Just Calling with AK on the Big Blind

[PDosterM]

Recently I tapped my option with AK in the big blind against 5 limpers. I wound up winning a nice pot, but after the game, someone I respect said he thought not raising was a poor play – being aggressive here and betting for value was what he advocated.

Normally I do this against a smaller field, and if there is a preflop raise – particularly a late one – I will generally pop it again. Here I reasoned that against 5 opponents, not raising was in order. My rationale is 1) AK is a drawing hand – a good one admittedly, but it still needs help on the flop, 2) If I don’t connect with the flop, I’m not going to bet into five opponents, and will almost certainly fold if someone bets, and 3) it’s nice and deceptive. Actually the deception is what got me the pot. I wound up with Broadway, and K9 didn’t put me on AK until way too late.

One of the things I am working on is to appropriately increase my aggression – I’m tight enough, but I have to fight my natural tendency to be passive.

In your opinion, is my logic correct? Or is this an appropriate place to be more aggressive?

[James282]

Hey, I would raise because you make money on every $ that goes into the pot against most limpers here, and you take control of the pot. Assuming that no one has AA or KK, you will win the pot well more than your fair share by the end to make this a profitable raise. Also, deception isn't as useful when against 5 opponents.

Understand that there are currently 6 people in the pot, so in order to show a profit on money put into the pot before the flop, you need to be expected to win more than 16.6% of the time. I did a quick simulation on twodimes.net against hands that decent people would limp in with(although in most games, this simulation gives them FAR too much credit, but it should give you an idea). I included K9o since it was in this particular hand and I tried not to have your opponents forfeiting eachothers outs too much.

Kc Ad .232
7s 7h .212
6d 5d .205
Ks 9c .049
Qh Jh .240
9s Jc .063

As you can see, your EV of about .232 puts you in a position where you like every penny that goes in against these opponents increases your expected value more than your fair share(which would be .167). Combine other things like the fact that your raise will have opponents incorrectly fold if they make a middle pair when you flop an A because they "know" you have an A and I think the raise becomes a good one.

This raise has even more value against opponents who will enter with miscellaneous crap, and not legitimate hands like the ones I speculate. If you are playing against loose-passive opponents(which it seems if 5 limpers is typical) then your raise will often show value, regardless of whether your hand is a "drawing hand" or not.
-James

[astroglide]

i never call with ace king. i do not expect to make 5 small bets because of "deception" in your situation.

[Dynasty]

[ QUOTE ]
1) AK is a drawing hand - a good one admittedly, but it still needs help on the flop

[/ QUOTE ]

AK is not a drawing hand. AK is likely the best hand dealt out. All the other hands are drawing against AK.

[ QUOTE ]
2) If I don’t connect with the flop...will almost certainly fold if someone bets

[/ QUOTE ]

That's far too weak.

[ QUOTE ]
3) it’s nice and deceptive.

[/ QUOTE ]

I'd rather make money than be deceptive.

[deacsoft]

I like to raise with it from the BB, but checking it every once in a great while is not such a bad play. It should be raised because against a hand full of limpers it is almost certainly the best hand.

[SinCityGuy]

I hope that it wasn't suited.

As Mr. Sklansky said, failing to raise AKs from any position against limpers is a crime against humanity.

[GuyOnTilt]

i do not expect to make 5 small bets because of "deception" in your situation.

You do realize that he doesn't have 100% pot equity preflop, right? I'm not saying it's right, but he doesn't need to make up 5 SB's postflop in deception because he's not gaining 5 SB's from a raise.

GoT

[beachbum]

How about checking AK in the BB preflop and check raising the flop if you hit the ace or king? That way you knock out gutshot straight draws, backdoor flush draws, middle pair, etc. or at least let them make a mistake on the flop by cold calling 2 bets when the pot is now not big enough since you haven't raised preflop.

I guess the bottom line is with specifically 5 limpers is there +EV in gaining 5 SB preflop by raising, as opposed to losing an occasional pot to someone who drew out on you who otherwise have folded to cold calling 2 bets on the flop? Thoughts?

Of course, with AKs, you raise preflop without a 2nd thought.

When i first saw this post i dismissed it, of course you raise... especially since the 2+2er's here have drilled the pot equity idea into my head.

Then i was reading HEFAP (again /images/graemlins/cool.gif) and found this in the loose games section.

Page 168 Looking at some odds, the last paragraph and the very last sentence.

On the other hand, if you are in a loose, passive game where they usually call, but only occasionally raise, you should play any Axs UTG. You should also be playing a hand like J /images/graemlins/spade.gif9 /images/graemlins/spade.gif UTG, and anything better. You should play these hands because you are going to win a lot when you hit them. That is, you take advantage of bad play. You would also play all pairs. Conversely, you often shouldn't raise with your AK or AQ in spots where you would raise in tougher, tighter games.

I didn't put this here cause i agree with PDosterM, but because i don't understand why you wouldn't raise. Can someone explain the logic behind this sentence?

[LetsRock]

FWIW, I play AK/AQ from the blinds as you did.

I know, it's not the "prefered" method and it strays from THFAP advice, but it's what has worked for best for me. I also have to fight passive urges and I'm working on it. (It seems like every time I try to get aggressive, my timing is off and I end up throwing money away /images/graemlins/crazy.gif) But I don't think you lose any bets by limping with AK from EP.

If you hit TPTk on the flop, you'll usually get plenty of action by betting out - nobody will put you on that strong of a hand because you're expected to raise it. You'll often get raised by someone with a weaker Ace (or K) and then you can 3-bet since your hand has been "made" (to a much better degree).

If you miss the flop it's much easier to release the hand since you don't feel the need to represent a big PP. True, sometimes you can take down a pot on a missed flop just by betting, but I rarely find that you can get away with bluffing your way through a hand with AK without betting every street - this can be a very expensive proposition.

I'm sure I'll get flamed (I always do when expressing this opinion) and I'm not advising that my style is correct; just offering my view on the subject.

[deacsoft]

If you are going to limp in; you should almost always check-raise the flop if you hit it. Just be careful if there is an obvious straight or flush draw out there. You don't want to give any free cards.

[squiffy]

I guess the conventional wisdom is that if you hold AK 10 million times and raise every time, or if you hold AK and call 10 million times, you will, on margin, make more money by raising with what is likely the best hand going in.

[squiffy]

You know, Doyle Brunson repeatedly calls AK a drawing hand in his book. And if the average winning hand is top pair or two pair, AK has to improve, on average, to win.

Even if it is ahead preflop, against 5 opponents playing trash, if AK doesn't pair up by the river, it's not likely to win.

You are focusing on semantics. If you isolate the preflop advantage of AK, then fine, it is likely strongest and other hands need to improve.

But there are 5 community cards to come.

Brunson distinguishes AK from AA and KK and QQ, big pairs, which are more likely to win without improvement. Whereas AK is very unlikely to win without improvement by the river.

[PDosterM]

First, I thank everyone for their input – I really like this forum.

James282 makes a very good point. Essentially I am being offered odds of 5 to 1 when the correct odds are 4 to 1. That’s a 20% edge, and you could not be faulted for taking that bet (i.e. raising) all day long. James282’s use of twodimes.net to establish a 23% win rate is useful, and I would like to use that figure to respond to Astroglide and explore this issue a bit further.

Astroglide says [ QUOTE ]
I never call with ace king. I do not expect to make 5 small bets because of "deception" in your situation.

[/ QUOTE ]

I agree that it would be fantasy to expect the deceptive power of not raising to earn 5 small bets each time the hand wins, but the bar is actually not set that high. Consider 100 hands where the big blind raises with AK. Everything else being equal, the 23 winning hands will earn an extra five small bets each (thank you James282) for a total of 115 small bets. But there is also a cost in that the 77 losing hands each lose one extra small bet for a total of 77 small bets. The net effect by raising compared to not raising is a gain of 38 small bets per 100 hands.

That’s still a good argument for raising. But please notice that if by using deception each winning hand makes up (on average) 2 small bets, you gain back 23x2 = 46 small bets of the 38 small bets lost by not raising. That produces a slightly positive EV. (1.65 small bets is actually the breakeven point.)

So you don’t have to make up 5 small bets (on average) each time you win, just two. Now the question becomes, is this a reasonable thing to expect?

About a third of the time, an A or K will appear on the flop. In these cases, it is often reasonable to check-raise the flop, or in ideal situations (like K-rag-rag-rainbow), check-call the flop and check-raise the turn. Also, as LetsRock points out, someone who limped in with AT will give you plenty of action that he wouldn’t have if he were raised preflop. So the deception obviously has some value, but it’s not clear exactly how much.

I am curious if anyone would change their advice given that two small bets are all that are required to be gained by deception.

[astroglide]

it's money that he collects every time he wins the hand

[astroglide]

you mean the book written in 1979?

[nykenny]

[ QUOTE ]
Recently I tapped my option with AK in the big blind against 5 limpers. I wound up winning a nice pot, but after the game, someone I respect said he thought not raising was a poor play – being aggressive here and betting for value was what he advocated.

Normally I do this against a smaller field, and if there is a preflop raise – particularly a late one – I will generally pop it again. Here I reasoned that against 5 opponents, not raising was in order. My rationale is 1) AK is a drawing hand – a good one admittedly, but it still needs help on the flop, 2) If I don’t connect with the flop, I’m not going to bet into five opponents, and will almost certainly fold if someone bets, and 3) it’s nice and deceptive. Actually the deception is what got me the pot. I wound up with Broadway, and K9 didn’t put me on AK until way too late.

One of the things I am working on is to appropriately increase my aggression – I’m tight enough, but I have to fight my natural tendency to be passive.

In your opinion, is my logic correct? Or is this an appropriate place to be more aggressive?


[/ QUOTE ]

u did the right thing. don't let the "someone you respect" fool you. well unless you had AKs then you should have raised, hehehe.

now am i someone you should respect? /images/graemlins/laugh.gif

Kenny

[nykenny]

i agree with ur other points, except

[ QUOTE ]
by the river

[/ QUOTE ]

because i think if one always take AK to the river, he will lose money. am i wrong?

Kenny

[nykenny]

[ QUOTE ]
i never call with ace king. i do not expect to make 5 small bets because of "deception" in your situation.

[/ QUOTE ]

you never? i never never. my kung fu is stronger than yours! /images/graemlins/grin.gif

Kenny

[astroglide]

i could care less about what people think about "always" and "never". i always try to do what i believe is the most profitable, and i think checking in the big blind here is never correct. i'm not trying to win a prize for being cool with my enviable chameleon-like adaptive preflop strategies, i'm trying to win chips.

[pokeryogi]

Let me try this one...
You don't raise with AKo or AQo in BB of a loose passive game, because your making the pot big enough (cuz they will all call) to make it more correct for your thin drawing calling stations to call subsequent bets. You want them in position to incorrectly call.

Still learning,
PY

[Tosh]

AK is only a drawing hand against a pair. Say your opponents hold JT, 78s, Axs and KT in this position. They are drawing to beat you not the other way around. I don't raise AJo with these limpers but AK and AQ certainly.

[nykenny]

[ QUOTE ]
i could care less about what people think about "always" and "never". i always try to do what i believe is the most profitable, and i think checking in the big blind here is never correct. i'm not trying to win a prize for being cool with my enviable chameleon-like adaptive preflop strategies, i'm trying to win chips.

