You are viewing a read-only archive of the Blogs.Harvard network. Learn more.


Charles Nesson to James, Peter
Sep 22

jim, peter

i’ve got a good crew of students ready to play duplicate poker (sept 28). what equipment do i need and where can i get it?

Joe Barnard to Don, me, James, Peter
Sep 22

Many thanks for the introduction Jim, and hi from the UK, Charlie. I am very interested to know more about the Global Poker Strategic Thinking Society. I founded the Oxford University Poker Society back in 2002 and would love to encourage its current members to help with research into duplicate poker and mind sports in general.

So what equipment does one need/where to get it?
Well, as you can imagine, it is the preparing of the decks which poses the biggest problem in practical terms. We have a duplicate poker trial taking place in London tomorrow evening where 600 decks of cards are being pre-arranged for us! Our friends at PlayBridgeUK currently control this process using an adapted Duplicate Bridge deck-ordering machine, but they are working on the design/build of a new duplicate-poker-specific Dealer4Poker machine. I realise this doesn’t help with your 28th September event, however – was it a full tournament you had in mind (i.e. 6 teams of 6 for example, playing a large number of hands)?

For now, please find attached some literature we’ve put together on Duplicate Poker (if you haven’t yet seen this).

Kind regards,
Executive Director, UK Poker Federation

Charles Nesson to Howard, Joe, James, Peter, Don
Sep 22

howard, would you have any idea how i might go about getting the decks properly ordered to play duplicate poker. joe in UK is using an adapted Duplicate Bridge deck-ordering machine. who could i connect with here? incidentally, sept 28 is our first meeting with new students. we are not playing duplicate then, just planning it. our first effort should be two weeks later.

Howard Weinstein to me, Joe, James, Peter, Don
Sep 22

Hi Charlie, all

I can describe some potential methods other than using a dealing machine. I know nothing about dealing machines themselves or their operation, as these are done behind the scenes under the auspices of the directing staff. If desired, I can research into dealing machines, but can’t really add any useful knowledge in that pursuit.

The following are off the top of my head thoughts, trying to relate it to current duplicate bridge mechanisms without dealing machines (as in most bridge clubs and tournaments), while incorporating the inherent differences that hold-em presents. Oliver may have more & better thoughts, having had to deal with these issues directly for the London tournaments, and I suggest you may want to pick his brain if he has any time (probably not much for the next couple of months.)

If it is formatted as heads up duplicate poker (Jim or I can forward you some of my volumes of e-mail on this subject), it would potentially make the logistics somewhat easier.

However, assuming a London Eye situation, with several players at the table and necessarily more than one table involved to make it duplicate (see my footnote below), there are a couple of ways to approach this.

Say the table had 6 players. Two approaches.

The first would require an apparatus for each hand played, which designates each players hand, and which hand is the button (which preferably should rotate every hand). Each player would have a pocket or other receptacle for holding his dealt hands, and one for the 5 board cards. At the end of the hand, each player replaces their two cards into their slot, and the board has to be very carefully replaced in its proper order, with the three flop cards on top, then the turn card, then the river on the bottom. Eliminating the use of burn cards would probably also simplify matters. Even better might be to use dummy burn cards, similar to the cut cards used in Blackjack, especially to cover the river card, which might easily be otherwise exposed if using a partial deck. In this approach the initial hand can be dealt at the table, or created in advance. At the end of the hand, the apparatus would be passed to the next table, to play the same hand.

Another approach, which would require much less in the way of apparatus, but involve a bit more work at the end of each hand, is to fully reconstruct the 17 cards (assuming 6 players and no burn cards). At the beginning of each hand, the dealer would physically deal out the hand. Here, realistically one would need a printout of the pre-dealt order of the 17 cards, and replace them in that order. At each table, one of the players, or a non-player/dealer can be assigned to this task. My wild guess is it should take about 30 seconds to reconstruct the proper sequence of cards, and return the 17 cards into its receptacle. Here there are advantages. It only takes one “receptacle”. My suggestion would be a folded piece of paper or such, listing the sequence of cards inside & noting which player is the dealer and the hand number on the visible portion, the 17 cards (or including the other 35 cards might be just as easy), and a rubber band. Low tech, but effective.

In the full 17 card reconstruction approach, someone will likely need to make up the hands in advance. If using the multiple slots approach, then either method is fine. If pre-dealt, which also allows a printed recap of the hands after the session, then preferably someone would use a random deal generator to print out a list of hands and the proper sequences of the 17 cards for each hand before being dealt out.

Scoring is another issue. Low tech, each player should keep track of his gain or loss on each hand on a private tally sheet. If a team event where everyone sits in every position, the teams can just add up their winnings/losings for a net team total. However, this will not work in most imperfect formats, not to mention this provides no check on miscalculation or tallying errors.

