Swiss Pairs

[ArtK78]

I find this to be very interesting. Has anyone ever participated in this type of event?

 

Quoting from The Dummy, November, 2009, the quarterly publication of the Delaware State Bridge Association:

 

The Wilmington sectional to be held February 26 through 28, 2010 will feature an event on Saturday, February 27 which is very rarely run anywhere in the United States, a two session Swiss Pairs, at 1 PM and 7 PM. Yes, Swiss Pairs, not Swiss Teams.

 

During the course of the two sessions, each pair will play a series of 6 to 8 board matches against other pairs. The number and length of the matches and the number of boards in each match will depend on the number of entries, but the number of boards played each session will be much the same as in a typical pair game. Everyone will play the same boards at the same time, and the results will be match pointed across the field. Each pair’s percentage score from the 6 to 8 boards in the match will be converted to victory points. The scale varies depending on the length of the match, but normally a score of about 66 % over the course of match will produce a blitz in that match.

 

Pairing, just as in a swiss team, is random for the first match. Thereafter, each pair will be paired against another pair with a similar record. The pair with the most victory points wins!

 

This is an event designed for pairs of all abilities. Master points will be awarded for each match won, and the event will be stratified so that everyone will have a real chance to win an overall award. As in a stratified Swiss team event, players will tend to play against other players in their own stratification after the first round or two. Of course, a flight B or C pair which is having a very good day may have to play against flight A players as a result of doing well, but generally, players will feel comfortable playing against players of their own ability.

 

Traditional one session pairs events will be available for players who are not able to play two sessions or don’t want to try this event.

[RMB1]

Very common England. If a county (region) or two counties organise a weekend event awarding green (national) master points then there will be one day of swiss pairs and one day of swiss teams. If the EBU organises a 3-day congress there will be one day of swiss teams (7x7) and two days of swiss pairs (13x8). Clubs play one-session (5x5) swiss pairs.

 

The default was to assign a round in arrears, so the scores after round 1 determined pairing for round 3. Using bridgemates and/or small fields allow current round assigning, as in swiss teams, the scores after round 1 determine the pairing for round 2. Current round assigning is becoming more common.

[TylerE]

I'm not a big fan really. I dislike any event that essentially punishes you for having a good game going into the final round.

[TimG]

I'm not a big fan really. I dislike any event that essentially punishes you for having a good game going into the final round.

How is anyone "punished" for having a good game late in the event?

[gnasher]

I'm a fan. I like any event that rewards you for having a good game by giving you better opponents on the later rounds.

[mrdct]

Over the last 10 years or so imp-scored swiss pairs has become the overwhelmingly most popular form of two-session pairs event in Australia with many, if not most, congresses (two day red-point bridge event typically with pairs on Saturday and teams on Sunday) ditching the traditional matchpoints pairs event for imp-scored swiss pairs.

 

This format, where you are still matchpointing across the field, I haven't seen before; but it might appeal to players who still want to risk their contracts for overtricks and play in anti-percentage NT contracts.

[jdonn]

I love events like this. Why not mix it up and play different ways? And how could someone mind that pairs doing well play each other near the end, would they rather that whoever is in first after reach round plays whoever is in last? That would be really funy for any team trying to catch the leaders...

[Vampyr]

Current round assigning is becoming more common.

Glad to hear that.

 

Hope to see this in Brighton this year!

[Vampyr]

Over the last 10 years or so imp-scored swiss pairs has become the overwhelmingly most popular form of two-session pairs event in Australia with many, if not most, congresses (two day red-point bridge event typically with pairs on Saturday and teams on Sunday) ditching the traditional matchpoints pairs event for imp-scored swiss pairs.

I am surprised to read this. I am sure I am not alone in finding IMP pairs a rather unsatisfactory format in many ways. I often play in a cross-IMPed club game and I think that this is the best way to run the Ladies' Trials, but I would not like a steady diet of it in congresses.

[TylerE]

I'm not a big fan really. I dislike any event that essentially punishes you for having a good game going into the final round.

How is anyone "punished" for having a good game late in the event?

Because the top two pairs play in the last round, there is more room for the pairs a bit below them to play weaker opps and have a bigger round and win the event.

[TimG]

I'm not a big fan really. I dislike any event that essentially punishes you for having a good game going into the final round.

How is anyone "punished" for having a good game late in the event?

Because the top two pairs play in the last round, there is more room for the pairs a bit below them to play weaker opps and have a bigger round and win the event.

I don't see it that way. I'd rather go into the last round with a few VP lead than get the slightly weaker opponent. In smaller Swisses where there is potential for great skill disparity between 2nd and 6th place teams so that you'd rather be in 3rd place and play the 6th place team than in 1st place and play the 2nd place team, it's very likely the 1st and 2nd place teams have already played each other before the last round. More likely that 1st is playing 6th and 2nd is playing 5th or some such.

