blinds.


That being said, it is possible I slightly underestimated the hand especially against poorer players. However keep in mind that my reason for the quiz was to keep readers from even thinking about merely calling with high suited cards against weak players in late position. I'm not worried about anybody throwing away AK.

In your answer, you stated that AK was worth about 8 bucks UTG.


You also posted, "Remember that I assume a typical LA ring game, which is pretty weak and loose. If the game was tougher all four errors would be about equal with a negative EV of about five bucks."


Now, your posting, "Against tough players...... Given tough opponents", etc.

So tough that when you get played at its by one opponent that is about break even (as in rarely dominated for starters). This helps justify the AKo approximation which you now say may have been underestimated


Then, in the same post, "However keep in mind that my reason for the quiz was to keep readers from even thinking about merely calling with high suited cards against weak players in late position."


Which implies that the large EV loss you gave for the ATs being a result of weak players. These same weak players that when they come in against AKo UTG raiser they are about break even.


The original quiz stated that all the hands were from 15-30 LA game. This would generally mean that it was the SAME 15-30 game.


Or am I just really tired?

i like the perspective you took on this problem...conceptual blockbusting...gl

Angelina asked me about what I thought here. Before I saw David´s answer, I replied A,D,B,C (AK, ATs, QT, JT). I replied in private that off the top of my head AK was worth $15 to $25 UTG in a typical LA $15-$30 game.


David´s answer took me by surprise. $8 for AK? From his assumption of 75% chance of a caller behind stated in David´s answer, I can deduce that the players are playing 20.6% of their hands versus an early raise, or roughly 33/76s/98 and up. (As an aside, Angelina pointed out to me that Gary Carson's book seems to advise cold calling with 98o versus an early raise.) This means that AK is playing quite often versus weak/dominated hands against which AK has an excellent expected value, so AK´s value here is certainly in excess of $15. If we instead assume the players are very tight, then AK would steal the blinds very often, and so AK will still be worth a good portion of the blinds - probably a bit less than the big blind in the worst case.


-Abdul

"From his assumption of 75% chance of a caller behind stated in David´s answer, I can deduce that the players are playing 20.6% of their hands versus an early raise, or roughly 33/76s/98 and up."


in the 15-30 LA games ive played in this seems pretty accurate actually. and there's a strong schooling effect where if one player cold calls the rest fall in behind. it's lovely. really.


"This means that AK is playing quite often versus weak/dominated hands against which AK has an excellent expected value, so AK´s value here is certainly in excess of $15."


out of position against 4 players or more? really? you need to hit your hand on the flop. period. if you miss with AKo on the flop in a typical LA 15-30 then dont even bother wasting another bet. just check-fold.

The way I read David's post, he is saying that, against tough players, AK is worth less than half the blinds, "just a few bucks". He then gives an explanation as to why he thinks this is so. He then says that he may well have underestimated the value of AK against poorer players. I don't see any inconsistency here.


The point he is trying to emphasize is that you should raise with the suited high cards behind several weak limpers. Even if his figuring of the value of the AK UTG against weak players was more than "slightly understimated," it's not important since no one is going to fold the A-K. But many will not raise with the A-Ts and he is trying to show that this is at least as big a mistake as folding the A-K would be. This is also the point, I gather, for the other two hands; those plays, more "obviously" wrong than merely calling with the A-Ts, are not actually as wrong as not raising with the A-Ts.

Does Gary Carson's book actually advise cold-calling with 98o versus an early raise? Abdul used the phrase "seems to advise," so I'm quite curious and looking for a bit more clarification here.


Thanks.

better 98o than KJo.


I have Carson's book, and I like it a lot. I have skipped the pre-flop sections so far, however. They are not as interesting to me as the other stuff. I will look into that tonight.

I'm particular interested in your opinion of the EV on D (not raising with ATsuited on the button after 4 middle limpers). David says that it's MORE than -$10 (and he initially placed A=folding AK utg as -$8).


