Tryes for fixed roster team tournaments

Running Chess Tournaments
1

Yes, I invented the word trye, as an alternative to a bye.

I first suggested it as a way to avoid a team bye in a team-vs-team event with, say, 4 players on each team. I think others have called it a “3-team monty” or something like that.

If there are an odd number of teams, instead of giving one team a full-point bye, which would force all 4 players on the team to sit out, you give a trye to 3 teams. Pair board 1 of team A against board 2 of team B. Pair board 1 of team B against board 2 of team C. Pair board 1 of team C against board 2 of team A. Do the same thing for boards 3 and 4, perhaps in reverse order. That way, every player on all 3 teams gets paired.

Bill Smythe

2

So the team scoring is based on the points your team earned (i.e. 2.5 is a win, 2 is a draw, < 2 is a loss? Theoretically you could have two wins or two losses or a win, loss and draw. With a standard bye, you either get two wins and a loss or a win and two draws. Interesting impact on the dynamics! How has it played out in practice?

3

Interesting. Given a standard swiss pairing, but for teams, how would you pair the 3 teams. I mean which teams would you select? Let’s say given 11 teams in the bottom score group, normally 5 plays 10 with 11 as the bye. Would you pair 5,10, and 11? Or use a different set?

4

Just as I would have assigned the bye to the lowest-rated team, I would assign the trye to the lowest 3 teams – 9, 10, and 11. The remaining pairings would then be 1,2,3,4 vs 5,6,7,8 respectively (making slight transpositions for colors, etc, as usual).

I think it’s actually been tried a few times, maybe even in the USAT Midwest. As far as I know it worked OK.

Bill Smythe

5

Your idea of a “trye” is interesting, but other than the obvious problem of having to explain this admittedly complicated scheme to people who not infrequently have trouble understanding how the basic ordered board format works, I don’t see how you could work this with a computer pairing system. Neither SwissSys nor WinTD will allow you to interweave three teams in a pairing for a single round. I can’t believe any other pairing program will, either.
The only way I can see to handle this would be to assign byes in the computer to the three teams in the “trye”, and then to score those byes as 1, 0, or 1/2 depending on the aggregate score of each team’s players for that round. That will give you the correct team score for that round, but you will lose the audit trail in the computer for individual results. Without those you cannot use the computer to calculate most tie-break systems. You would have to calculate those by hand. I wonder if anyone has figured out a way around this problem?

This system is designed for a 4 board team event, but it could work almost as well for a 5 board event. You would have to have player #5 on each team sit out, but that seems preferable to having an entire team sit out. You would also have to make sure to award the three players sitting out only 1/2 point each for that round. If you give them a full point each you not only increase the chances of two of the three teams winning that round (scoring at least 3 out of 5), but you introduce the possibility of all three teams winning(!), which would happen if each team’s top four players scored two points each in their played games. Not only is that logically absurd, but it would truly mess up the pairings for the following round.

6

In WinTD:
Each round of a trye do the following:

  1. create three dummy teams (if not done already) of four dummy players (no US Chess IDs, named non-player 1, non-player 2, etc.). Pair the trye teams with the dummy teams and put in the results for the real players. That takes care of tie-breaks.
  2. create six two-player teams, each consisting of two interwoven players from each team and two dummy players. Pair the correct set of interwoven players with each other (i.e. new team A1 boards 1 and 4 vs new team B2 boards 2 and 3, new team A2 boards 2 and 3 vs new team C1 boards 1 and 4, new team C2 boards 2 and 3 vs new team B1 boards 1 and 4. Enter the results of the games and withdraw the teams (A1/A2, B1/B2, C1/C2 are the three trye teams).
  3. after calculating trophies and tie-breaks and before submitting the rating report convert the matches with dummy teams into the appropriate forfeit/non-played results for the paired teams.

You thus have a file that can be accurately used for tie-breaks and which has the correct actually played results for ratings for players (a player listed twice in the same section will have the games combined for the ratings calculation)

MSA does not have team information, so the loss of the step one information is not an issue there. If you post results on your own website then you can use the pre-step-three file.