[/ QUOTE ]

hahah, whatever works, brother. i respect u enough to say that you make a very good point. i will try it some times.

but as "cool" as i am, i standby my blieve of "never say never". u sure u get more +EV? /images/graemlins/wink.gif

Kenny

[James282]

Bad players will call with ubiquitous draws regardless of whether or not you raise preflop or not. Can you make up 2 SBs with deception postflop? Possibly. In certain situations you are also much further ahead than in this example and in certain situations you are not quite as ahead. The problem with planning for a check-raise is that it only works if people cooperate. What if it's a bet and a raise by the time it gets to you? Raising lets you take control and start kicking ass. It only depends on you. If you have a very good grip on your opponents you can try to be deceptive, but I think that the majority of the time you will be ahead enough that you might as well raise preflop.

On an important note, this raise becomes more correct the worse the players are. This is because you will be even further ahead in a lot of examples because your opponents won't need what 2+2 books consider playable hands. The better the players are, the more deception will be useful. You don't need to deceive stupid players. Something to think about.
-James

[Schneids]

[ QUOTE ]
You don't need to deceive stupid players.

[/ QUOTE ]

Let's get out our highlighters.

[Nate tha' Great]

[ QUOTE ]
On an important note, this raise becomes more correct the worse the players are. This is because you will be even further ahead in a lot of examples because your opponents won't need what 2+2 books consider playable hands. The better the players are, the more deception will be useful. You don't need to deceive stupid players. Something to think about.

[/ QUOTE ]

I don't necessarily know that that's true. To provide two somewhat contrived examples:

1) http://www.twodimes.net/poker/?g=h&b=&d=&h=Ac+Kd%0D%0AAd+Jh%0D%0AKs+Ts

2) http://www.twodimes.net/poker/?g=h&b=&d=&h=Ac+Kd%0D%0AAd+3c%0D%0ATc+8s%0D%0A9s+2 s

One reason that AK is so powerful is because it dominates other "decent" Ax and Kx hands; against very bad players who will come in with much less than that it will get outdrawn more often.

[gunboat]

From twodimes

[ QUOTE ]

Kc Ad .232
7s 7h .212
6d 5d .205
Ks 9c .049
Qh Jh .240
9s Jc .063


[/ QUOTE ]

twoddimes is a great resource. I use it myself. However, it is only a beginning of the analysis. From the above it is clear you want to get as much money into the pot - IF the action were to stop right there. Unfortunately, the action does not stop there.

There are 3 more betting rounds and you are out of position for the rest of the hand.

Now we all know that late position = +EV and early position = -EV. A complete analysis would include the -EV from being EP for the rest of the hand. Does the -EV from position overcome the +EV of the hand? My gut feel is that this is one of those decisions which is so close that either choice can be considered correct.

That said, you should raise about half the time so the more intelligent opponents will not always put you on big PP when you raise in the BB. Same for AKs, AQu, AQs which I would also put in the category of a very close decision to raise or check. The only reason for choosing one over the other is to vary your play.

[James282]

Ax and Kx are crappy hands unless they are suited, but regardless your crappy players are going to be playing FAR more crappy As and Ks than decent players. Bad players aren't going to fold these hand...and good players that play As and Ks probably play ones with other paint, and are likely raising with them preflop. Bad players also call when they miss(or with backdoor flush draws, etc) so the time to start charging them is preflop. They will make plenty of bad mistakes postflop whether you raise or not, so let's get the money in.

Also, in both of your examples, it's still correct to raise preflop. The example with AJ I think is unlikely because a lot of players raise this preflop so you don't need to make that first raise.
-James

[James282]

Hey gunboat, your assessment is correct that early position is -ev. This chart shows that you are likely to win abuot .23 bb for each bb that goes into the pot right now, so you are getting your money in with enough of the best of it to make it +EV preflop. If you play poorly postflop you will lose money, not because of your correct preflop raise.
-James

[HajiShirazu]

I'm sure there's probably a ton of value in raising AK from the BB and that's a reason to do it too, but the reason why I raise AK/AQ/JJ and some of the big suited hands is because if I only raise with big pairs from the blinds, that's being just too predictable. I am bad enough as it is.

[gunboat]

[ QUOTE ]
Ax and Kx are crappy hands unless they are suited, but regardless your crappy players are going to be playing FAR more crappy As and Ks than decent players. Bad players aren't going to fold these hand...

[/ QUOTE ]

While what you say is true, we are talking about a hand with 4 or 5 limpers. If there are players in with Ax and Kxs, your AK is going to be less likely to hit.

Result
http://twodimes.net/h/?z=272077
pokenum -h ah kd - ad 6s - kc 5c - qd jc - th 9h - 7s 7d
Holdem Hi: 658008 enumerated boards
cards win %win lose %lose tie %tie EV
Kd Ah 97357 14.80 552980 84.04 7671 1.17 0.154
6s Ad 45586 6.93 608277 92.44 4145 0.63 0.072
Kc 5c 88659 13.47 565438 85.93 3911 0.59 0.138
Jc Qd 140639 21.37 516984 78.57 385 0.06 0.214
Th 9h 147246 22.38 510377 77.56 385 0.06 0.224
7s 7d 130850 19.89 526773 80.06 385 0.06 0.199

With a couple of crappy A and K hands in the mix, your EV goes into the toilet against other reasonable limping hands. Add the -EV of EP and raising would be clearly wrong in the above example.

With one or two limpers, the raise is clearly correct. With more, the decision is about even. With 3 or 4 I will raise if the AK are the same color to vary my play. With 5 I will normally just check.

One more point. A raise with 4 limpers will make the pot big enough so it is correct for anyone with a piece of the flop to take one off if you hit and lead.

[James282]

A couple things, gunboat.

[ QUOTE ]
While what you say is true, we are talking about a hand with 4 or 5 limpers. If there are players in with Ax and Kxs, your AK is going to be less likely to hit.

[/ QUOTE ]

I hope this is obvious to anyone playing poker. I was responding to Nate's point that he would want to raise people who play crappy As and Ks. His point was that these people will pay you off more, giving your preflop raise more equity. Since you can never ever know what your opponents exactly have, a raise is going to be correct far more often than not, and WAY WAY more often than making it EV neutral. In your example where both an A and K were out against 5 limpers(this is unlikely but certainly plausible) your raise is EV neutral. On average, in unraised pots, you will be against less than 2 of your outs, making this a clearly +EV raise in itself.

[ QUOTE ]
A raise with 4 limpers will make the pot big enough so it is correct for anyone with a piece of the flop to take one off if you hit and lead.

[/ QUOTE ]

The pot is already big enough for people with as little as one pair to keep drawing after the flop. Also, I'll repeat myself, bad players are going to draw with any piece of the flop regardless, do NOT worry about "manipulating pot size" against bad players. I think this part of HPFAP should be completely ignored unless you are playing against a table of player who are basically breakeven or better(and you shouldnt be playing at this table). [ QUOTE ]


[/ QUOTE ]
No need to get fancy here. Just raise and start pounding away. The only reason I could think of not to raise is that you feel like most of the people in the pot play much better postflop than you, in which case, you should once again not be at the table. By the way, not raising AK against three crappy limpers would just be terrible IMO.
-James

[t_petrosian]

Can someone explain what UTG means! And don't call me a complete idiot for not knowing... /images/graemlins/smile.gif

[t_petrosian]

It occurs to me that if you already have 5 limpers, you have a decent amount of money in the pot already. Considering that, I think it makes sense to then raise them all. If anyone hangs around after that, more power to you, since you probably have the better hand. If they all fold, then you won a decent pot with 6BB in there. The fact of the matter is, however, I would prefer to be head-up or close to it with AK, regardless of position, than up against 5 players.

[NoChance]

UTG = under the gun (first player to act)

[gunboat]

[ QUOTE ]
In your example where both an A and K were out against 5 limpers(this is unlikely but certainly plausible) your raise is EV neutral. On average, in unraised pots, you will be against less than 2 of your outs, making this a clearly +EV raise in itself.


[/ QUOTE ]

I can't do probability calculations, but with 5 limpers I would expect to be against two or more of my outs over half the time. And if its only one, that means you only dominate one of the limpers. With just one of your outs in the field, you can still be a dog to suited connectors and unsuited broadway connectors when the field is this large.

pokenum -h ah kd - ad 6s - 4c 5c - qd jc - th 9h - 7s 7d
Holdem Hi: 658008 enumerated boards
cards win %win lose %lose tie %tie EV
Kd Ah 117616 17.87 536094 81.47 4298 0.65 0.182
6s Ad 39334 5.98 614376 93.37 4298 0.65 0.063
5c 4c 124398 18.91 533258 81.04 352 0.05 0.189
Jc Qd 121475 18.46 536181 81.49 352 0.05 0.185
Th 9h 133539 20.29 524117 79.65 352 0.05 0.203
7s 7d 117348 17.83 540308 82.11 352 0.05 0.178

Of course, in a field this large, there will most likely be a couple of REAL crap hands. Perhaps enough to make the raise +EV.

Does anyone have enough pokerstat hands that would give a meaningful number here? Probably not, but maybe if we combined everyone's stats....hmm, that would be an interesting project.

[Doc JS]

Player on the immediate left of the big blind...

[morello]

Curious,

How do you play your AK when it misses the flop and you're first to act vs 5 others in this situation? If you bet the flop and the turn doesn't help, (and say 2 opponents called the bet on the flop), do you put another bet out on the turn?

Basically, at what point are you willing to let the hand go in to the muck?

[John Feeney]

I don't have time to read the whole thread, but I'll say I think automatically raising in this spot is typically justified with "fair share" arguments. But I believe the application of the fair share concept is not nearly as cut and dried as its advocates often seem to suggest. There are many times when you can expect your EV to be greater by sacrificing immediate profit (based on fair share) on this round in the interest of greater profit to be had on later rounds and the hand as a whole. Winning the pot somewhat more often, for instance, can mean quite an increase in EV.

[PDosterM]

I think you have to be careful with AK in the big blind when you don’t hit the flop against a large crowd. I’m willing to lead if I flopped a 4-outer nut straight draw, but leading into 5 opponents with no flop help seems overly aggressive. I prefer to check-fold the flop for several reasons:

1) I’m out of position for the entire hand,
2) Against 5 opponents, I am probably beat, and many times peeling off an A or K still doesn’t win.
3) Being raised is a distinct possibility, and I would surely have to fold to one.
4) I like my opponents to know I will check a poor hand so that when I do come out firing (against fewer opponents) I am given more credit.
5) Future check raises will work better as well.

If I did choose to lead on the flop, (say with an inside straight draw) and the turn is poor as well, I’m done.

[astroglide]

the fair share argument stands the test of math. there are all sorts of seemingly subjective assumptions about winning more small pots by esteemed authors such as yourself, but i still feel that in an absence of light, dark prevails. i have yet to see any non-anecdotal, non-singular evidence that it is correct. this could probably be (ROUGH) tested with sims. how do you objectively justify your position in light of fair share facts?

[Spyder]

If the table has gotten tight and you'll lose callers when you raise, I can see a reason for limping (Early) with AK.

Other than that, I have to believe you're better served raising.

Spyder

[James282]

Fair share arguments aside, I am not totally sure how not raising causes you to win the pot more often against typical opponents. Unless you are playing against a group of thinking players(which you shouldn't be unless you are at the maximum limits) then you will undoubtedly run into people who will play ubiquitous draws the exact same way whether you raise or not.