Do you agree that D is MORE negative EV than A? By a significant amount or marginal amount?


David rates B (calling with QTo UTG) at -$5 and C (Raising with JTo on button after 4 middle limpers) at -$2. Both B&C seem about right and fairly self-evident.

That's exactly what I was trying to say in the other thread. It's just not that great a hand with 5, 6, 7 way action out of position. Its still a good (+EV) hand, but I don't think its worth $25. I would rather have almost any playable suited hand when the fish are schooling (and I am out of position).


I think that AK under the gun will have the most value in a semi-tough, tight passive kind of game perhaps with a couple of players that play a little too loose before the flop (calling UTG raises with A-J and the like).


I haven't done any analysis to support this theory, other than playing a lot of poker.

Andy,


In order to assess the relative strengths of the situation, you have to use the same assumptions.


Sklansky's AKo answer appears underestimated more than a little. This places it as the worst mistake in the game described, IMO.


It really doesn't matter that much in reality, cuz like DS says, we are not going to fold it. But its an interesting theoretical question.


Now, if his answer on AK was wrong(and being a little wrong is no big deal), what makes you think the ATs isn't? (Im not saying it is or isn't...I don't have a problem with the ATs analysis because there are a lot of variables at work with 4 limpers).


Lets not lose sight of the idea that raiser after 4 limpers with a relatively stong hand is good for you.


Regards.


None of this diminishes Andy's reasoning and your willingness to part with the crowd on your answers.

OK, well, is he advocating the call in this spot with 98o, or merely saying it is not as bad as calling with KJo?

If we define that as a constraint for the problem set, then yeah limping with AKo is vulnerable. But if you raise utg and are getting 7-way action, you are in a game that really needs suited cards and pocket pairs. Most of the games I play in aren't like this, and I'd much rather be bringing AKo to the fight as opposed to say 77.


JG

From time to time I have encountered 10-20 and 15-25 games that are this crazy in Washington, and 15-30 and 20-40 games that were that crazy in the Bay area (of course usually the maniacs won't just call the raise, and the betting will be capped 6-7 ways). I don't play in LA, but I just assumed from all that I've heard, that the games were all like this there. Perhaps it isn't quite that sweet.

Before seeing David´s answers, I put the relative EV´s for misplaying at (negative) $15-$25, $10, $5, and $0-$3 for AK, ATs, QT, and JT, respectively. Angelina is skeptical that ATs is worth $10ish for raising here, so we are running some sims, stay tuned.


-Abdul

"out of position against 4 players or more? You need to hit your hand on the flop. period. if you miss with AKo on the flop in a typical LA 15-30 then dont even bother wasting another bet. just check-fold."


This is good to hear. I play california $3-6 and play like you advise, but I often feel like a puss for dumping after i check and there's a bet and 5 calls.


Question: You specify A-K offsuit, but if it were A-K suited, and flop has one of my suit, is it then okay to take off a card like I will often do for one small bet(with a bunch of callers & a preflop raise)?

> Against tough players, AK must be worth less than half

> of the total of the blinds. This is obvious once we notice

> two things:

>

> A. The EV IS the blinds when everyone folds.

>

> B. Given tough opponents (who are putting you on a hand

> that is on average about what you have) almost every call

> or reraise reduces your EV from that of a blind steal. In

> other words if a tough opponnent offered to fold without

> looking, you should jump at the offer.


Are you claiming that, against very skilled opponents, there should be no hand in which the value is greater than the blinds, since those players always have the option of simply giving up the blinds?


Sometimes I wonder about you, David.

If the game was played with all cards exposed then this would be true. Of course, it would be very boring as well.

a three flush, backdoor broadway, and if your call closes the action are all things to consider when deciding if you will call the flop bet.

The fact that good players will stay away from hands AK dominates and will outplay unpaired AK's with position when no ace or king flops.

I'm going to read Abdul's post's below and David's in response.