Note: Possible is not the same as easy. It can be done if you are determined to avoid team byes and are willing to do the data entry.

7

Thanks for the suggestion, Mr. Wiewel, on how to handle this with WinTD…I think in step #2 you mean “create six four-player teams…”, not “create six two-player teams…”?..If I am understanding this correctly, you would only have to do step #1 at the tournament. You could do steps #2 and #3 after the tournament, but before submitting the result to USCF, if a USCF rated event. It’s a fair amount of work, but giving a bye to an entire team is not something I like doing.

8

I really did mean “create six two-player teams”. Each of them are half of one of the three four-player team in the trye and the six half-teams can then be paired together for the six games between the twelve players (three half-matches of two games each). You are correct that step 2 could be done after the tournament. Step 3 must be done after the tournament. The advantage of doing step two during the tournament is that it would allow players to verify the cross-tables while they are still there. If you don’t do step 2 during the tournament then it would be possible to skip 2 and 3 entirely and instead do step 4: after uploading the results manually changing the opponents for all players in the trye matches. Using players without IDs on the dummy teams means that those games would fail ratings validation anyway so the fix would have to be done.

One caveat is if one of the players one a trye team is also in the running for a board trophy. That may require manual adjustment of the tie-breaks for those few players since the entry that is in the running is the one on the four-player team that is paired with a dummy opponent, and not the one on the two-player half-team.

9

Of course it is also possible to use 25000000 for the dummy IDs.

A question, though, would the initial proposal fall foul of the match rules?

Alex Relyea

10

I am not understanding how you can create a two player team “consisting of two interwoven players from each team and two dummy players.” Two interwoven players, and two dummy players is four players, is it not?

11

The following would be the pairings
3 trye teams vs 3 dummy teams
Harding vs Dummy A
Harding 1 vs Dummy A1
Harding 2 vs Dummy A2
Harding 3 vs Dummy A3
Harding 4 vs Dummy A4

Wilson vs Dummy B
Wilson 1 vs Dummy B1
Wilson 2 vs Dummy B2
Wilson 3 vs Dummy B3
Wilson 4 vs Dummy B4

Taft vs Dummy C
Taft 1 vs Dummy C1
Taft 2 vs Dummy C2
Taft 3 vs Dummy C3
Taft 4 vs Dummy C4

Interwoven matches between the three Trye teams
Two-player teams (the results from these matches are copied to the three matches above on a player by player basis)
Taft A vs Harding B
Taft 1 vs Harding 2
Taft 4 vs Harding 3

Harding A vs Wilson B
Harding 1 vs Wilson 2
Harding 4 vs Wilson 3

Wilson A vs Taft B
Wilson 1 vs Taft 2
Wilson 4 vs Taft 3

I guess an alternate would be to merge the last two of the matches as one four-board match with Wilson vs a mixed Harding/Taft team and leave the two-board match between half of Taft and half of Harding.

12

Or, instead of having all three 5th-boarders sit out, you could pair two of them against each other, so that only one sits out.

That’s the best you can do with an odd number of players per team. If number-of-teams is odd (3) and players-per-team is odd (e.g. 5), then total-players-to-be-paired is odd-times-odd, which is odd (e.g. 15).

I don’t see how, as long as no two players are paired against each other more than twice.

Bill Smythe

13

In the interwoven matches you could also reverse the colors on the second board in each of the three matches as you have them listed. So that the match -ups would be:

Taft 1 vs Harding 2
Harding 3 vs Taft 4

Harding 1 vs Wilson 2
Wilson 3 vs Harding 4

Wilson 1 vs Taft 2
Taft 3 vs Wilson 4

That way each team is white on the odd numbered boards and black on the even numbered boards, or visa versa if you want to reverse these colors on all six boards. This corresponds with how the colors are allocated in a normal ordered board match, and therefore should be less confusing to the players, who are likely to be a bit confused by this whole process anyway.

14

Does pairing two teams in a trye preclude pairing them head to head later in the event?

Assuming it doesn’t, does it preclude pairings of players who faced each other in the trye?

