Proposal for Knockout Pairs and Swiss Pairs

by Marvin L French

It is time we had a new event at NABCs. This proposal is for two such events, running almost concurrently, Knockout Pairs and Swiss Pairs.

Matchpoint Pair Movements

To set the stage for the KO pairs, a matchpoint event, we must first agree on what sort of contest is the best test of ability. A common answer is a one-winner movement in which all pairs meet and all compare board results with the others an equal number of times. An example is the “perfect” Howell movement used for the Cavendish Invitational IMP-Pairs. However, in such an event competing pairs have played different sets of hands and have not compared with all others on every board. An ideal contest would have competing pairs playing the same set of hands and comparing with each other on every board.

How is that possible? Start by imagining a super computer of 2050 programmed to play bridge at any selected skill level. If pairs were to play the same hands against such a computer, its skill level set appropriately, comparing with each other on every deal, you would have the ideal contest. While pairs do not meet, it is much more important to compare with a rival pair on 52 boards than to meet them on two boards and compare on at most half. As we were told when learning duplicate bridge, one’s real opponents on a board are those sitting in the same direction, not those at the table.

Look at the numbers. Say you figure to beat a rival pair 14 to 11 matchpoints per board (25 top) when face to face. Meeting them on two boards nets you an average six matchpoints over what they get (with a high variance). Comparing with them on half the other boards, you gain on them let us say 1/2 matchpoint per board (too low a number, probably). Twenty-six comparisons will gain 13 matchpoints against them, for 19 matchpoints total. However, if you don’t meet them but compare on all 52 boards, your edge is 26 matchpoints!

We don’t have that super computer today, so we must compromise with human opponents in a nearly ideal contest. Is there any such game today? Yes, in a most unlikely place. The lowly non-championship club game has two winners, North-South and East-West. In a completed 13-table Mitchell with predealt boards, pairs in each field have (1) played the same hands, (2) faced the same opponents, and (3) compared with each other on every board. You can’t have a better contest than that, can you? Well yes, if the opposing fields are strong and of approximately equal strength, reducing the element of luck.

That version of the ideal is a two-winner game, not a one-winner game. Such a game is now included in the qualifying sessions of six-session NABC pair events. Sections are paired, each pair qualifying pairs independently of the others, and each of the two fields in the two sessions (kept intact) qualify pairs independently. The result is that pairs have compared with all rivals on every board, have played the same hands, and have met the same opponents, a very fair selection process. The next step is a very unfair final two sessions.

The six-session NABC+ event finals, with overall ranking in what is a two-winner game, are not very good contests. It often happens that the top two pairs have not compared on a single board! Also, the finals are too large, resulting in a quality of play that ought to be higher for an NABC+ championship. The rule is that at least 40% must be qualified at each stage, with a goal of 50%. This policy is reasonable, but does not lead to high-quality finals, which should comprise no more than two sections.

Another minus for those events is the partial carryover of qualification scores into the finals. The qualifiers have not played the same set of hands during the qualifying sessions, so treating their scores equally is inappropriate. Instead, they should be assigned matchpoints at the start of the finals in accordance with their qualifying rank, not their scores.

The principle of counting only previous rank, not scores, when positioning finalists is near-universal, nothing new. For instance, Indianapolis 500 drivers are given pole positions based on their qualification standings. They are not spaced according to the speeds with which they qualified. Tennis players get seeding rights, football and baseball teams get home field advantage, basketball teams get home court advantage, track and field athletes get positioning preference. Hardly anyone gets carryovers. If the qualification sessions were run with one-winner movements, carryovers might make sense. However, one-winner movements are not quality contests, as discussed above.

Knockout Pairs

KO Pairs will not have the negatives described above. A six-session KO works like this: Eliminate 40-50% at each qualifying stage, yes, but make each session a stage, similar to KO teams. For instance, if starting with 16 sections, go to 10 sections in the second session, six in the third, four in the fourth, and end with two sections in the two-session finals. If the event gets larger, the percentage of eliminations could increase without exceeding the 50% maximum.

Divide the entrants are divided into two qualifying fields of approximately equal strength, using the current seeding procedure augmented to "anti-seed" the weaker pairs. The two fields, facing each other in Mitchell sections, qualify separately, since each plays entirely different hands against entirely different opponents, and compares only with pairs within the field. Matchpointing is across two sections, the same two for both sessions (the current practice). With an odd number of sections (first session only), matchpointing across three sections is acceptable.

After the first session all sections comprise 14 tables (13 if preduplication is possible). The fields are not mingled, with everyone switching direction for the next qualifying session, except perhaps for those few who must remain stationary. The current carryover formula is used for qualifying sessions only.

It may be argued that players will not tolerate elimination after just one session, but we have a nice event for them, described later.

Tiered Mitchell Final

The next step addresses the finals, a “Tiered Mitchell.” Place the top 26 qualifying pairs, a “subfield” of 13 from each qualifying field, North-South (First Tier) in two sections, facing the next 26 pairs East-West (Second Tier). Since the pairs have come from what are essentially two different games, they should not carry over even partial scores into the finals. Instead, recognize their qualifying achievements by starting them off with four matchpoints for every surviving pair in their qualifying subfield who is below them in the standings. That gives the top pair in each subfield 48 matchpoints and the bottom pair zero, a reasonable spread of nearly two boards (25 top).

The First Tier is playing for rankings 1-26, the Second Tier for rankings 27-52. After two Tiered-Mitchell sessions, with all pairs switching direction for the second session, and half switching sections, each Tier will have (1) played the same hands, (2) faced the same opponents, and (3) compared on every board with all other pairs in their Tier. The Second Tier will not be significantly weaker than the First Tier, so the overall quality of play will be much higher than in a four or five-section final, and the contest will approach the “ideal.” Can a better pair competition be imagined?

It may be argued that players will not like being in the Second Tier, playing with no chance to win. The answer to that is to look at a Swiss Team’s last few matches. Some good teams are out of contention, but they don’t mind playing on, trying for a better ranking. Besides, those wanting to drop out are welcome to do so.

KO Pairs can be a four-session event if the number of entries is not too large.

Swiss Pairs

The next step is to arrange a separate event for those knocked out of the KO Pairs. I suggest Swiss Pairs, very popular in Europe but seldom seen here. The last Swiss Pairs in Brighton, England, had 600 pairs. A tough event? Yes. The Brighton winners are always awesome pairs.

The Swiss starts with pairs eliminated after the first KO Pairs qualifying session, plus any others who care to enter. To these are added pairs knocked out after each qualifying session. Assign the incomers Swiss Pair Victory Points commensurate with their accomplishments in the KO Pairs. Participating is not mandatory, and pairs can drop out after any session.

Matches consist of eight duplicated boards, three matches per session. Brighton runs three matches in the afternoon, four in the evening, perhaps too much for ACBL players. Since this event ends on the same day as the KO Pairs, it is a five-session event, the last two of which have the same contestants.

The Swiss Pairs should be rated an NABC+ event, not given a merely regional rating for masterpoint awards. The last day’s matches include all but the 52 pairs who are in the KO Pair final, making it a rather prestigious event.

Marvin L. French
San Diego, CA

This proposal was rejected by the ACBL Competition & Conventions committee in 2005.

Also available as a PDF file.