Your point here is a good one: if the answer on the A-K is more than a little wrong, then it would lead one to question if his answer on the A-Ts is also. But I think David's point was exactly as you put it here: raising with a relatively strong(suited) hand behind four (weak) limpers is indeed good for you and not raising could be a mistake of even greater consequence than the other more obvious mistakes.


Thanks for the words of encouragement in your last sentence. You'll note my "courage" was so strong that I entitled my original post on this subject "a dissenting GUESS." :-)


Regards,

Andy

Of course there are hands that are worth more than the blinds even against experts. Those are the hands that gain EV when they are called or raised by said experts. AK out of position, rarely does.

better 98o than KJo is my thoughts. i see many people in my games chuck stuff like 98o to a raise but then play KTo or QJo. not that i mind...


there is an instance where he lists the possibility of cold-calling an eary raiser with 98o on the button, but the text explains what he means by his groups. the way it is stated in the book makes it reasonable. I think he does a good job explaining his thoughts on this matter(and most others). i think its definately worthwhile for you to check out his book.

Huh? When they stay away, your AK takes the blinds earning $25. What happened to your estimate of 75% chance of getting played with?


Angelina Fekali

Studying People Inc.

Ljubljana, Slovenia

http://www.fekali.com/angelina

First of all, I agree with all of the other posters who are perplexed by DS going from a "typical LA ring game, which is weak and loose", to discussing the value of AK when up against extremely tough players. Just curious.... were these the same four extremely tough players who limped in the play of the ATs? And why aren't we comparing the value of the ATs in an extremely tough game? (other than the fact that that situation never comes up in a tough game)


So, let's say we humor DS and let him compare apples to oranges in an effort to make his point. I disagree with both of his values for the ATs and AK, but let's first work on the AK.


You claim that these extremely tough players are not going to call you with dominated hands. What distribution of hands would you like them to give you action? If they are only playing TT+ and AK, then they are only playing 2.6% of their non-blind hands. So it will be folded to the blinds approximately 85% of the time. The blinds are $25... you are certainly going to be able to get more than $8 from this situation.


Also, in an effort to minimize our ev of AK utg, you've made these players play so tightly that we now have positive ev for any holding. It sounds as though we are stealing the blinds on the order of 40% of the time, thanks to our opponents fear of our holding AK or better. I can make money with any two cards if I can win the blinds 40% of the time with a raise. Suddenly, QTo utg looks quite playable.


Your group of tough players is not that tough. They are a task force assigned to minimizing the value of a strong hand utg, with no regards to the fact that you might raise with weaker hands.


Your task force is said to play against our AK postflop with equally amazing insight. Again, if they are so quick to fold whenever an ace or king flops, I can play many more hands preflop from that position. Your postflop play specializes in breaking down AK, but at the cost of losing more to all other holdings.


You simply cannot design a flawed poker strategy for your opponents in an effort to get results that make your point.


I'm not certain of the change in value for raising the ATs. I do know the value I have gotten from playing AK utg in games much tougher than the LA game you describe, and that value is close to one small bet.

Everyone should keep in mind that the sims will probably be based on Turbo Hold 'em results. This is a program that I think we all agree plays absolutely terrible. As one poster put it if you can't crush TTH you have no chance in any casino poker game. Thus results from these sims which he seems to base so much of his theory on may be very inaccurate.

On the other hand, hand history analysis seems to corroborate the simulations results.


On the other other hand, I know a lot of people who say the same thing about "mathematics and poker" that Mason says simulations and poker".


On the other other other hand, those of us "in the know" know that simulations are just applied mathematics.


Kind of makes you wonder eh?


- Andrew

Mr. P:


I think it's a fact that TTH plays terrible. Thus certain sims should be way off. However, it doesn't mean that all sims will be way off. Even though it at times may be consistent with what you might think, or be consistent with other supporting data, it doesn't mean that it is reliable.