 

Perhaps that is a penalty, if I'm in 1st place going into the last round, I'd prefer to play the 2nd place team and have better control over the result rather than rely upon a lower team to give the 2nd place team a good match.

[awm]

There is a general trend in swiss events that weaker teams don't fall out of contention over time. Basically what happens is that since each team is presumably playing a team of comparable level, we'd expect a good team and a bad team in the middle rounds to obtain (in expectation) roughly the same score. This means that as the number of rounds increases, the amount of separation in score doesn't increase as in other formats. Thus there will often be a bunch of teams "within striking distance" of the top even after quite a few rounds of play. If the number one team plays the number two team in the last round, it is normally the case that if one team wins by a reasonable margin that team wins the event. However, the loser of the "number one versus number two" match usually finishes quite a ways down (i.e. not second or even third) with a weaker team (often one that played a much weaker strength of schedule) charging up the ladder into the 2nd or 3rd place spot. If the "number one versus number two" match results in a tie or near-tie then neither team will win the event and a mediocre team can charge into the lead by blitzing a mediocre opponent.

 

This problem is much worse if rematches are allowed, but it only really makes a difference when the number of rounds is large relative to the number of teams (which it often is not).

 

Really this format is very random, especially the version that seems to be popular in England. Essentially Swiss is more random than round robin (or a pairs type movement), IMPs are more random than MP/BAM (if we fix the number of boards), and pairs are more random than teams. Putting these together, a swiss imp pairs is about the most random format available. Of course, formats with a high degree of randomness are popular because "anyone" can win! For example this is why two-session BAM has been essentially replaced by two-session swiss teams in most of the US.

 

Anyway, I think it's fun to have different types of events and get to try different things. But in general I prefer the "less random" formats.

[cherdanno]

I would rather play an event that gives me a high probability of playing other good opponents, than playing an event that gives me a good chance of winning.

I agree that IMP swiss pairs sounds a bit too random anyway - but if I read the OP right the event in question is a matchpointed swiss pairs.

[Echognome]

I'm not quite sure I understand all of what Adam is asserting. From my experience in England playing in the Swiss Pairs at Brighton and in various congresses, it sure seemed that the best pairs were winning. Of course, which of the top pairs that won varied, but I attributed that to who was playing well at the time and of course there is always some luck.

 

I actually made a long post some time ago how I thought that we lacked some variety in bridge in the U.S. (although since I play very little club bridge these days I really have no complaints).

[Vampyr]

I actually made a long post some time ago how I thought that we lacked some variety in bridge in the U.S.

Here in the UK we have a pretty good mix of formats, with the exception of teams -- these are almost always either Swiss or all-play-all with matches of 7+ boards. It seems that we rarely play multiple teams, and I think that that is a shame.

 

We also don't play board-a-match, which I think is also a shame.

[awm]

My complaint about swiss events is basically the following. Suppose team A is better than team B. We would hope that, on average, as the number of rounds played increases team A will get further and further ahead of team B in the standings.

 

If this does not happen, then there is an opportunity for team B to "get lucky" with a good last couple of rounds and overtake team A, even though many previous rounds were already played and A is the better team. This therefore creates more of a luck factor in the event.

 

The problem in swiss is, once team A is a bit ahead of team B in the standings, team A will be playing stronger opposition than team B. Thus team B's "expected score" will be roughly the same as A's "expected score." So at this point adding more rounds does not increase team A's lead over team B in the standings.

 

A really bad team will get far enough behind that one or two "lucky" rounds still isn't enough to pass a really good team, and even though they won't fall further behind, it won't really matter. So you won't likely get a truly bad team winning the event. But this does substantially increase the degree of randomization amongst then "pretty good" teams as to who ends up winning, and the problem appears to be more severe a bit further down the standings (i.e. swiss does a moderately bad job determining the best team and a really bad job determining the "upper quarter" of the teams).

 

Notice that random match-ups don't have this issue with expectations, since on any given round the expected opposition of team A is the same strength as the expected opposition of team B (actually slightly weaker because they might play each other) and so A's expected score remains higher than B's expected score each round. Thus you'd expect the gap to get wider and wider, such that eventually B is simply out of "striking distance" to make up the ground with one or two lucky blitzes.

[Fluffy]

2 last rounds of a swiss are the key, howver it pays to be on the top at the start, to avoid a really tough match.

 

The swiss we played last week in the NEC with 3 rounds to go running in position 9-11 or so we had to go to table 2 to play against Zimmerman, our penalty for mediocre scores in the beginning. Unlucky match meant 25-3 loss and we had little to no chance after it.

[NickRW]

I thought it was well understood that the swiss format, be it teams or pairs, imps or mps, is well suited to finding the correct winner - but that it is pretty terrible at sorting out who should be the correct 2nd, 3rd, 4th.

 

This is innevitable - the way that matches are arranged is that whoever is 1st plays the highest placed team/pair that they haven't met yet. Similarly with 2nd place if they are not playing number 1 and so on. So you get, in general, better teams/pairs playing each other and weaker ones playing each other. So this must depress the score of the better participants, except perhaps for the very best, compared to other formats.