Frankly, if people raised for their fair share every single time and never ever worried about "keeping the pot small" to "make their opponents make mistakes" or "not draw as often" they would be doing themselves a great service as opposed to attempting to "keep the pot small" too often. Astroglide is correct, saying one should not raise when one has more than his fair share of pot equity can not be justified by speculation, regardless of how experienced someone is, unless someone knows the exact table texture of the hand being discussed.
-James

[John Feeney]

[ QUOTE ]
the fair share argument stands the test of math.

[/ QUOTE ]

As does the “increased wins” argument. No different. All sorts of poker concepts do. But often one concept is outweighed by another in a given situation. I’m suggesting that this can sometimes be the case here.

[ QUOTE ]
there are all sorts of subjective assumptions about winning smaller pots more frequently

[/ QUOTE ]

And objective ones too.

[ QUOTE ]
i have yet to see any non-anecdotal, non-singular evidence that it is correct. this could probably be (ROUGH) tested with sims.

[/ QUOTE ]

That suggests you’ve never seen such evidence to show that always raising is better than anything else either.

It’s nothing mysterious. To estimate the EV for raising you need only the usual assumptions concerning pot size and how often you’ll win the pot. Plug them into the standard EV calculation and you have your EV. To make the “increased wins” argument you simply adjust the same assumptions and do exactly the same thing.

Would someone like to offer an estimate for the situation at hand concerning how often you should win the pot after raising preflop? I haven’t thought it through and don’t want to bother with it. But if we can agree on that, then someone can run the math comparing the two scenarios. It takes a little while to work through the estimates of how many bets should be in the pot at various stages, the consequent average pot size (when you win it) in each scenario, etc. So I’d rather not bother with it myself. I assume someone who’s more interested in it might want to. (I worked this out to my own satisfaction long ago.)

Yes, I suspect sims for this would be very rough at best.

You almost don’t need such work, though, when you think about how obvious it is sometimes that your chance of winning the pot will skyrocket when you refrain from raising. This can be in a situation like the present one, or any of a few others. It may be so obvious, for instance, that you’ll be able to bet and get raised by the player immediately behind you, or to work in a check-raise to thin the field, that to do otherwise would just be silly.

[John Feeney]

[ QUOTE ]
Fair share arguments aside, I am not totally sure how not raising causes you to win the pot more often against typical opponents.

[/ QUOTE ]

It doesn’t. It causes you to win it less often. Maybe that’s what you meant?

[ QUOTE ]
Unless you are playing against a group of thinking players(which you shouldn't be unless you are at the maximum limits) then you will undoubtedly run into people who will play ubiquitous draws the exact same way whether you raise or not.

[/ QUOTE ]

Maybe the games are different now, but I used to run into lots of thinking players at limits as low as 10-20, a great many at 40-80 or so.

[ QUOTE ]
Frankly, if people raised for their fair share every single time and never ever worried about "keeping the pot small" to "make their opponents make mistakes" or "not draw as often"

[/ QUOTE ]

None of those are the reasons to sometimes refrain from raising, IMO.

[ QUOTE ]
Astroglide is correct, saying one should not raise when one has more than his fair share of pot equity can not be justified by speculation, regardless of how experienced someone is, unless someone knows the exact table texture of the hand being discussed.

[/ QUOTE ]

Nothing I’m suggesting is more speculative than “fair share.” It’s all equally mathematically demonstrable.

What I'm suggesting is that players (well, posters) often apply the "fair share" argument simplistically, indscriminately really. Often it is outweighed by other equally sound concepts.

[astroglide]

[ QUOTE ]
As does the “increased wins” argument. No different. All sorts of poker concepts do. But often one concept is outweighed by another in a given situation. I’m suggesting that this can sometimes be the case here.

[/ QUOTE ]
...and i'm suggesting no proof has ever been provided for this assertion. it is very easy to prove out "fair share" preflop raises, against random or specific hands, and it has been done in this very thread already.
[ QUOTE ]
And objective ones too.

[/ QUOTE ]
where?
[ QUOTE ]
That suggests you’ve never seen such evidence to show that always raising is better than anything else either.

[/ QUOTE ]
no it doesn't. if the hand wins more than its fair share against random hands and virtually all legitimate limping hands (barring attempted AA limp-reraises and such), that is proof, not just evidence.
[ QUOTE ]
(I worked this out to my own satisfaction long ago.)

[/ QUOTE ]
how specifically did you do that, and did you discard your results?
[ QUOTE ]
Yes, I suspect sims for this would be very rough at best.

[/ QUOTE ]
i think they're probably as good as it gets. to know if limping for the intention of maintaining a low pot size, checkraising people off of draws, etc you not only have to put them on specific hands but put those specific hands on a specific flop and make assumptions about their play. fair share math can be roughly proven out with simple showdown poker, and the rest can be inferred.
[ QUOTE ]
You almost don’t need such work, though, when you think about how obvious it is sometimes that your chance of winning the pot will skyrocket when you refrain from raising.

[/ QUOTE ]
it is quite obvious that i can win smaller pots more often by checking here, but it is not at all obvious that i will make at least as much or more as i would by raising. this advice has been around since before the days of sims, and i have never seen literature that even attempts to prove it out with anything other than blanket statements akin to "would you like to win $1 5 times or $4 1 time?."

i could certainly see doing it in a limit a little higher than i should be playing. i could see doing it on my last round when i've booked a small win and i'm playing mother hen with my checks. i could see doing it even before those scenarios just because i like to book winning sessions. none of those reasons are correct, though. i play online only, 6 tables at a time now and the long run happens much faster so it's a lot easier to drop that baggage.

i would love to be proven wrong here, because it will make me more money.

[John Feeney]

[ QUOTE ]
...and i'm suggesting no proof has ever been provided for this assertion.

[/ QUOTE ]

Well, you didn’t quote the part of my post where I laid out for you exactly how to show it, just as you can show the EV for raising. I said it’s as simple mathematically to prove the value of the “increased wins” argument as it is the “fair share” argument. Can you make it worth my while to post such a comparison today? ($$$) /images/graemlins/wink.gif

[ QUOTE ]
[ QUOTE ]
That suggests you’ve never seen such evidence to show that always raising is better than anything else either.

[/ QUOTE ] no it doesn't.

[/ QUOTE ]

Yes it does. You say:

[ QUOTE ]
if the hand wins more than its fair share against random hands and virtually all legitimate limping hands (barring attempted AA limp-reraises and such), that is proof, not just evidence.

[/ QUOTE ]

That’s not proof that that raising beats not raising in a case like the present one. It is proof only that raising is +EV. We know that. The question is: Are there instances when not raising has better EV? Again, your statement indicates that you have not seen any proof that always raising is better.

[ QUOTE ]
[ QUOTE ]
(I worked this out to my own satisfaction long ago.)

[/ QUOTE ] how specifically did you do that, and did you discard your results?

[/ QUOTE ]

Exactly as I described in the part of my post that you didn’t quote. A simple EV calculation, trying out a variety of reasonable assumptions to find the conditions under which not raising beats raising, the points at which they intersect, etc. I don’t know if I discarded my results. They could be in a pile of papers in a box somewhere. I can recreate a short version of the same thing to post, but I’ve already spelled out how to do it.

[ QUOTE ]
to know if limping for the intention of maintaining a low pot size,

[/ QUOTE ]

That’s rarely the reason I’d do it, btw.

[ QUOTE ]
... checkraising people off of draws, etc you not only have to put them on specific hands but put those specific hands on a specific flop and make assumptions about their play.

[/ QUOTE ]

You don’t really have to do all that. Most of that gets averaged out in the assumptions you plug into your EV calculations. You ask something more like, “What will I win on average, over many trials, when I win the pot in this scenario. What will I lose?”

[ QUOTE ]
fair share math can be roughly proven out with simple showdown poker

[/ QUOTE ]

Much better to do it with a realistically playing simulator. However, I’m not at all sure there is any widely available simulator that plays realistically enough to test the question of whether not raising (and the subsequent tactics) might at times be more profitable.

Again, I can post an example, but why not do the math yourself (why make me work? /images/graemlins/frown.gif )or even just take my word for it? I’m not lying or making it up when I say you can show, with reasonable assumptions, that the EV for not raising can be better than that for raising.

[astroglide]

it is straightforward to refrain from raising in order to win more (smaller) pots, but it is more straightforward to get more money in the pot (especially with the best hand) when one will win more than their fair share. in showdown poker, against random or specific hands, ace king will show a much greater profit by raising than it will by checking. while we don't simply play showdown poker, i believe it is this simple fact that puts the onus on those with the checking theory to provide absolute proof. if you cannot or will not do this, how can you continue to advocate it?

[John Feeney]

“Complete proof” is impossible short of an extremely realistically playing simulator that is also capable of playing very skillfully, or, for the online case, very specific online data from a pool of very good players. I submit that you (or any player) subscribe to a great many poker strategies and tactics for which you do not have such proof.

[ QUOTE ]
in showdown poker, against random or specific hands, ace king will show a much greater profit by raising than it will by checking.

[/ QUOTE ]

As you basically concede, this is not complete proof either. So you’ve been subscribing to the “fair share” notion without complete proof.

Moreover, it does not even approach a comparison with checking in the situations we are discussing. Obviously AK is so often going to be the best hand preflop that if you simply run the cards out it will make money. In a showdown sim a preflop raise is only adding to that money as no one folds no matter what. The action goes exactly the same whether you raise or not. All go to the end. What we’re talking about it precisely that the action will not go the same way in the raise versus the no-raise condition, and that in neither will all go all the way all the time.

Can I assume you aren’t interested in the mathematical comparison of EVs as that is admittedly not irrefutable proof?

Edit: As a general reply to your post as a whole, I will ask: How much more likely to win the pot do you think checking can, at best, make you versus raising? Is that enough for you to check on those "best case" occasions?

Put differently, how obvious does something have to be, short of irrefutable proof, for you to accept it?

[John Feeney]

Just to see if we're close to the same page, if we could get anywhere with math, what do you think is a half way reasonable estimate of how often you should win the pot after raising preflop against 5 average limpers (say at 15-30 or something) plus the small blind? Would 10% be reasonable? 15%? 20%? (I don't know the exact answere, but I doubt it would go any higher.)

And what might a remotely reasonable estimate be for the same thing when you don't raise? i.e., what would the 10% or 15% or 20% turn into when you increase the chance of winning by not raising?

If we went with 15% for raising and 25% for checking, would that be a reasonable place to start?

[astroglide]

[ QUOTE ]
“Complete proof” is impossible short of an extremely realistically playing simulator that is also capable of playing very skillfully, or, for the online case, very specific online data from a pool of very good players.

[/ QUOTE ]
okay, decent proof? anything higher than your word would be a start.
[ QUOTE ]
I submit that you (or any player) subscribe to a great many poker strategies and tactics for which you do not have such proof.

[/ QUOTE ]
without such proof one should go with what would seem to be best. it is easily proven that i can win good pots often by raising ak in the blinds. to my knowledge it has never been proven that i can equal, much less best my long-term raising results by checking (winning a higher percentage of the time with smaller pots). given the lack of any other empirical data, the showdown results would certainly seem to suggest that raising is superior unless proven otherwise.
[ QUOTE ]
How much more likely to win the pot do you think checking can, at best, make you versus raising?