To your probable surprise we have in the past used "hot and cold" simulations in our work, particularly for seven-card stud. (See the appendix of such simulations in SCSFAP.) That's because in stud it is ofen correct to chase to the river. However, hold 'em plays so differently that we feel this type of sims has virtually no value. But one thing is for sure, in the hands of an expert who understands the game, he should have a pretty good idea what adjustments need to be made since he knows exactly what the sims are producing.


When using TTH you don't know that. Perhaps in this particular case they do a good job. Perhaps not. But you do know, based on the fact that even poor hold 'em players can easily crush the game, even when all the tough flags are turned on, and that some of the advisor advice is very poor, that it must be making some significant mistakes. (MR P: I'm sure that even someone who likes to be as contrary as you will agree.)


So when you see us use statistical analysis which you call "mathematics and poker" you are overlooking the fact that we know that what we are producing is either extremely accurate or at least have a very good understanding of the shortfalls and weaknesses of the mathematical model. With TTH you don't know this. (Well perhaps you do, or the people who produce it do, but I and certainly most of the posters on these forums don't.)


By the way, I want to point out that even with the above being said I think that TTH is an excellent product for someone relatively new to the game or for others who just want to have a little fun. But I strongly recommend that it not be used as a serious research tool even though it may at times give reasonable results.


Finally, if a future version of TTH does begin to play reasonable well, then its simulations results would have much more credibility in my opinion.

Mason writes:


"To your probable surprise we have in the past used "hot and cold" simulations in our work, particularly for seven-card stud. (See the appendix of such simulations in SCSFAP.)"


Gee, you make it sound as if I've never looked at that book at all. To your probable surprise, I own this book. In fact, I bought it with real money.


"So when you see us use statistical analysis which you call "mathematics and poker" you are overlooking the fact that we know that what we are producing is either extremely accurate or at least have a very good understanding of the shortfalls and weaknesses of the mathematical model. With TTH you don't know this. (Well perhaps you do, or the people who produce it do, but I and certainly most of the posters on these forums don't.)"


I find this hilarious, especially in light of DS estimating the worth of AK UTG as $8 in a 15/30 CA game. I also find it telling that you basically admit that you don't know what the strengths and weakness of TTH are. They are, for all practical purposes, just as useful as your "hot and cold" simulations, except that what fits between the quotes is a little more complex than "hot and cold".


It's nice that you finally admit that you don't really understand TTH, and how to properly use it.


- Andrew

"This is a program that I think we all agree plays absolutely terrible."


Nice argument technique. Now can you prove it Mr. Gra^H^H^HMalmuth.


"As one poster put it if you can't crush TTH you have no chance in any casino poker game."


Have you ever crushed the current version Mason?


"Thus results from these sims which he seems to base so much of his theory on may be very inaccurate."


Ah, the issue is that "his theory may be very inaccurage". This isn't a problem with TTH per se, but rather a problem with Abdul.


When reading a 2+2 book all the readers should remember that Mason Malmuth botches basic logic this bad.

Well Mike, since you have obviously played it extensively, why don't you tell us exactly how well it plays? I strongly suspect that you will admit that TTH doesn't play very well even though you'll insist (and I understand this to be true) that the current version is much better than it use to be.


Furthermore, I hope that some of our other posters who have used this program also give their opinions. And by the way, how about commenting on the playing advice that the advisor gives.


Best wishes,

Mason

The issue at hand are you assertions about TTH and how they reflect in Abdul's "theory". You should find some way to support your assertions and logic other then attempting to change the subject or attacking your opponent.


FWIW I havn't kibitzed against TTH since they released their latest advisor ("sid") so I can't comment on that. But in the past I did take one of your suggestions and watch my TTH opponents for errors. They were of the same class and number as I see in the casino in 6-12 through 15-30 games (I can't comment on higher limit games as I don't have enough experience and I stopped kibitzing against TTH some time ago).