If it does, how do you decide who has to yield?

That can get complicated!

15

This topic is a spin-off from New match-play format for club events.

16
  1. It may or may not. If an organizer & chief TD opt to use tryes then they have to decide that. My gut feeling is that they could face each other (later or earlier) in a normal pairing.
  2. Since the trye pairings are cross-board (1s vs 2s and 3s vs 4s) they would only end up playing each other if there are alternates used in one match and not the other. The captains determine their line-ups.
  3. If you avoid teams in tryes from meeting each other (or having met each other earlier) you’d use the bottom teams available for the trye and you’d do transpositions to avoid pairing teams that were previously in tryes.

PS I have never used tryes. Most of the fixed-board team tournaments I’ve done have had schools that could have one or two teams and an individual section that could be tapped to make a scratch team to even things out if no school was available to field a second team.

17

What I think is a bit easier is to create “TempOppoA”, “TempOppoB”, “TempOppoC” with the actual interleaved teams. (You have to be careful about getting the board orders correct on them, since they will very likely be out of rating order). Pair “trye” team A with TempOppoA, etc. Put in the results. Then “split results” on the matches (from the team vs team pairings list) and delete the one that applies to the “temp” team. That eliminates the duplicated game. What will happen when you submit for rating is that each player on the three teams will have two entries, a standard entry and one where he has no actual game. You’ll get mismatched results and you’ll have to override that, but I believe it will get the rating calculations correct.

As Jeff said, “possible” <> “easy” (or pretty—you’ll get quite a messy cross table if you have to do that for every round, since each round will need a new set of “tempoppo” teams).

18

The “Hilton” pairings described in

englishchess.org.uk/wp-conte … irings.pdf

is a general system for doing interleaved team pairings when the number of teams is small.

19

I’ve never been a fan of avoiding a procedure just because “the computer won’t let me do it”. If, for example, I were using pairing software that made no provision for cross-round pairings, I would still use cross-round pairings. I’d just do it by hand, and wait until the final reporting stage to figure out how to prepare the rating report.

In fact, I’m wondering how good the popular pairing programs are anyway, when it comes to team-vs-team events. I’m treading on dangerous ground here, because it’s been many years since I’ve played in, let alone helped to direct, a team-vs-team tournament, so I don’t really know how the software handles team events.

I’m imagining that the procedure presently works along the following lines. After the data is entered, the pairing software makes the team pairings, assuming the top 4 players on each team will be playing (i.e. assuming no alternates will be used). After the round starts and it is determined which players on each team are actually playing, the TD then makes changes to accommodate alternates. The pairing software ultimately generates a crosstable, which looks something like this:


1  THE GRAND CHAMPIONS OF WESTERN PENNSYVANIA AND EASTERN OHIO
                                   wbwb  2199  3.5   W 72   W 40   D 28   W109
   Players:                                          (4.0)  (4.0)  (2.0)  (3.5)
   1a  Adams, Abigail     12233445 wbwb  2400  4.0   W 72a  W 40a  W 28a  W109a
   1b  Bensen, Bobby      13344556 bwbw  2300  4.0   W 72b  W 40b  W 28b  W109b
   1c  Campbell, Charlie  14455667 wbwb  2200  2.5   W 72c  W 40d  L 28c  D109c
   1d  Davidson, David    15566778 bwbw  1896  3.0   W 72d  W 40e  L 28d  W109e

2  THE DAKOTA RANCHERS, FARMERS, AND POLKA PLAYERS
                                   bwbw  2195  3.5   W 75   W 41   D 29   W110
   Players:                                          (3.0)  (3.0)  (2.0)  (3.0)
   2a  Edahl, Ernest      16677889 bwbw  2300  1.5   W 75a  L 41a  D 29a  L110a
   2b  Farmer, Florence   17788990 wbwb  2200  2.5   D 75b  W 41b  L 29b  W110b
   2c  Gibbons, Gertrude  18899001 bw-w  2190  2.5   D 75c  W 41c    -    W110c
   2d  Holmes, Hector     19900112 wbb-  2090  2.5   W 75d  W 41d  D 29c    -
   2e  Iverson, Inez      10011223 --wb  2010  2.0     -      -    W 29d  W110d

Only the top two teams are shown here. The top line is reserved for the team name. The second line shows the team’s color history (i.e. the colors on boards 1 and 3), the team’s average rating, the team’s match score, and round-by-round team results. The third line shows (in parentheses) the team’s game score, round by round.