More realistically (and for other reasons as well), the piece of paper included with each hand should contain a spread sheet like form, where at say table 1, someone enters the +/- result for each of the six players on that hand. The director collects all these pieces of paper at the end, enters the scores into his scoring system, and generates the results for each team or player, for both each hand & overall. Then he ideally would print up a results sheet, or if enough time and paper, provide each player with a copy of the matrix of results. However, scoring can be done by hand until the fairly simple program exists to accomplish this by a laptop (or maybe even smartphone?) and printer.

Eventually, there are easier ways if each table had some kind of entry device. Even beyond that, if each player had a computer like device, if one was okay with virtual dealing — this could generate images of the hands, and one could dispense with all of the the other required logistics. It would essentially emulate on-line play at the table. One step further, if one made the bets on-line, displayed to the others, it would also obviate all scoring issues.

Also eventually, one could play duplicate, either in person or on-line, even without more than one table, once we have constructed a sufficient database of results (this could be very easily initially computer generated). If you want, I can forward the applicable parts of my past e-mails on this topic to Jim (and maybe Peter, as well).

Got to run. Hope this helps.

David Lander to me, Dane
Sep 22

I was just making some notes on the duplicate rules and here are my thoughts on these issues:

I very much dislike the slowness of the reconstruction method. There is also the danger that it reveals players’ hole cards during reconstruction, giving information about other players not usually available in poker.
I like the pocket method, and I think it could be implemented more easily with some marker cards. Basically, we have marker cards for each position that are the size of a playing card (1, 2, 3, 4, 5, 6–these could be Uno cards). To fold, each player puts the marker card over their two facedown cards and pushes them slightly forward. This makes it clear who has folded and eliminates the tendency to muck by tossing them, which could cause us to lose track of who had what hand, while pushing them only a little makes it unclear who still has cards.
After the hand is over, operate just like the pocket method. When the hand is over, each player makes sure the marker card is over their facedown hole cards, and these are collected in order. There could also be a marker card for each of the three streets put in front of them, so these serve as a clear way to make sure they are in order as well as burn/protector cards. The rest of the deck could be held in a rubber band to make sure all decks end up being recombined correctly at the end of the tournament.

I think this would work with 6 tables dealing out six rounds each. All decks (marked as indicated above) are rubberbanded with a piece of paper saying the table number and deck order number (1st hand, …).
The tournament manager collects decks after each round of 6 hands and redistributes them (giving the set of decks a different label than table number, in case students can see other tables they don’t know which table’s previous cards they are getting at each round).
It may also be worth changing where the button starts each roundof6 if students are near/can see other tables. Each deck within a set of 6 decks is still played in the same relative order and with correct positions, but which deck is played first is somehow set. (This is mainly to prevent information leakage between tables in the same room due to seeing action, timing, chip countups etc.)
After sets of decks, and deck order within a set, are randomized like this, it should minimize the leakage of information in a room like Hark South.

On scoring:

One danger of the capped approach advocated in the paper is that it relies on someone counting and enforcing the cap during the hand, which is easier said than done (are we to keep our prior bets in front of us instead of combining into a pot each round?–seems there is not space for this). It also only allows scoring at the end of the tournament (or rounds or half, depending on when chips are counted).
I think we give everyone 1000 chips in the same recognizable stack, so it’s clear how much is in play. At the end of the hand, each player figures out what chips they are missing (should be pretty easy visually) and moves chips from the pot to the “betting circle” in front of them that would give them their original stack. If there are any mistakes, they will quickly be discovered and resolved as every chip is necessary to complete someone’s stack. These amounts are tallied on a scoresheet and the winner’s total circled. (The winner’s profit from the hand is simply the sum of the other 5 numbers, and this can be calculated later after scoresheets are passed in so no one is adding or counting a pot at the table.)

The virtual approach would obviously be ideal, but obviously would take a huge investment in technology. And it’s just more fun to play at a table with cards and chips, and looks better if this is every recorded or broadcast.

Charles Nesson to David, Dane
Sep 23

i love your analysis for handling the cards, scoring not so sure

David Lander to me, Dane
Sep 23

Without going the route of the spreadsheet and individualized hand tracking, then I think the best way to score is to use the method described in the handout: each player is given a (hands * cap) stack of chips at the start, bets are capped each round, and stacks are counted at the end.

I don't think there is any way to segregate individual players' bets during hands to maintain the cap (either we run out of room in the middle, chips fall into each other, or some sort of containers would block the view of the pot). I think the the solution is for a player who folds preflop to be responsible for tallying the total bet during each round and writing it down on a piece of paper (showing it to another player to confirm on each street). Any player still in the hand can ask for a count of remaining stacks (equal of course) at any time. We can pass out four function calculators to each table to assist.

While this introduces some more verbal indication of how play is proceeding, I think this should be fine as long as tables are segregated some, voices are required to be kept low, and, of course, there is randomization (and concealment) of the order in which tables get deck sets, and in the starting hand played first in each set at each table.