This leads me to belive that somebody with a reasonable understanding of poker and simulation (as i believe Abdul has) can use TTH results to positive benefit. 2+2's continued assertion to the contrary without any current experience doesn't make a lot of sense to me (thus my request of a proof from the person making most of the assertions). Its _not_ correct logic to point to somebody deriving incorrect information from a TTH simulation as an indictment of TTH (the K3 thread).

I do not use TTHE nor am I familiar with the methodology for the sims, but you say that the sims converge to the results.


Does this really prove that the sims are accurate? It seems to me that all it proves is that the quality of player in the actual results roughly approximates the player input on TTHE. For a large group of tough players this assumption may not be at all accurate.


So, isnt it a fact that the quality of the players is the most important aspect of estimating the EV of a situation? If so, then if a good player does not play in a similar, self-weighting fashion, then the simulation is per se incorrect, or is it? At the very least it seems that it cannot be correct against a player who will not only vary his strategy but who will use counter strategies, and the sims may therefore be valuable for bad players, but not for tough players.


Of course I may be totally off base here due to my unfamiliarity with TTHE.


Pat

This leads me to belive that somebody with a reasonable understanding of poker and simulation (as i believe Abdul has) can use TTH results to positive benefit. 2+2's continued assertion to the contrary without any current experience doesn't make a lot of sense to me (thus my request of a proof from the person making most of the assertions). Its _not_ correct logic to point to somebody deriving incorrect information from a TTH simulation as an indictment of TTH (the K3 thread).


I agree with you that interpreting TTHE results is open to error. And it is not correct logic to indict TTHE for those who do draw false conclusions.


I am also aware of flaws in TTHE due to my extensive use of it, and MIGHT post something on this in the future, if I have the time, and inclination to document it.


Regarding the K3, that series of simulations was put through controls which helped minimize the errors that affect some of the other simulations.


Although K3 isn't the issue, Abdul for some time advocated that according to his website, although I know he has backed down from that, was a call against a steal raise. It still appears form his site that this is correct, he hasn't changed it. I could be misinterpreting it. It doesnt matter, the TTHE sim results support it, and if they support that, and are wrong, then what of the other sims?


From a RGP thread:"


joshg27 writes:


> Last night I was playing $10-20 holdem and had quite a few situations come up

> where I was in the big blind, facing an obvious steal raise (on different

> circumstances from either fairly solid or loose players), and holding

> marginal holdings. I found myself often unsure how to act.

>

> BTW, the game is raked.

>

> I don't have anything in front of me, but I seem to remember that Abdul

> recommends folding JT in the BB versus a middle position raise from a decent

> player (in a raked game). Also, if I recall correctly, the recommendations

> differ tremendously if the game is not raked - i.e. defending with much

> weaker hands.

>

> The raises which I was facing were from obvious steal positions. My question

> is what type of hands should I be defending with in the BB?


abdul "You've got odds to flop a pair or draw (since a draw often has semibluff

potential or pair outs), so call with almost any two cards. Play

hands with flush or straight potential and avoid dominated

offsuit disconnected hands such as Q5. A2 is okay, though,

as long as you are heads up against a stealer. Oh, except I forgot

you had a rake; you'll have to play a bit tighter, maybe

a minimum of 43s, A2s, A3, K2s, K9, Q5s, Q9, 76, 86, 22, etc."


--

Abdul


+++++++++++


I think this his updated advice, although it tends to change with the wind.


Notice his specific warning against big little cards like Q5, of which k3o is one.


But also notice his inclusion of k2s. Now, its not hard to see that with a bigger small blind, a smaller relative rake in a 30-60 game, the specific fulcrum I gave, and my caveat to avoid better players, that my conclusions were actually in line with ABdul's and in fact more specific, more thorough and less open to misapplication.


Would you rather play k2s against a "steal raise" in a 10-20, or K3 off in a game with less rake, more dead money, against a known hand range, and knowing you are the better player? I'll take the latter.


And all the worry about these weak kings is silly anyway, I can recall three occasions in the last eight months that I called with a weak king in that spot and I play shorthanded (with seat rental), live.


