November 16, 2005
A Poker Puzzle
Posted by Victor Fleischer

Reading Christine Hurt and Thomas Schelling while watching the WSOP on my TiVo can only lead to one result:  blogging about poker.  I'm puzzled by the following behavior:  at the end of a tournament, when a short stack is all in and two others call, the two others routinely check it down on the turn and the river unless they improve.  Why?  More details below the fold, for those who share my poker habit.

I've followed this strategy before, and it seems to work, but I'm not sure why.  It should be obvious to someone who really knows game theory, but I don't.  The commentators on ESPN called it an implicit agreement.  A poker table would be a strange place to find implicit agreements, as there is no honor among thieves. 

So here's the puzzle.  Suppose Ivey is the small blind with $100, Hellmuth is in the big blind with $500, and you are on the button with $600.  The blinds are $10-20, and the tournament pays out $50,000, $30,000, $15,000. 

You are dealt A-6 offsuit and raise on the button to $60.  Ivey calls all-in for $100, and Hellmuth calls, and you call.  (You put Hellmuth on a real hand, so you don't want to re-raise.) 

Suppose now the flop comes 8-5-2.  No flush draws.  The convention is for both you and Hellmuth to check it all the way down, assuming you don't improve on the turn or the river.  On one level, the reason is obvious: if either of you bet the flop, one of you might fold, which increases the chances that Ivey wins the pot.  On the other hand, if you think you have the best hand, why not bet?  The implicit agreement seems to be that it's more important to knock Ivey out even if it means some foregone profits in the side pot.  But not only do you give up possible profits in the side pot, you decrease your odds on the main pot. 

To illustrate, suppose Ivey has J-10, and Hellmuth has K-Q.  You'd rather see the turn and the river heads up than three-way. 

(A variation.  Suppose Hellmuth has 7-7.  Is he allowed to bet it?  After all, he doesn't need to improve to make his hand.  But of course by that logic, you should be able to bet your A-6, which you might think is good after the flop.) 

So there's two possible explanations for the check-it-down convention, as I see it.  The first is that this is a prisoner's dilemma, and poker players follow a tit-for-tat strategy, cooperating until they know the other person defects, at which point they may retaliate.  But that seems unlikely. 

The second possibility is that it all turns on the tournament structure, and the marginal increase in the likelihood of winning the $300 main pot and/or getting a call on the side pot is worth less the the marginal increase in the probability of knocking Ivey out.  Unless my math instincts are wrong, though, you're giving up a lot by not betting on the flop, esp. because it increases your odds of winning the main pot, and winning the main pot increases your likelihood of winning the whole thing vis-a-vis Hellmuth. 

I'm interested in this puzzle, in part, because there are a surprising number of norms that seem to develop at a poker table, despite the fact that the game seems downright designed to make people into selfish, rational, wealth-maximizing machines.  If it's true in poker, it might be true for things like hedge fund compensation.

Gambling | Bookmark

TrackBacks (0)

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8345157d569e200d83496462169e2

Links to weblogs that reference A Poker Puzzle:

Comments (10)

1. Posted by Joshua Wright on November 17, 2005 @ 9:12 | Permalink

Lets take a stab at the math.

Ok, pre-flop odds are (approximately):
A6 off = 36.7%
J10 = 28.6%
KQ = 34.7%

After the flop, nobody improves and there is $300 in the pot. The odds are now:

A6: 53.7
J10: 23.4
KQ: 22.9

Eliminating Ivey is worth 15K to each player as it secures second place.

If you bet out your A6, and KQ folds, you face the J10 heads up in which case you are 73% to win the hand. So the question is whether the marginal gain of getting heads up is worth the trade-off of lowering the chance to secure 2nd place. Vic's instincts are correct that going heads up improves his OWN chances, but it decreases the chances of eliminating Ivey.

Expected value (check through) = 53.7% (15,000+ 300) + 23.4% (-300) + 22.9% (15,000 - 300)= 11512.

EV (bet out) = .73(15000+300) + .27 (-300) = 11088.

This is a rough calculation as I am making a number of simplifying assumptions here, i.e. not counting the marginal increase in probability of getting the additional 20K at stake for first place. I am also not counting the marginal impact that the additional $300 might have in increasing your chances of later securing 1st or 2nd (which Vic mentions). But I think it suffices to show the value in the play.

For the record, I would not think, however, that it would be an egregious violation of this norm to bet out Ax in this situation with such an ugly flop. Of course, betting out with that 6 kicker is a problem since Hellmuth may have called (instead of raising) with AJ or AQ because he IS abiding by the norm. How is that for game theory?

Likewise, I do not think it would be a violation of this norm for Hellmuth to raise with a made hand. It may not be the right play, but I dont think it would work against the norm --- there is nothing that says a player cannot protect a made hand (which is consistent with the exception that allows one to protect a hand that improves on the flop or turn).