The remaining lines show the data for the individual players, lettered a-d (or a-e if alternates are involved).

Now let’s suppose that, due to withdrawals after round 3, there are now an odd number of teams, and that teams 97, 98, 99 are not doing well, all having match scores of 0 out of 3. The TD pairs these three teams in a trye, so that the crosstable ultimately will look like this:


97 THE FORD TRYE MOTORS
                                    wbwb  1050  1.0   L 46   L 65   L 71   W 98
   Players:                                           (0.0)  (0.0)  (0.5)  (2.5)
   97a  Johnson, James     10001234 wbwb  1200  1.0   L 46a  L 65a  L 71a  W 98b
   97b  Kardashian, Kathy  10002345 bwbb  1100  0.0   L 46b  L 65b  L 71b  L 99a
   97c  Lightfoot, Linda   10003456 wbww  1000  1.5   L 46c  L 65c  D 71c  W 99d
   97d  Monson, Martin     10004567 bwbw   900  0.5   L 46d  L 65d  L 71d  D 98c

98 TRYE AS YOU MIGHT
                                    bwbw  1000  1.0   L 47   L 66   L 72   W 99
   Players:                                           (1.5)  (1.0)  (0.0)  (2.5)
   98a  Noble, Nicholas    10012345 bwbw  1150  2.0   W 47a  L 66a  L 72a  W 99b
   98b  O’Shea, Otto       10023456 wbww  1050  0.5   D 47b  L 66b  L 72b  L 97a
   98c  Peterson, Paula    10034567 bwbb   950  0.5   L 47c  L 66c  L 72c  D 97d
   98d  Quincy, Quayle     10045678 wbwb   850  2.0   L 47d  W 66d  L 72d  W 99c

99 TRYE TRYE TRYE AGAIN
                                    bwbw   950  0.0   L 48   L 67   L 73   L 97
   Players:                                           (1.5)  (1.0)  (0.0)  (1.0)
   99a  Raddison, Richard  10011234 bwbw  1100  1.0   W 48a  L 67a  L 73a  L 97b
   99b  Smith, Sean        10022345 wbwb  1000  0.5   D 48b  L 67b  L 73b  L 98a
   99c  Thayer, Thomas     10033456 bwbw   900  0.0   L 48c  L 67c  L 73c  L 98d
   99d  Unruh, Ulysses     10044567 wbwb   800  2.0   L 48d  W 67d  L 73d  W 97c

Note the non-matching opponent numbers in the round 4 column. This is due to the trye.

Now, how does the TD get the pairing software to produce this eventual result? I would suggest temporarily marking all three teams as withdrawn, then printing and posting the round 4 pairing sheet, and hand-writing the three team names at the bottom with a note saying “see TD”. The players on these three teams are going to need to see the TD anyway, to have the situation explained to them.

Then, once the games have started, the TD edits the data to put the correct opponent numbers into the database, just as he has done in previous rounds to handle alternates. Only this time he edits not only the opponent letters a-d, but also the opponent numbers, so that the team’s opponents can come from different teams, as required by the trye concept.

Eventually, perhaps, the pairing programs could be modified to automate some of the above – but not the “see TD” part, which will be a necessary step anyway from a tournament management standpoint.

Bill Smythe

20

I don’t know about SwisSys, but with WinTD, you can’t put a single game record into two locations, so you would need to do two separate game records, and make sure the results match (as described a couple of posts up). One thing to note is that no matter how you do this, the individual board standings will be a mess—you have board two’s and four’s playing above their position, and board one’s and three’s apparently playing on two and four, which (ordinarily) would mark them as board two or four. One should probably hope that (as with, for instance, a cross section pairing) that the player involved isn’t really in contention for anything.