[/ QUOTE ]
my honest to god answer is "i do not care." poker is a high card game, and i have the highest two cards. i will win a showdown versus all of my opponents more than my fair share. i realize you fully understand that. even if i were to ballpark a percentage, it would largely have to do with table context. i think it's pretty clear that a complex simulation is necessary to get remotely reasonable checking results.

as you and i have both mentioned, i do not have irrefutable proof that raising is superior - just logic and showdown results. that is a lot better, i think, than situational logic and no showdown results.

[John Feeney]

So what about the math? Would a mathematical comparison of EVs for each scanario be of interest? Would it be convincing if the assumptions looked reasonable? I can provide it (grumbling all the way), assuming no one else has ever done so here. (which, btw, is amazing to me as this is a topic that comes up so often) Please also reply to my other post about the percents of the time you should win the pot. Without some rough agreement on those assumptions we won't get far.

[astroglide]

unfortunately, i think any kind of assumption about increased wins by checking would be too great of a wild-ass-guess situation. i do have a copy of turbo texas holdem around here, though, and i may look into producing some sims on the situation soon. it should be easy to produce a lineup, deal the big blind ace king, tell him not to raise with it, and only count the hands which were unraised preflop by the others.

[astroglide]

as you and i have both mentioned, i do not have irrefutable proof that raising is superior - just logic and showdown results. that is a lot better, i think, than situational logic and no showdown results.

does that make sense to you in the apparent absence of all other data that we have? i am not asserting that checking will always be the wrong play, just that it is the worse choice knowing what we do know right now.

[John Feeney]

[ QUOTE ]
unfortunately, i think any kind of assumption about increased wins by checking would be too great of a wild-ass-guess situation.

[/ QUOTE ]

But you can run through the math with many different assumptions. First one, then another. See where one approach wins out, where the other takes over, etc. The results could be so pronounced that you'd be convinced one way or the other. You might even intentionally make the assumptions seem a little biased in favor of raising. I'll bet not raising still wins out.

[ QUOTE ]
it should be easy to produce a lineup, deal the big blind ace king, tell him not to raise with it, and only count the hands which were unraised preflop by the others.

[/ QUOTE ]

It might be a start if you run separate sims where he follows up with the appropriate tactics on the flop (or 4th). i.e., you'd need to create the right scenario with, say, a very aggressive player on the button, a tendency for the others to check to that player, or what have you. I don't recall whether that can easily be done with TTH. There would still of course be the problem of the general quality and realism of play. But it might be a start. (Such sims provide a certain level of realisim in that they don't have to rely on the same sorts of assumptions. But the math provides absolute answers without the vagaries of the black box workings of TTH.)

[John Feeney]

[ QUOTE ]
[ QUOTE ]
as you and i have both mentioned, i do not have irrefutable proof that raising is superior - just logic and showdown results. that is a lot better, i think, than situational logic and no showdown results.


[/ QUOTE ] does that make sense to you in the apparent absence of all other data that we have? i am not asserting that checking will always be the wrong play, just that it is the worse choice knowing what we do know right now.

[/ QUOTE ]

Sure, I just think showdown results without a lot of good interpretation are weak.

Note as well that what you have with the showdown results is simply proof that raising is +EV. So in this case they really tell you nothing. I mean, you knew it was +EV before you or anyone else ever produced any showdown results, no? What you don't have is any comparison with checking in the right situations.

How about if we make your chance of winning the pot 15% in the raising condition and only 20% in the no-raise condition? That's only 5 more wins out of 100 for not raising! I think in reality the effect on number of wins in the no-raise condition has to be far larger than that. But if even that showed not raising to be better would it start to look convincing? (Obviously you know I've looked into this by now. /images/graemlins/smirk.gif)

Edit: How about this idea? I could produce some math with certain assumptions. If we can agree that such conditions (as described by those assumptions) could at least occasionally exist, then we have a case for at least occasionally not raising. Right?

[John Feeney]

[John Feeney]

Do you realize that by saying something like, "I always raise in this spot," you're suggesting, "I don't think not raising can ever increase my chance of winning the pot enough to make it worthwhile"?

If I could show you that in some instances you need only increase your chance of winning the pot by a very modest amount to show greater EV for not raising, well, how would you like them apples?

[astroglide]

[ QUOTE ]
But you can run through the math with many different assumptions. First one, then another. See where one approach wins out, where the other takes over, etc.

[/ QUOTE ]
i agree that it is a cool approach, but how are you going to mathematically handle the post-flop play?

[astroglide]

we "knew" that it was +ev before just using common sense, but the showdown proves out that AK wins more than its fair share. it isn't at all surprising, but not useless.

while i am busting your balls over the recommendation to check, it's really a lot of people (including s&m) that i'm addressing. i don't particularly want this to be your mathematical cross to bear, but i don't understand how the advice can be given when we all agree that there is no proof that it is correct.

do you have a copy of tth?

[astroglide]

yes, i realize that i am saying that. as a default play, i still believe that it is correct. i am perfectly willing to concede that situations may exist where it isn't, but i have not been shown any proof of it. i'd still bet that raising is correct in a majority of the instances even if it were shown that checking is sometimes correct. i love them apples any time it enables me to make more money playing poker, and that would be one of those instances.

do we both agree that it's going to be a lineup-specific play? e.g. raise against a BLAH table, check against a BLAHBLAH table.

[John Feeney]

[ QUOTE ]
agree that it is a cool approach, but how are you going to mathematically handle the post-flop play?

[/ QUOTE ]

Well, once you define the basic situation you're talking about (e.g., in the big blind with AK, 5 limpers plus small blind, passive players except button who's aggressive but wary of your check if you raise preflop...), then you have to estimate an average number of bets you'd expect to win on each street in each scenario to come up with an overall average for each scanario. You just ask, for instance, "If I bet all the way through in the raising scenario, until I'm convinced I'm beaten, about how many bets will I win on average when I do win?" Stuff like that. You don't have to work through the details of lots of different specific variations because they all get averaged into one estimate that covers the basic situation you've defined (above).

[astroglide]

there just seem to be too many guesses/estimations/assumptions for it to be believed. i would trust tth's crappy play over that. i think it would be best to simulate before attempting to math it out.

[John Feeney]

[ QUOTE ]
i don't particularly want this to be your mathematical cross to bear, but i don't understand how the advice can be given when we all agree that there is no proof that it is correct.


[/ QUOTE ]

I'm happy to post the math. I did it hours ago. I just don't want to post it if it's going to be dismissed out of hand. That's why I've been trying to get a sense of what assumptions you'd see as at least remotely reasonable.

There is no absolute proof that it's correct, though the math can be mighty close to it. Similarly, there's no proof that raising is better in the situations in question.

I no longer have TTH.

[John Feeney]

[ QUOTE ]
do we both agree that it's going to be a lineup-specific play? e.g. raise against a BLAH table, check against a BLAHBLAH table.


[/ QUOTE ]

Yep. And often more specific than the table, involving what kind of action you can expect from specific players.

[astroglide]

i could locate tth, run some sims, and we could exchange our results hostage-style!

[John Feeney]

[ QUOTE ]
there just seem to be too many guesses/estimations/assumptions for it to be believed.

[/ QUOTE ]

Having played a lot, you can make most of those assumptions very reasonable. e.g., you can know beyond any doubt that you're not going to average winning over some number of bets or under some other number. Then you can test it out for various numbers in between those two. Sometimes the results will be very clear in the sense that one tactic keeps proving superior to the other, almost all through that range - that sort of thing. That's why I said in another post that sometimes the math can come mighty close to "proof."

I'm not sure TTH players can be made to respond to things like check-raises realistically enough. Not that it would have no value, but if one could show something almost irrefutably through the math, that's more convincing to me. I haven't carried the math out to that point, but what I've seen so far looks strongly in favor of not raising in this spot (with the specifics as I defined them when I did the math).

[John Feeney]

Maybe. I haven't looked at TTH in a long time, and may have trouble assessing how you set up the sims. That can be very frustrating. I recall someone once doing some sims on a somewhat related question, and posting that they showed clearly that a particular tactic was better than another. Once I looked closely at how he set up the sims, though, it was clear that he didn't effectively simulate the question at hand. But it took a lot of work to tease that out, even though I was more in touch with TTH at the time. Still I'd be happy to see any sim results (along with detail as to how it was set up). Also, I may be able to get someone else who I know is very good with TTH to run similar sims. As for my calculations, I've already basically told you my results. All I can add now is the exact numbers and the assumptions I plugged in. No need to do it hostage style. /images/graemlins/shocked.gif I'm an open book. /images/graemlins/grin.gif I need to sleep, but I'll just post an example tomorrow, barring anyone else doing so.

[Bozeman]

John,

I believe that for most of these situations when you raise you will win the pot ~20% of the time. I find it difficult to believe that you could nearly double your win rate by not raising. More than 5% increase seems too large to expect.

Not to say that I disagree with you, but I imagine that you will also need to increase the amount you win at least a reasonable fraction of the time to make raising preferable,
Craig

[UWWJAMES]

What is HEFAP and THFAP...I need these for my acronyms list...

[scrub]

I've been spending the past week trying to write a stochastic simulation for a much less complicated system than a poker game, and it's incredibly difficult to do well. That being said, I think it's a better method than EV calculations that John Feeney proposed, and strangely enough, the method that is more likely to provide convincing support for his argument.

I'm looking forward to seeing both, though. I go back and forth on this situation and think it's a very open question. I didn't have time to go through the entire thread, but the part that I read was some of the most interesting stuff I've seen on twoplustwo in a while.

Thanks guys.

scrub

[MicroBob]

"No need to get fancy here. Just raise and start pounding away."

i haven't read all the response...this is one long damn thread!! but this is the mentality you should be taking at LL HE imo.

it's amazing the number of pots i have won since my semi-recent added aggression. inducing all kinds of folds and keeping my opponents off-balance. it's great.

play your AK like it's worth something and your opponents will in turn think you have something.

you dont want to get too crazy and chase all the way with nothing of course....but it wouldn't be unheard of for your raise and then bet on the flop to scare away 2 or 3 of the players....and then to win when the other two miss their draws.

it doesnt happen all the time...but your raise can scare players who otherwise might have stayed in long enough to catch their two-pair or trips.

the added deception defeats your goal. you want your opponents to think you actually have a bigger hand then AK...not a smaller hand.


manipulating the pot gives the draws a reason to stay....but your PF raise and post-flop aggression can scare them into leaving...or at least playing too passively when they have you beat and you would have stayed in anyway.

when you DO hit top pair....they will be more inclined to slow-down with their two-pair or trips because they still have to fear AA or KK. and when you do see real strength you know to get out of the way assuming they are the type of player observant enough to notice that you raised.


regarding the 2-dimes analysis of different hands, etc.
it's not juabout how many pots you win....it's about the size of the pots you win as well as the amount you will lose when you don't hit.
as contradictory as it seems, i believe that raising the AK will prevent you from losing as much on the hands that you don't hit.

[Bartholow]

I haven't read all the responses, so I don't know if these 2 quick points have been covered or not.

1) AK is, in many ways, a big drawing hand. Just as with a lot of limpers you should raise big draws for value, AK can certainly raise for value, not just to get people out/set up semi-bluff on the flop.