Re: tit for tat. It is my experience that a player who violates this norm in a tournament is likely to be punished. At the end of a tournament, of course, the incentive to deviate may be very strong since future punishment is not likely.


2. Posted by Joe Liu on November 17, 2005 @ 10:06 | Permalink

I'm strictly a limit player, so my comments may be off, but ...

Joshua's calculations seem to capture what is going on here, i.e. the idea that the expected value of checking through in terms of payout dollars exceeds the expected value of winning more chips (in either a side pot or the main pot).

But in those calculations, is it correct to combine payout dollars and chip values in the same calculation? I think it would be more accurate to express the chip values in terms of expected payout dollars. Again, I'm not a tourney player, but I seem to recall standard calculations for valuing chips in terms of payout dollars, based on the payout structure and the number of total chips in play.

Anyway, as you probably are already aware, if you want to get a more detailed answer to this question, you should post it at:
http://forumserver.twoplustwo.com/ubbthreads.php

Joe.


3. Posted by Joshua Wright on November 17, 2005 @ 11:22 | Permalink

Joe. Good point. You are absolutely right. You would really want to replace the pot value of chips (300) with its analogous value in terms of expected payout dollars. It is probably more simple to think about it without making the conversion, and as you point out, the analysis would not change.


4. Posted by Steve Bainbridge on November 17, 2005 @ 16:52 | Permalink

Vic: I'm puzzled. At the outset, you opine that "A poker table would be a strange place to find implicit agreements, as there is no honor among thieves." Yet, as you acknowledge at the end of the post, there are many social norms in poker, even (especially?) professional. There are obvious norms against cheating. Here's one that's purely social, however: Watching WSOP and WPT, I've noticed that if everybody else has gone out, you go all in, and I'm holding A-A, it's regarded as very poor sportmanship for me to slow-play my aces. There seem to be any number of other sportsmanship norms, especially relating to how one reacts to a bad beat and so on.

Now on a different puzzle. How on earth does PokerStars.net make money when the don't have ads and don't charge?


5. Posted by Joe Liu on November 17, 2005 @ 21:11 | Permalink

Steve, PokerStars.net, like most other on-line poker rooms, makes its money via the rake, i.e. a percentage of the pot taken out of every pot over a given size. Thus on-line poker is not, strictly, a zero-sum game, as you must not only win money from the other players, but win enough to beat the rake.

As for norms, I would think that they would be fairly strong at the final tables of large live tournaments, since these are likely professional poker players and therefore likely to encounter each other again in the future.

Even at the (very) low limits I play, I have always been surprised at how friendly people generally are, given the fact that we are all trying to take each other's money. In fact, I find folks more friendly at the poker tables than at the blackjack tables, where we are all ostensibly on the same side. (I assume this changes at the higher limits, but I doubt I'll ever get a chance to find out first-hand.)


6. Posted by Steve Bainbridge on November 17, 2005 @ 21:29 | Permalink

Joe: I know how online casinos make money off the rake. But at PokerStars.net, you're using play money. It's a free non-gambling site. So I still don't see how they make money.


7. Posted by Joe Liu on November 17, 2005 @ 22:55 | Permalink

PokerStars offers both play money and real money games. I assume they use the play money games to draw folks to the site, hoping some of them will convert to real money players.


8. Posted by zzyz on November 19, 2005 @ 22:29 | Permalink

I think the intuitive way to think about it is that by checking, they won't force the other player to fold, and with two people there besides the all-in player, it's more likely that one of them will beat the all-in player and eliminate someone, which is in both their interests.


9. Posted by Robin Sullivan on February 11, 2008 @ 5:22 | Permalink

Pokerstars.net makes no money. You are correct they have no cash games to take a rake from nor do they provide ads to make revenue from. Their "sister site" Pokerstars.com is the money maker. This is where people go to play for cash and as with most "gambling" type aveunues I'm sure they make quite a bit.

There are rules against advertising gambling for profit so it is pokerstars.net which is "all free" and therefore publisized on television and the like. But the people who spend their cash go to pokerstars.com


10. Posted by card counting on July 10, 2008 @ 20:38 | Permalink

There are lots of rules against the advertising gambling for profit which is all free and therefore publicized on television and the like. The people who spend their cash go.

Post a comment

If you have a TypeKey or TypePad account, please Sign In

Bloggers
Papers
Posts
Recent Comments
Popular Threads
Search The Glom
The Glom on Twitter
Archives by Topic
Archives by Date
February 2012
Sun Mon Tue Wed Thu Fri Sat
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29      
Syndicate The Glom
Subscribe

The Glom's Blog Network on Facebook:

Miscellaneous Links
LexisNexis Top Business Blogs 2011

 LexisNexis Tax Law Community 2011 Top 20 Blogs