I think this also solves one of my other concerns regarding match scoring. I very strongly dislike the "points system" (1) of ranking players at each position. One issue is that a very small differential in chips could lead to a very large change in the rank. The bigger issue is that creates terrible incentives to make moves that are +EV for the purposes of this scoring but not +chipEV. To explain, say a player plays a set of hands during a short tournament that are relatively straightforward and likely to create little variation. Toward the end he butchers a hand or two and knows he'll most likely be down a buyin or two from the rest of the field. The incentive at that point is to "game" the tournament by trying to bluff huge pots or get in coinflips, because 6th is still 6th, but it may be possible to use volatility to get oneself back into the running. I don't think we're looking to encourage this type of play.

I like the chip count only system for a competitive tournament, as it's like a pure cash game where very hand is independent and the only incentive is to make +chipEV moves. In a club setting, I think it could lead to one terrible player's loss of many buyins counteracting the entire team's winnings, which is a worry if we are going with random teams at the meeting.

The hybrid solution (3) is no good as explained because there could be significant compression of scores that under-rewards good play if one person is far below the rest of the field, creating a large range that means that everyone else ends up scored from something like 85-100. I think it could be made better by dividing by standard deviation instead, but I'm still not sure how I feel about total match scoring.

Charles Nesson to David, Dane
Sep 24

with you all the way

Lisa Carlivati to me, Dane, landerad
Sep 26
Hi all,

Charlie’s two cases of poker chips are still missing, I believe. As I wrote to Charlie last week, Dane and David estimated we’ll need 6-10 chip sets for the HPSTS. On Amazon, a case of 500 chips is ~$35. Should we order 6 of these, which would be over $200?

I assume we need at least one or two cases for this Wednesday, so we should figure out budget ASAP! Or perhaps just for this week, we could borrow from… Dane or David?


Charles Nesson to Lisa, Dane, landerad
Sep 26

how many chips do we need to do the 6×6 duplicate tournament with a 1k cap each deal?

David Lander to me, Lisa, Dane
Sep 26

I can drop off some chips for Wednesday (I can't be there until 7 after class).

I'm thinking we label the chips as 1/2 (not 5/10 as in handout) to use traditional value/color matchups. So cap of 200. Total stack of 7200.

1 -20. 20
5 -26. 130
25- 14. 350
100 -7. 700
500 -12. 6000

This will take 6 sets of We'll also need to order some 500 chips or could use some sort of fake bills for these since they're more of a counter than a viable betting chip.

David Lander to me, Lisa, Dane
Sep 26

We should also discuss at some point whether we like the 100 BB cap. A deeper game involves a lot more skill, so I think at least for Harvard-Yale we'd want to be playing deeper like 200 BB. For this trial 100 BB should be fine though.

James McManus to me, landerad, Peter
Sep 29

In London we played with blinds of 25 and 50 with a 5,000 cap on each hand: 100 BB. Absolute minimum number of chips for 36 players would be 36 x 5,000, which would require dealers to replenish all stacks to 5,000 after every pot: onerous and time-consuming. I recommend at least 10,000 per player, in denominations of 25, 100 and 1,000. No 5,000 chips (or whatever the max bet is) are necessary. Encourage players to make change and replenish each others' stacks between hands and while the action proceeds after they've folded, freeing the dealer to open the next deck and recommence dealing. It's weird to give back you chips after winning a pot, but you will get used to it.

You should also have the tournament director make a friendly but firm suggestion to play with alacrity: thinking time should be limited to tough all-in decisions; routine folds should be made quickly; no dawdling or talking so much that you lose track of when it's your turn. (Many folks can play fast while talking; the ones who can't are the problem.) The reason, of course, is that you want to play the maximal number of hands, and a single table of Chatty Cathies will yield a whole lotta waiting by everyone else to hear the results. We only completed 60 of the 100 hands we'd prepared decks for, mainly because one table was about twice as slow as the next slowest.

Meanwhile, have FUN!

Charles Nesson to James
Sep 29

how long did it take to complete the 60 hands? was the 100BB cap for each hand about right? did people have fun?

James McManus to me
Sep 29

A little more than 3 hrs for 60: way too slow. Most tables could've finished the entire hundred, but at the midway pt one table was way behind, so they decided to cut the number: extremely unideal bc that meant the strong hands for many players were lost in the 40 unplayed decks. So make em play fast.

Yes, 100 BB is a very good number for testing skill hand by hand.

Fun was had by nearly all. Especially by those who enjoy playing efficiently and well; less so by folks for whom poker is a chance to get all loquacious and drink. So present it as a mental competition, not as poker nite with beer pizza and football on the TV. Not heavyhandedly, of course. Just make sure they remember it's a serious competition and to play swiftly whenever possible. Longish thinks remain acceptable on the obviously tough decisions.