2) When you raise preflop, you also tie YOURSELF on more postflop. Since you have to pay off more bets in situations where you could otherwise fold, you increase your reverse implied odds. Just looking at PERCENTAGE of losing hands is wrong here; while your wins are larger, your losses are also larger. This problem is exacerbated by your position.

[John Feeney]

Okay, here’s a simple comparison. I put a moderate amount of time into the estimates/assumptions, but a good deal less than I could have. It can get tedious (though not half as tedious as setting up a TTH sim /images/graemlins/crazy.gif ). Also, some sort of valid hard data, perhaps good online data, could obviously improve the estimates. As they are, they’re a bit quick and dirty and certainly vulnerable to criticism. Still, I think they’re generally within reason.

First let’s go directly to the EV comparison calculations for the Raise case and the No-Raise case:

RAISE

(.80 * -4.78) + (.20 * 23) = 0.78

NO RAISE

(.75 * -2.27) + (.25 * 12) = 1.30

Now here are the estimates/assumption that went into those. (Remember that they’re averages. Think of typical scenarios on each street and ask yourself if the numbers I submit are not reasonable as averages, understanding that averaged in are times when you win without a showdown, times the pot gets bigger or not as big, etc.):

Raise scenario:

1) What you win when you win
You average winning a pot of 23 small bets. A typical scenario here being: 13 bets preflop (2 bets each from 5 limpers + small blind, + your first bet) + 4 bets on flop (one each from 4 callers) + 4 bets on the turn (2 big bets, one from each of 2 callers) + 2 bets (1 big bet) on the river.

That looks kind of generous, but I’ll give that to the raiser, making it a bit tougher for the non-raiser to overtake him.

2) You win the pot 20% of the time. (Figure you might win half the time you hit plus a few times when you don’t hit and win without a showdown = 15 +5 = 20. Things like winning in a showdown with ace-high, having started with six opponents seeing the flop, is probably infrequent enough to ignore and counterbalanced by infrequent losing scenarios that I ignored as well.)

3) What you lose when you lose
When you lose you’ll average losing 4.78 small bets (beginning with your preflop raise). This estimate, for each scenario, takes some work to come up with. You have to look at what you’ll lose when you hit a pair or better but lose, and what you’ll lose when you don’t hit and lose, the former occurring about a third of the time that you lose. This involves working through a variety of ways of playing the hand out and losing, considering about how often each should occur and finding an average taking into account those frequencies.

Note as well that while it’s true, as one poster pointed out, that your preflop raise actually helps you save some bets some of the times you’re beaten, it also ties you on (as someone else pointed out) at other times. More importantly, it is far outweighed by the savings in the No-Raise scenario of having invested nothing preflop all those times you don’t hit and lose.

No-Raise scenario

1) What you win when you win
You average winning a pot of 12 bets when you win. A typical scenario here being: 7 bets preflop + 2 on flop + 2 on turn (one big bet from one caller) + 1 on river (0.5 big bets) (Note that the numbers are brought down partially by the increased times you’ll win the pot without a showdown in this scenario.)

2) You win the pot 25% of the time (Start with the same 15% that began the estimate in the Raise scenario, add fewer wins when you don’t hit, but quite a few more when you hit. e.g., 15+3+7 = 25).

3) What you lose when you lose
When you lose you’ll average losing 2.27 small bets. (See comments on this estimate for the Raise scenario.)

You can of course change the assumptions to the point that raising wins out. But I think mine are at least arguably reasonable. And there’s a good bit of room to bias them in the direction of raising and still show better EV for not raising. Moreover, some (such as what you win when you win in the Raise scenario) are probably already somewhat biased toward the Raise scenario. To get raising to win out, in fact, I find I have to either bias an estimate pretty severely from the above or start biasing multiple estimates. If you change them to the point where raising wins, I’m not sure it would be easy to make them look reasonable anymore.

Someone will probably take issue with some of the above. I could have made an arithmetic error or two as well. But I think there’s a more important issue than the precise details of the assumptions I used. That is that even if my assumptions were fairly significantly off, the EVs for the Raise and No-raise scenarios would still at least be close, suggesting that in situations especially suited for the No-Raise approach (more limpers, a player mix that should be ideal for a field-thinning check raise...), it is indeed probably correct. Here I based my assumptions on a decent but not great situation for not raising. (Sorry that I can’t provide a whole lot more detail right now. I’m not trying to be deceptive; I’ve just put way too much time into this already and have other things to do.) So I would argue at least that always or virtually always raising in these sorts of spots looks pretty convincingly to be wrong. Hope it’s of some use!

[Senor Choppy]

How exactly did you come up with the -4.78 and -2.27 figures? It seems that your entire argument here hinges on the accuracy of those in particular, but all you say about it is it "involves working through a variety of ways of playing the hand out and losing" etc.

From what I can tell, the calculations you posted here are based on assuming a 5% increase in win rate, along with an EV of -1.51 small bets postflop compared to not raising. I don't buy either of these, nevertheless, all you've really shown here is a few assumptions and not how you arrived at them.

I would also imagine that raising will win virtually the same amount of pots as not raising, which is possible because of the added times you have to committ to seeing the turn (your opponents aren't the only ones that will go farther in the hand because of improved odds).

[John Feeney]

[ QUOTE ]
How exactly did you come up with the -4.78 and -2.27 figures? It seems that your entire argument here hinges on the accuracy of those in particular, but all you say about it is it "involves working through a variety of ways of playing the hand out and losing" etc.


[/ QUOTE ]

That's exactly what it involves: coming up with all the basic variations on playing the hand out and losing, accounting for the frequencies of hitting, not hitting, etc. If you want me to copy and post my work (from paper) then you're asking for more than I'm willing to invest in time. (I'll give you all that just after I give you the shirt off my back, ya know?) I can only say that I used numbers that seemed reasonable for every situation. There is room for some variation. So my exact figures don't matter much anyway.

But if you want to see how it works, you can run through the same stuff yourself. It's the only way to come up with reasonable estimates for what you'll lose when you lose. Feel free to take a few hours to come up with it yourself, then type up and post all your work. No doubt your numbers will differ some from mine, but I doubt they'll be dramatically different.

[ QUOTE ]
From what I can tell, the calculations you posted here are based on assuming a 5% increase in win rate, along with an EV of -1.51 small bets postflop compared to not raising.

[/ QUOTE ]

I don't know what your -1.51 refers to.

[ QUOTE ]
I would also imagine that raising will win virtually the same amount of pots as not raising

[/ QUOTE ]

This is where you're way off base. Even the staunchest advocates of raising here understand that by not raising you are making room for more pot winning tactics from the flop on. I've never even seen anyone question this. I'll check back later. Maybe someone else can get into that with you.

[Senor Choppy]

Since no one else posted anything, I ran my own TTH sims.

I didn't want to let my bias enter into it initially, so I took the lineup that was set at the time (one that was actually used to approximate the 15/30 on Party FWIW), and used the player that happened to be in the 10 seat (Conan). I made two new profiles, one where Conan would always check from the bb with AK, one where he would always raise.

I gave him AKo in the big blind and froze the button at seat 8. I specified 5 opponents exactly, with the decision point being whether or not seat 10 saw the flop. Since I told him to fold all hands to any raises when in the big blind, this insured that the pot would be unraised (but also that no one would limp reraise, FYI). Both used deal code 7 and were for 100,000 hands. Anyway, the results:

Raising: No call other 6347, Split Pots 469.3, Whole Pots 25333, Total Pots 32149.3, Total $ Won $2,232,436 ($22.3 per hand)

Checking: No call other 6500, Split Pots 493.9, Whole Pots 25531, Total Pots 32524.9, Total $ Won $901,079 ($9.0 per hand)

Raising came out almost 1 sb ahead, which is more than I would've guessed. The number of overall pots won were pretty close between the two, which I don't feel is too surprising.

General thoughts: There are a few flaws with this approach, although I'm not sure which are worthy of real concern. First, I didn't know of a way to include only hands where it was always checked to the player initially, but included the times where it was reraised back to him. So the times when the hero is up against slowplayed aces are excluded. This was actually worse than it initially seems, because it removed all the times an opponent actually had aces from the equation instead of just causing them to play them more weakly. Only one of the profiles actually limped preflop with aces or kings, but there were only 80 times this player had aces, (assuming a loss of 4 small bets on average, with the wins being mostly insignificant, this works out to $4800). If all players open limped with aces 100% of the time and reraised if it was raised behind them, I believe this would work out to ~800 hands where the first limper had aces (assuming the same range of hands played), and if the hero loses 4 small bets on average, that comes to $48k, which isn't a significant amount overall. Even doubling it to an 8 small bet loss on average it still only shaves $1/hand off the earn of the player that raises. You would have to assume that players are limping with aces even after 1 or more other limpers, probably that the whole table was under all circumstances, or assume players were limping with a very narrow range of hands, which includes aces.

Secondly, the scenarios where there are 5 limpers given the tight/aggressive lineup that I use for 15/30 sims before, most scenarios involve small pocket pair vs. small pocket pair vs. small pocket pair and that sort of thing. Most profiles would always raise AJ, AQ, etc. and wouldn't play t6s, which is usually how these types of situations normally occur (a bunch of loose/passive players at the game instead of the normal, decent lineup). This should actually skew the results in favor of checking, since AK should actually be in better shape on average (not against 55 vs. 66 vs. 88 but more like K9o, Q8s, etc. which probably play much, much worse overall against AKo). The elimination of players with dominated trash hands because of the relatively good profiles involved here probably hurts the raising scenario.

As far as postflop play goes, I'm not really sure how to judge the accuracy of it. Most of the players are the best of the default TTH profiles, and probably are good enough to approximate normal opponents play. Because many of the profiles were the same character (Brett and Conan), this should lead to pretty even results, although if they play poorly in some situations, like being out of position after raising preflop and missing, and better in others, like having position on a preflop raiser in a large pot, then obviously the results can be way off.

Overall, there have to be some pretty big holes that I overlooked initially to swing things enough in the direction of just checking to make it preferrable to raising. I didn't take the time to run the best sim possible because the time it would take to comb the various profiles to see what the behavior was in certain specific situations didn't seem worthwhile, and once I start making adjustments to that sort of thing, my own bias becomes more significant in understanding the results. I was just shooting for an initial unbiased look given the default setup from the last time I fired up TTH.

(For anyone else that uses TTH, if there's a way to include situations like players that limp reraise that excludes hands where it is raised before it gets to the hero, I'd love to hear how).

[Senor Choppy]

If you don't want to copy and post your work here that's fine, but the entire argument seems to be based on those calculations. If everyone believed that the assumptions you used to come up with those 2 figures were accurate, there wouldn't be any disagreement here.

[ QUOTE ]

There is room for some variation. So my exact figures don't matter much anyway.

[/ QUOTE ]

Again, these figures are what matter most in your calculations.

[ QUOTE ]
But if you want to see how it works, you can run through the same stuff yourself. It's the only way to come up with reasonable estimates for what you'll lose when you lose. Feel free to take a few hours to come up with it yourself, then type up and post all your work. No doubt your numbers will differ some from mine, but I doubt they'll be dramatically different.

[/ QUOTE ]

If I believed there was a method for doing this that would provide somewhat reliable results, I would attempt it, but the problem I have, and I believe anyone else in the raise for win share camp has, is attempting this sort of thing relies on too many unreliable estimates, which make the likelihood of accurate results very low.