Mason's objections to TTHE would hold more water if he actually knew more specifics or could make an actual showing of error that applies to the sims. The human player game and the sims are to different components of TTHE.


Regards.

"So, isnt it a fact that the quality of the players is the most important aspect of estimating the EV of a situation?"


Sort of. It's the RELATIVE quality of the players that determines EV. This is why TTH is a lot better than some think. If you can get a lineup where the relative quality differences are close to correct, then you'll get ev results which are close to correct.


- Andrew

Like everyone else, I was disturbed by Sklansky secretly replacing all the LA 15-30 opponents with Vegas 80-160 pros. However, for this case of facing the very tough players I did not overlook the "fact that good players will stay away from hands AK dominates and will outplay unpaired AK's with position when no ace or king flops." I assumed the tough players would play with 99-AA, AKs/AK, and AQs versus our UTG raise. (The 99 is a bad play, but it hurts the AK slightly anyway.) I used S&M-approved no-fold'em simulations, fudged to account for position and the way AK plays against each hand. Based on this, my detailed and pessimistic calculations suggested an EV for AK UTG about equal to the big blind prior to deducting the amount the blind players take. I was unsure of how much to deduct for the players in the blinds - quite a bit if we assume our opponents know we have AK, but not much if we assume the players don't know we have AK and they are really as tough as you say (because they have to fold extremely often preflop and also on the flop and turn for fear we hold a big pair.)


I don't think there is a necessarily a disagreement for the worst case value of AK UTG - I say a bit less than than the big blind, and Sklansky could probably be talked into slightly more than 8/15's of the big blind. However, IMO, AK UTG normally will be worth the big blind or more in any games we face and especially $15-$30, and this is all that mattered for the quiz.


For the case of ATs, I suspect that Sklansky's number there comes from no fold'em simulations - the chance of ATs winning in no fold'em versus 5 random hands is considerably better than 1/6 - run the sim, do the calculation, and you'll get an EV for raising compared to calling of about $10, as I recall. I am amused that some think such a no fold'em simulation would be better than a pair of TTH simulations, in which the only thing changed is raising versus folding with ATs versus several limpers. However, it's true that the strength of the limping hands is not a whole lot more than the strength of random hands, since most of the strong hands are stripped out of the limping hands (since they would have raised with most strong hands.) With a TTH sim, at least you get some effects from the limpers tending to hold tens more so than aces. Any sort of sim is subject to flaws, misinterpretations, etc., so a sim is never the final answer on a subject.


In any case, what I think gets lost here sometimes is that the truth is the most important thing. How much better is it to raise rather than limp with ATs after several loose limpers? I would like to know the answer, even if Sklansky and I were wrong.


-Abdul

I do not think what you are saying is an accurate statement of the "official" 2+2 position, if there is such a thing. The problem with these simulations isnt that a good player cant use them well; I think that a good player certainly can.


The problem arises when people use the simulations to derive strategy without regard to the type of player or game, or without proper consideration of the types of hand that people will play. In my opinion I think this problem exists with many authors, and I see it in Gary Carsons book and Lee jones as well.


If you have success in your games (and I dont mean you specifically but the general "you") by deriving a strategy or following one from sims then by all means keep doing it. But for most people in most games this is likely a bad mistake. A good example is Carsons advice that you should not be looking to thin the field with high pairs because they make more money with large fields. If that works for you and you believe that a simulation that proves it is correct then more power to you. I will never be convinced that a strategy is a good one merely because a computer says it is, but that is just me.


Pat

Most of the players in the game condition this whole discussion is predicated upon 1) play terribly 2) have no chance in casino

poker.

"I do not think what you are saying is an accurate statement of the "official" 2+2 position, if there is such a thing. The problem with these simulations isnt that a good player cant use them well; I think that a good player certainly can."


Mason's post today about not using simulations in HE seems to contradict that. In any case I agree with what you said. Monte Carlo simulation is just another tool. Those that choose to use it correctly have another source of information to use at the table.