[ QUOTE ]
I don't know what your -1.51 refers to.

[/ QUOTE ]

You estimated that when raising you be losing 4.78 small bets on average when losing, and checking would only lose 2.27 small bets, 1.51 being the difference between the two that comes from postflop play (subtracting 1 small bet because of the preflop raise).

[ QUOTE ]
This is where you're way off base. Even the staunchest advocates of raising here understand that by not raising you are making room for more pot winning tactics from the flop on. I've never even seen anyone question this. I'll check back later. Maybe someone else can get into that with you.

[/ QUOTE ]

If you look at the sim results I posted, this doesn't seem to be the case (checking preflop accounted for an increase of 1.2% in total pots won). It's true that you can make more tricky plays to win pots that you wouldn't have otherwise, but because you'll be putting in more money postflop after raising, simply seeing more turn and river cards will increase your win rate as well. How much more you'll win because of pot winning tactics vs. simply firing a bet into a bunch of players on the flop and getting to see a good turn card is open for debate. But it's not written in stone that by checking you're going to win a great deal more than you would've by raising. Whatever the number is, I'd be willing to guess it's lower than 5%.

[MarkD]

Casual observer piping up here;

The one thing that I have seen looking from the outside in is that John seems to be discussing a fairly specific table texture (which appears ideal for the type of play being discussed), whereas the table you are simulating is nothing at all like the one he describes.

Specifically, the table texture seems to lean towards some weak passive looser players up front an aggressive opponent on on your right on the button. This type of texture would increase the amount of times you are in a situation where your AK dominates your opponents hands, as well as giving you many more opportunities to check raise and thin the field compared to the table texture your simulation used.

Just something to think about it. I have a feeling it would change your results a lot by changing the table texture. By how much I wouldn't want to guess, but I think it would be significant.

[John Feeney]

Unfortunately your sim did not address at all the question at hand. First, your sim looked at the comparative results of always raising versus always checking under identical conditions. That is not at all what we’re talking about. The times when a check is indicated are quite different from the times when a raise is indicated. (I see MarkD just posted something about this. I’ll add to his comments that another situation in which the check makes sense is when you can bet into an aggressive player to your left who will raise and knock others out.) Second, I see nothing in your post to indicate what tactics were used postflop. This is key to the checking condition, and probably the reason you got only a small difference in the percentage of pots won.

Maybe you can rerun some sims to address those things, but as I mentioned before I’m not at all sure TTH profiles can be made to play post flop the way they need to. Maybe you can simply isolate the times they do.

BTW, I’d respectfully suggest that you try not to be in any “camp” on this as that will only bias your thinking. Personally, I don’t care which camp is right; I just think the question is less settled than many would suggest, and that there’s too little questioning of the indiscriminate application of the “fair share” idea in preflop situations.

[John Feeney]

[ QUOTE ]
If you don't want to copy and post your work here that's fine, but the entire argument seems to be based on those calculations. If everyone believed that the assumptions you used to come up with those 2 figures were accurate, there wouldn't be any disagreement here.


[/ QUOTE ]

It's based as well on how much you win when you win in each condition. Those estimates were a far less protracted process which I showed.

[ QUOTE ]
If I believed there was a method for doing this that would provide somewhat reliable results, I would attempt it

[/ QUOTE ]

I've expained the method to arrive at the EV calcs. You ask to see all my work. I say it's a lot of work to copy it all, but that if you really want to play with the assumptions you can work it out yourself. If you're not willing to, I guess we're stuck.

However, I will post a little bit of one of the situations I looked at to give you a feel for it. The rest is just estimate after estimate aimed at coming up with rough but realistic averages.

I see now what your 1.51 was. Just to be clear, that's not an EV figure.

[John Feeney]

Okay, to look, for instance, at how much you lose in the No-Raise condition when you hit but lose, you need to work through a variety of the major kinds of sequences of play in which you lose the hand after hitting a pair or better. I envisioned a game in which you anticipated a decent (nothing special) check-raise field-thinning opportunity on the flop, based on the button’s tendencies and the kinds of players behind you.

In one common scenario you of course lose nothing preflop. On the flop you check-raise the button. He (and possibly other players) calls and you bet out on the turn. He (or someone else) raises, you feel confident that he has you beaten, and fold. So you lose 4 small bets. I came up with six common situations of this sort, basically ignoring very rare situations. Thus the six situations represented “100%” of all the times you hit and lose. I estimated that this particular one would happen on 20% of the occasions you hit and lose. You can represent it numerically thus (Each set of brackets represents a street (preflop, flop...).):

[0] + [2] + [2] = 4 bets lost

Another of these six is when you play as above, but are not raised on the turn. However, due to the action and developing board you conclude that you are beaten and check-fold the river:

[0] + [2] + [2] +[0] = 4 bets lost

I estimated that this would happen 10% of the times you hit and lose.

The object, I think, is not to come up with all possible betting sequences, but enough to come up with a realistic average of what you lose when you hit and lose.

So there were four more for that condition, six as well in the No-Raise condition where you miss and lose, seven in the Raise condition where you hit, and seven in the Raise condition where you miss.

As I noted previously, you have to account for the frequencies of the situations when working out the averages. And you have to account for hitting but losing representing only a third of the times that you lose.

You can move a couple of those numbers up or down within reason (e.g., change one of those 4s to a 5 or a 6 or a 3...) , and typically (from what I’ve looked at so far) it doesn’t alter the end conclusion very much. If you change a whole lot of them by a fair amount that should be enough to shift the results.

More important is something I said earlier:

“...even if my assumptions were fairly significantly off, the EVs for the Raise and No-raise scenarios would still at least be close, suggesting that in situations especially suited for the No-Raise approach (more limpers, a player mix that should be ideal for a field-thinning check raise...), it is indeed probably correct. Here I based my assumptions on a decent but not great situation for not raising... So I would argue at least that always or virtually always raising in these sorts of spots looks pretty convincingly to be wrong.”

[astroglide]

if your situation has the be that specific, i cannot begin to understand how you arrived at the conlusion that, in general, one will win 33% more pots by checking (20% vs 15%). was that conclusion only for scenarios where one has somebody aggressive to their left, and that sort of thing?

i think it's good that senor choppy has jumped in here because he's much more experienced than me with tth. in a simple bulleted-style list, what parameters would you have him add/change in order to satisfy the "rough sim" prove-out?

[astroglide]

[ QUOTE ]
there’s too little questioning of the indiscriminate application of the “fair share” idea in preflop situations

[/ QUOTE ]

well, nobody has given a credible/objective/reproduceable reason why we should question it. until somebody does, i think it only makes sense.

[astroglide]

[ QUOTE ]
I estimated that this particular one would happen on 20% of the occasions you hit and lose.

[/ QUOTE ]

...and these estimations are the plinth that your math is standing upon. how did you come up with that number?

[John Feeney]

[ QUOTE ]
i cannot begin to understand how you arrived at the conlusion that, in general, one will win 33% more pots by checking (20% vs 15%).

[/ QUOTE ]

Well, it was 25% vs 20%. That's only 25% more. I don't think giving the checking scenario five more wins is much.

Anyway,how I arrived at it was as I posted: First, you hit a pair or better on the flop about a third of the time. Now, I assumed that when you raise preflop you you'll win about half of those pots where you hit. Add to that about 5 that you pick when you don't hit and you have 20%. For the No-Raise condition I started with the same 15% and proceeded as I described previously.

[ QUOTE ]
in a simple bulleted-style list, what parameters would you have him add/change in order to satisfy the "rough sim" prove-out?

[/ QUOTE ]

I'm not sure. I'm pretty out of touch with TTH, and was never intimately familiar with it as I felt its value was marginal at best due to it's poor play and "black box" features.

[Senor Choppy]

I looked over this thread last night and didn't respond until today, I didn't remember him discussing specific table conditions :/

I can understand how an aggressive player on the button can change the results in favor of not raising. But the fact that the players were not only tougher, but had better than average cards in the sim I ran should mean that raising should actually show more of a profit, all other things being equal here, than the sim suggests.

I'll try running the sims again with a hand-picked lineup and see what happens.

[John Feeney]

[ QUOTE ]
[ QUOTE ]
I estimated that this particular one would happen on 20% of the occasions you hit and lose.


[/ QUOTE ] ...and these estimations are the plinth that your math is standing upon. how did you come up with that number?

[/ QUOTE ]

It's an estimate. You have to use them to do most EV calculations.

[Senor Choppy]

I responded to something I read 24 hours ago and didn't remember you saying that certain conditions had to be met for a check to be correct. I reread some of your posts again tonight and understand your point about a check being better under the right conditions, but still couldn't tell if you were saying that raising AKo with 5 limpers as the default showed more of a profit on average than just checking. You said it was obviously +EV, but that could be interpreted in a few different ways. I wouldn't argue that a check can be right in certain situations, but only with the specifics of what that situation might be.

You're right about postflop tactics, it's something I don't trust entirely in TTH. I have spent a decent amount of time with the program, and have even gone through and built a profile from the ground up. But I don't have the patience to make sure all the profiles I use are playing the way I want them to in every possible situation, and so the validity of the results is based on the assumption that the profiles play somewhat rationally /images/graemlins/crazy.gif

When I'm researching something away from the table, I don't have a camp. But when I sit down to play I gotta choose a side /images/graemlins/smile.gif

[John Feeney]

I think I did mention conditions, but if I didn't it was because I saw that as a given. It's what the entire argument is based on.

You'll need to put that aggressive player on the button in the No-Raise condition and not in the Raise condition. And how about players behind you that respect your raises and are unlikey to call two bets cold on the flop without something quite strong in the No-Raise condition, but not in the Raise condition?

While you're at it, how about getting TTH well and realistically? /images/graemlins/wink.gif

The conditions for the two scenarios should be very, very different. The idea is to see if there are at least sometimes, maybe only occasionally, but sometimes conditions which warrant not raising. If so, then we have at least shattered the "always raise due to fair share" argument.

You sound like you're squarely in the raising "camp," though. I'd be more comfortable with someone running the sims who didn't care which camp was right.

Edit: You just clarified your view of the "camp" thing abovce. And it seems we don't really disagree. So... maybe we'll be winding down here.

[John Feeney]

[ QUOTE ]
I wouldn't argue that a check can be right in certain situations, but only with the specifics of what that situation might be.


[/ QUOTE ]

Well damn. We agree then! Nuff said. /images/graemlins/shocked.gif /images/graemlins/laugh.gif

[John Feeney]

[Senor Choppy]

[ QUOTE ]
It's based as well on how much you win when you win in each condition. Those estimates were a far less protracted process which I showed.

[/ QUOTE ]

I understand the overall formula you posted, but the numbers plugged needed to be explained before it would have value. I'm not sure knowing the general method one could use is enough. I think the formula is going to succeed or fail based entirely on the accuracy of a few estimates, and I guess my big problem is that this approach never seems to work out too well.

[ QUOTE ]
I see now what your 1.51 was. Just to be clear, that's not an EV figure.

[/ QUOTE ]

Well it is a difference in EV, if I'm not mistaken.

[Senor Choppy]

I tried envisioning all the various situations that both scenarios involved, but it's evident that, at least for me personally, there's just too much guesswork with too little support for it. One of the 6 scenarios you list is betting the flop and turn and check folding the river, and an estimate that this happens 10% of the times you lose. I don't know exactly how often this happens to me, but I know I don't play a hand like this very often and 10% seems way too high (at least for online games). In any case, my scenarios would be a lot different along with the frequency that they occur.

Regardless of the numbers you come up with, I can't really disagree with them because I wouldn't trust my own. I think the approach itself is flawed, and think a simplified version would be necessary to come up with something I would really trust. Short of that, I think the reasons given for raising are more reliable than those that could be given for other courses of action.

[John Feeney]

[ QUOTE ]
I think the formula is going to succeed or fail based entirely on the accuracy of a few estimates, and I guess my big problem is that this approach never seems to work out too well.


[/ QUOTE ]

I think it works pretty well in some instances, though others are too complex to use it easily. (though clever and more sophisticated mathemaeicians can take it farther) For instance, it works very well IMO to show that raising on the button preflop with a small pair can sometimes be right. For that the assumptions are pretty straight forward and simple to work with. For the present situation it gets more complicated, and is a little less clear. But that's why I keep emphasizing that it looks like the two approaches (raise/no-raise) are at least sort of close, even without requiring extremely ideal no-raise conditions. That, to me, at least puts to rest the "always raise" argument.

[ QUOTE ]
Well it is a difference in EV, if I'm not mistaken.

[/ QUOTE ]

Well, it's not important, but I think that was just the difference between what you lose (or win or whatever) when you lose in one condition versus the other. EV has to take into account both wins and losses to come to a kind of average result.

[John Feeney]

[ QUOTE ]
One of the 6 scenarios you list is betting the flop and turn and check folding the river, and an estimate that this happens 10% of the times you lose. I don't know exactly how often this happens to me, but I know I don't play a hand like this very often and 10% seems way too high (at least for online games).

[/ QUOTE ]

It most likely is a little high. I believe I left that one at 10% knowing that if I reduced it it would shift the results further in favor or not raising. (It would make the loss smaller.) And I wanted to be sure to leave in numbers that were a little biased toward raising. That way, if I found a result in favor of not raising I could be more confident of it's validity.

[ QUOTE ]
Short of that, I think the reasons given for raising are more reliable than those that could be given for other courses of action.

[/ QUOTE ]

Again, the key here is simply whether there are some occasions when not raising is correct. You've said you think there are. If so, a lot of "always raise due to fair share" folks need to think again.

[astroglide]

i maintain this: raising is the correct default line of play based on what is known, and while situations may exist where checking is correct, they have not been fully described or tested. showdown results + logic > unspecific situational logic. i too "agree with you" and i'm raising this one every time.

[John Feeney]

[ QUOTE ]
i maintain this: raising is the correct default line of play based on what is known

[/ QUOTE ]

Do you think it's correct in every single situation? If not, then we have some situations in which "fair share" is outweighed by something else.

[astroglide]

Again, the key here is simply whether there are some occasions when not raising is correct. You've said you think there are. If so, a lot of "always raise due to fair share" folks need to think again.

there are some occasions where playing aces preflop is incorrect. does this mean that people need to "think again" about playing them? an extreme example, but allow me to explain:

the fair sharers will never miss a higher ev raising opportunity if they raise all of the time, but they will cost themselves money if they are in a situation where it is not correct. looking at it arbitrarily, if raising is correct 90% of the time the always-raisers will cost themselves money exactly 10% of the time. those who do not always raise, and look for opportunities where checking is appropriate, will certainly misjudge from time to time.

even if you were to clearly define situations in which you believed that it is correct to check AND we took it on faith that you were correct, i still believe that people would make less money in the long run by misjudging those situations than they would by simply raising all of the time. while this assertion does not make always raising technically correct, it makes it more profitable, which is correct.

[Senor Choppy]

I altered the lineup a bit to hopefully approximate the conditions John mentioned earlier.

Seats 1-7 are all Jane Doe profiles (description: average before the flop (38%), passive after.) The button (seat 8) is Mr. Hyde (description: If Dr. Jekyll went 'on tilt', this brute would emerge). The small blind is Crusty Jack (The tightest 'player' in the whole world: Plays 0%). And in the big blind is Conan once again. His play has been altered preflop but I also changed his tactics postflop this time, as well.

I ran 100,000 hands again, all with the button frozen and the same deal repeated for both. Here are the results with no postflop changes:

Checking: No call other: 3735, Split Pots: 870.6, Whole Pots: 24159, Total Pots: 28764.6, Total $ Won: $780,125

Raising: No call other: 3577, Split Pots: 896.2, Whole Pots: 24063, Total Pots: 28536.2, Total $ Won: $1,880,331

Again, raising still shows over $10 more in profit per hand. As far as the lineup goes preflop, I feel this is as close to perfect as it gets.

I ran this test and then looked at postflop play and realized it wouldn't be too difficult to force the player to always check-raise and I could know with 100% certainty that this part of the play postflop was accounted for (what happens after he check-raises with top pair is harder to figure out).

Here are the results for checking after the behavior for top pair with an unpaired board has been changed to always check-raising on the flop:

Checking: No call other: 3918 , Split Pots: 862.4, Whole Pots: 24231, Total Pots: 29011.4, Total $ Won: $715,229

For all the uncertainties regarding postflop play, the main reason for checking preflop (to check to an aggressive button in order to check-raise) doesn't appear to make a very signficant difference (and in fact costs us $65k vs. betting out). You could argue that the lineup still isn't right for this play, that if it were altered a bit the play might show more of a profit. But considering how difficult it is to know the exact tendancies of your opponents in most games anyway, we probably shouldn't keep changing parameters even if we could come up with the desired conclusion by doing so.

Going to bed now, will look at the results a bit more in the morning.

[John Feeney]

[ QUOTE ]
[ QUOTE ]
Again, the key here is simply whether there are some occasions when not raising is correct. You've said you think there are. If so, a lot of "always raise due to fair share" folks need to think again.

[/ QUOTE ] there are some occasions where playing aces preflop is incorrect. does this mean that people need to "think again" about playing them? an extreme example, but allow me to explain.


[/ QUOTE ]

Don't be silly. You know what I mean.

[ QUOTE ]
let's look at it arbitrarily, if raising is correct 90% of the time the always-raisers will cost themselves money exactly 10% of the time. those who do not always raise, and look for opportunities where checking is appropriate, will certainly misjudge from time to time. i believe it would not take very many misjudges to make their overall stance less profitable, and i believe this is certainly the case if the stance's proponents are unable to clearly define the situations in which it is correct to check.


[/ QUOTE ]

Poker is filled with spots in which the correct play is not the "default" or most common play. If you ignore them and just always make the default play knowing that it will be +EV, and fearing that if you try to identify the spots warranting different play, your misjudgements will cost you EV, then IMO you're just playing blackjack by rote, and are doomed never to do as well as you could at the game.

[John Feeney]

[ QUOTE ]
I altered the lineup a bit to hopefully approximate the conditions John mentioned earlier...

[/ QUOTE ]

1) In the checking condition did you have the button check-raise when he hit a pair or better, but not when he didn't? It sounded like you had him check-raise all the time. Making sure it’s the former is crucial. He of course needs to play well when he doesn’t hit as well, semi-bluffing as appropriate, check-calling as appropriate, and so on.

2) You only looked at cases of 5 limpers, correct?

3) I’d rather see the small blind along for the ride too, though that’s probably minor.

4) How likely are the “Jane Does’” to fold to the BB’s check-raise on the flop?

5) When you forced the BB to always check-raise, did he only check-raise when the bet came from the button (or perhaps another late player), or did he do so indiscriminately? It needs to be the former.

6) I’m unclear on the implications of “Mr. Hyde’s” playing style.

At any rate, though I hate to come across as discounting TTH out of hand, as I mentioned I've never put a lot of stock in the realism (or skill when it's needed) of its play. I just have grave doubts that you can get the postflop play realistic enough for this. Importantly, that includes getting the players to react realistically to check-raises. I have some vague memory, for example, of TTH profiles reacting in some kind of mindless way to check-raises. I can't remember just what it was, but I remember being frustrated with something of that sort.

[ QUOTE ]
what happens after he check-raises with top pair is harder to figure out

[/ QUOTE ]

This sort of thing is a real problem for TTH, I think. Does he adjust appropriately to how his opponents react to his check-raise? Does he tend to follow up by betting, “knowing” that he has now taken the lead and established a kind of momentum. But does he back off if the reaction to his check-raise warrants it? Are there other ways in which it’s hard to know what he or the other players do after he check-raises? TTH should get rid of all its “black box” features. They can be worked around, I’m told, but it should not be so tricky to do. It just gets in the way of both good simulation and discussion of results.

Anyway, right now my numbered questions above are of the most interest to me. It’s important to set up the conditions precisely to address the question at hand. If you want to keep working at it I’ll read with interest. Not sure how much time I’ll have to continue with this thread though. I’ve put in too much already. But if the proper conditions can be really clearly simulated, and the difference between raising and checking turns out to be very large, then maybe that will show raising to be right nearly all the time. (And I’m sorry that I haven’t even spelled out those conditions very well. I can only give this so much time. But you appear to be on the right track, and hopefully my questions above will help with adjustments.) Without such a large difference though, it’s tough to draw conclusions due to the vagaries of TTH postflop play.

As it stands, what I’ve seen from the math of how the differences in EV change as I try out a range of different assumptions for each condition suggests a robustness in terms of the results that’s hard to discount.

[astroglide]

[ QUOTE ]
Don't be silly. You know what I mean.

[/ QUOTE ]

i said myself that it was an extreme example, and asked that you allow me to explain. apparently you didn't.

[ QUOTE ]
Poker is filled with spots in which the correct play is not the "default" or most common play. If you ignore them and just always make the default play knowing that it will be +EV, and fearing that if you try to identify the spots warranting different play, your misjudgements will cost you EV, then IMO you're just playing blackjack by rote, and are doomed never to do as well as you could at the game.

[/ QUOTE ]

man, now i want to go to borders and hide your books in the cooking section. i am very surprised and disappointed at that cop-out, but the good news is i think you may be able to pitch it to successories (http://www.successories.com/) for a poster. i already outlined my basic stance in my last post, and i am happy to elucidate and/or listen to your specific disagreements. i'm even happy to be proven wrong, but i will go to the grave believing that you hit rock bottom with that generic response.

[John Feeney]

I've been polite and civil toward you this whole thread, no matter how rude you've been and no matter how tranparently erroneous your assertions have been. That you now react that way to one post in which I merely shaded things a tad more bluntly and didn't take the time to spell out all my comments in detail says a lot about your character. But it's actually a good thing. It makes a good spot for me to be finished with a thread to which I've already devoted too much time.

[astroglide]

my post was specific, and your response was generic. while i believe you won't reply anymore, i think you will read this post. please honestly consider to yourself what value your post added to the conversation. it was blank, not blunt. you could have just as well said that grass is green, as we all like to make as much money as possible in poker. i outlined exactly why i believe raising makes more money.

"there are specific situations that are not clearly defined where checking will enable you to win a smaller pot more frequently. in those situations, the more frequently-won smaller pots might be worth more than less-frequently won larger pots in the long run." that's not good enough.

[Bartholow]

[ QUOTE ]
as contradictory as it seems, i believe that raising the AK will prevent you from losing as much on the hands that you don't hit.


[/ QUOTE ]

This might be true if you are playing vs a significant amount of scared opponents, but at higher limits I don't agree. Maybe if you elaborate more on why you think this it will help.

[Senor Choppy]

1) His play for all conditions other than top pair were Conan's defaults, which are standard mostly reasonable plays (better understanding of pot odds than most players, not perfect though)

2) Yes, 5 limpers which did not include the sb.

3) I excluded him to give the hero the worst possible position at the table and because that was the situation we were discussing earlier. I'd rather have the small blind out because it helps the check-raise scenario and I'd like to do what I can to see if it can ever show more of a profit than raising preflop and betting the flop can.

4) Jane Doe will passively call down almost any straight or flush draw for up to 3 bets. With overcards she'll call 1 bet but fold to more. With middle pair she'll call 1 but will not cold call 2. With top pair and so-so kicker she'll bet and will call 1 but not 2. With top pair and a good kicker, (a Q with a pair of jacks, a T with a pair of aces), she'll go to the end for just about any number of bets.

5) The BB always goes for check-raises when he flops top pair here. I know of no way in TTH to tell him to check-raise the button, but if a MP player bets the flop, to fold, check/call all the way, or wait until the turn to raise (not sure what you want his action to be in this circumstance).

6) I went with what I believed to be the most aggressive profile postflop for the button. I think how often he'll bet when checked to on the flop is a good approximation of a loose/aggressive player on the button.

If anything is wrong with how TTH reacts to check-raises, it should help the results for checking PF. With a small pot, there's no hand that SHOULD be folding to a check-raise on the flop that hurts us when they call. If Jane Doe is calling 2 cold with gutshots or mediocre pairs, that should be great news.

I'm willing to do what I can to tweak conditions, but at what point will it still lead to any sound conclusions? If we hand picked 5 opponents and script every single action that taken throughout the hand, and finally walk away with checking being the more profitable preflop action, what have we actually proven?

And Conan's reaction to how his opponents react to his check-raise should be pretty similar for when he bets and gets raised or when he check-raises and gets 3 bet. I think it's reasonable to say that his overall EV might not be accurate, but if he's giving up something because he doesn't take into account his opponents reactions on previous streets, it should hurt him to similar degrees for both scenarios.

You say it's "tough to draw conclusions due to the vagaries of TTH postflop play.", but I think given both scenarios are similar enough that, while the overall results might be too low or too high in terms of EV, it's hard to believe that one action (checking) would lead to a large number of postflop errors or didn't allow for more expert plays, while the other option has the hero making significantly fewer errors or more expert plays.

[Senor Choppy]

[ QUOTE ]
I've been polite and civil toward you this whole thread, no matter how rude you've been and no matter how tranparently erroneous your assertions have been.

[/ QUOTE ]

Regardless of who happens to be right here, I believe that astro has been discussing this in good faith throughout, and has provided more in the way of concrete details.

One final sim: I ran the checking scenario with the same conditions as last time, but noticed Conan valued overcards too highly if the pot is small, so his default behavior for all conditions with overcards on the flop is now to check fold (whereas in the raising scenario he would've bet the flop 100% of the time).

Checking: No Call Other: 3403, Split Pots: 786, Whole Pots: 20758, Total Pots Won: 24947, Total $ Won $1,048,902

[maurile]

This is been a fascinating thread.

I'm in favor of raising with AKo (or AKs) from the big blind against any number of limpers for two reasons, the second of which doesn't apply to online games:

1. A lot of people limp with hands that hate being raised. Where I play, people limp with J9o, 63s, etc. I know that if I were ever to limp with trash like that, I'd be praying not to get raised. With five limpers, there will be several hands with so little pot equity that forcing them to put in an extra bet pre-flop is too juicy to pass up. This is nothing more than the "fair share" argument, but I think it's helpful to look at things from your opponents' perspective as well as your own. What's bad for them is generally good for you; and sometimes discerning what's bad for them is easier than directly discerning what's good for you.

2. In the game I usually play in, the rake + jackpot_drop + toke = 1.67 small bets. [Despite this, my hourly rate has been slightly higher in this game than at the same limit at Party.] You don't make money in this game by winning lots of small pots (where the rake will be a huge percentage of the pot); you make money by winning a few large pots. This enhances the "fair share" argument since, in terms of relative importance, it tilts the scale in favor of pot-building [with the best of it] over win-percentage.

[brianmarc]

The most reliable assessment of hand strength are the Income Rate Rankings published by U of Alberta. In fact, they have an analysis that shows the IR's vs. the Sklansky groupings. Some major revelations there that can transform how you play pre-flop. Regarding AKo: This has an IR of 718, putting it below hands like KJs (767), ATs (736), and about equivalent with QJs (720) and TT (714). Based on this analysis AKo should be relegated to Sklansky Group 3. (Similarly for AQo (555) which is about as strong as A9s (538)). Bottom line: AKo is clearly not a raising hand from EP.

Since I made the IR modifications to Sklansky, tempered with some of Abdul’s ideas, my results have improved dramatically. This is the result f playing about 3% fewer hands, particularly unsuited connectors and one-gappers, coupled with being more aggressive with big-to-medium suited connectors.

[slavic]

The U of Alberta rankings were made by pure showdown play correct? If that's true I don't think we can conclude quite what you are reaching at.

[maurile]

[ QUOTE ]
The U of Alberta rankings were made by pure showdown play correct?

[/ QUOTE ]
Yes. It says here (http://216.239.53.104/search?q=cache:Konz7_NBPicJ:www.cs.ualberta.ca/~darse/Papers/AIJ02.ps+income+rate+rankings+alberta+sklansky&hl= en&ie=UTF-8) that they're based on "roll-out simulations," which consist of "playing several million hands (trials) where all players call the first bet (i.e. the big blind), and then all the remaining cards are dealt out without any further betting."

[brianmarc]

Clearly roll-outs are not perfect; but neither is S&M, since they are merely using their (admittefly superior) judgment. However, the results from these sims do mirror S&M closely enough through groups 1-3, 5/5 match in each of G1 and G2, 5/6 in G3). Matching begins slipping off in G 4 (4/8) and G5 (11/17). The benefit of this analysis is then that it allows us to rethink some of the S$M groups. As a practical matter, Abdul has already done some of this and his work is close to the sims. Many experts do seem to agree that S&M’s weaknesses lie primarily in the unsuited connectors and -1and 2- gappers, which is where Abdul also finds greatest differences. This suggests to me that the sims are, in aggregate, actually better than S&M, notwithstanding the apparent simplicity of the analysis.

[brianmarc]

See A New Guide to the Starting Hands in Texas Hold'em Poker.www-2.cs.cmu.edu/People/mummert/poker/
It has a fasciam

It includes a very interesting analysis of the AKo question. One of their main conclusions is that AKo is THE HAND MOST DEPENDENT upon number of players: "One hand, AK, suffers tremendously with nearly every additional foe who plays to a showdown. This is the only hand in the group that should be raised pre-emptively in an effort to weed out the competition." Implication: Do not raise in a family pot from BB!

Take it from there.

[astroglide]

it doesn't change the fact that it wins more than its fair share multiway.

[brianmarc]

The "fact" you claim is, in fact, not supported by their analysis. At least not for all game situations. For example, against 1 player at the river, AK is ranked as the 8th best hand to hold; whereas against 5 opponents it is ranked as the 40th best hand. Hardly a ringing endorsement of your contention of "more than its fair share multiway". In fact, this analysis seems to point out exactly the opposite of your claim!

Now, if their methodology is flawed, that's a different story - and, truth is, I don't know if it is. But you will need more than rhetoric to make a cogent counter-argument.

[astroglide]

you clearly do not understand what fair share is. if 10 people are in a pot and your hand wins 10% of the time, it wins its fair share. if it wins 15% of the time, it wins more than its fair share. whether or not another hand would be preferrable does not matter.

[brianmarc]

Please don't patronize me! Of course I do know what 'fair share' means. Just tell me how a hand that changes in rank from 8th to 40th retains the same value?

Feeney has you peggged about right-an opinionated, blustering ignoramus who is full of his own self-importance.

[astroglide]

it wasn't a patronization, it was a fact. you are applying preferential values where they are wholly irrelevant; the "ranking" of a hand you would rather have has know bearing whatsoever on the decision. if you win more than your fair share, you should like your hand. i would rather have aces than AK. who cares? it doesn't change the fact that AK shows a profit, and neither does the fact that AK is "ranked" higher with less opponents.

[brianmarc]

Maybe this will get you to focus: Tell me the likelihood of winning this hand against 1, 2, 3, 4 5, 6 river opponents. And expalin your methodology.

[astroglide]

you are completely missing the point. if it is a profitable hand, you raise to increase your profits. it does not matter how it does against other numbers of opponents in other situations. those are your cards, and that is your situation, so you do what you can to increase your profit.

[brianmarc]

Astro: This ismy last shot. I have more important things to do than beat my head against the brick wall of your obstinacy.

How do you know it's a profitable hand? Is it equally profitable against 1 as well as many opponents. Where do you show that this hand will win at least 50% of the time against 1 opponent, at least 33% of the time against 3 opponents etc.

[astroglide]

by using a showdown calculator (http://www.twodimes.net/poker/). against 5 completely random hands, AKo wins 27.5% of the time. plug in arrays of 5 specific hands to get a reasonable real-world estimation. in AKo vs 88 vs J9d vs A9o vs QTo vs 65o (http://www.twodimes.net/poker/mini/?g=h&b=&d=&z=500000&h=Ac+Kd%0D%0A8s+8c%0D%0AJd+9d% 0D%0AAs+9h%0D%0AQc+Td%0D%0A6s+5d), for a singular example, AKo wins 23.40%, which is more than the fair share of 16.66%:

pokenum -h ac kd - 8s 8c - jd 9d - as 9h - qc td - 6s 5d
Holdem Hi: 658008 enumerated boards
cards win %win lose %lose tie %tie EV
Ac Kd 154002 23.40 498773 75.80 5233 0.80 0.238
8s 8c 121316 18.44 536020 81.46 672 0.10 0.185
Jd 9d 95573 14.52 557377 84.71 5058 0.77 0.149
As 9h 45107 6.86 603282 91.68 9619 1.46 0.076
Qc Td 119613 18.18 537723 81.72 672 0.10 0.182
6s 5d 112778 17.14 544558 82.76 672 0.10 0.172

it should also be noted that the real-world win rate for the hand will be higher due to the fact that AK will generally see the turn at a minimum, and other hands which may have won with running 2 pair, trips, that sort of thing generally will not. for hands that intend to go far vs hands that do not intend to go far, showdown results only provide us with a minimum win rate.

this kind of calculation is the fundamental basis of all of the reasoning for raising in this thread, and it is incredible that you actually hold an opinion on this issue without an understanding of the tools involved or how the facts have been obtained.

[Jive Dadson]

Where can I find a copy of the Income Rate Rankings paper?

-- Jive

[brianmarc]

http://www.cs.ualberta.ca/~jonathan/Grad/papp/node75.html

[Akasha]

[ QUOTE ]
While what you say is true, we are talking about a hand with 4 or 5 limpers. If there are players in with Ax and Kxs, your AK is going to be less likely to hit.

[/ QUOTE ]

By that same logic... Wouldn't thier hands be worse also. Leaving you still in the lead?