 

However, if you're interested in playing against comparable opposition - which I would have thought ought to be a consideration - then swiss is likely to give you a greater percentage of the overall session playing matches that are worth playing.

 

Nick

[TimG]

If the number one team plays the number two team in the last round, it is normally the case that if one team wins by a reasonable margin that team wins the event. However, the loser of the "number one versus number two" match usually finishes quite a ways down (i.e. not second or even third) with a weaker team (often one that played a much weaker strength of schedule) charging up the ladder into the 2nd or 3rd place spot. If the "number one versus number two" match results in a tie or near-tie then neither team will win the event and a mediocre team can charge into the lead by blitzing a mediocre opponent.

I doubt many mediocre teams win events this way. It also is good to remember that good teams sometimes have off days and teams somewhere between good and mediocre play well some days.

 

This strikes me as a good place for a simulation. It's a bit beyond me, and I suspect it has been done before. But, perhaps someone will do one, or point us to the results of an old one?

[hrothgar]

If the number one team plays the number two team in the last round, it is normally the case that if one team wins by a reasonable margin that team wins the event. However, the loser of the "number one versus number two" match usually finishes quite a ways down (i.e. not second or even third) with a weaker team (often one that played a much weaker strength of schedule) charging up the ladder into the 2nd or 3rd place spot. If the "number one versus number two" match results in a tie or near-tie then neither team will win the event and a mediocre team can charge into the lead by blitzing a mediocre opponent.

I doubt many mediocre teams win events this way. It also is good to remember that good teams sometimes have off days and teams somewhere between good and mediocre play well some days.

 

This strikes me as a good place for a simulation. It's a bit beyond me, and I suspect it has been done before. But, perhaps someone will do one, or point us to the results of an old one?

Gerben and Alex Ogan provided a nice model a few years back analyzing the results of Swiss Teams formats.

 

A Swiss Pairs format will change the results somewhat, largely due to external constraints regarding the round length and an increase in the number of pairs; however, I don't think that the fundamental results should change very much.

 

Here's a posting on this same topic from a few years back. The most important line is probably the following:

 

With no SoS adjustment, tournament organizers need to run twelve 20 board rounds to have a 95% chance that the strongest team will place in any one of the top eight slots. (Simply put, Swiss Teams are a real crap shoot)

 

Swiss Team type events are one of the most popular formats in bridge.

We argue that the accuracy of Swiss Team style events can be improved

significantly if a Strength of Schedule adjustment is used to

complement the normal scoring system.

 

 

This hypothesis was tested using a series of Monte Carlo simulations.

A computer program generated 128 bridge teams with known strength.

These teams competing against one in a Swiss Teams type event. At the

conclusion of the event, the sample statistic - the ranking produced

by the Swiss Teams event - was compared with the population statistic

(the objective/known ranking of the team strength). We consider event

event formats in which the sample statistic closely mirrors the

population statistic superior to formats in which the sample statistic

deviates significantly from the population statistic.

 

 

Monte Carlo simulations can be used to test a variety of different

hypotheses. For example, are tournaments with a large number of short

rounds more accurate than tournaments with a small number of long

rounds. (None too surprisingly, the answer depends on the fixed cost

associated with the break between rounds) Alternatively, is there a

relationship between the number of teams entering a tournament and the

number of rounds necessary to accurately identify the winner. Our most

striking result involved using a Strength of Schedule adjustment to

the normal Swiss Team scoring system. We determined that a Strength of

Schedule adjustment allows tournament organizers to significantly

improve the efficiency of their events. Hypothetically, an event

organizer could reduce the time required to stage an event without

compromising the accuracy. Alternatively, an organizer could hold the

length of an event constant and significantly improve the accuracy of

the event.

 

 

Strength of Schedule adjustments can implemented in a variety of ways.

For the purpose of this study, we used a very simple SoS adjustment.

 

 

1. Run a normal Swiss Teams event

 

 

2. Calculate the total number of Victory Points won by each team

 

 

3. Sum all of the Victory Points won by each team that team i played

against, excluding the head to head competition between team i and

team j.

 

 

4.The Team's final rank is determined by adding the Victory Points

that Team "i" won in head-to-head competition and some fraction of the

total VPs won by all the teams that team "i" competed against. (This

fraction is a function of the number of rounds in the tournament)

 

 

We certainly don't claim that the SoS adjustment just described is by

an optimal implementation. However, even this very crude

implementation has a dramatic impact on the accuracy of the event.

 

 

Consider the following tournament format:

 

 

* 128 teams competing in a Swiss format

* The event consists of "N" 20 board rounds

* The primary statistic used to measure the accuracy of the event is

the percentage chance that the strongest team will land in any of the

top eight places at the close of the event. (We used other metrics

including the Spearman rank coefficient and how many of the top eight

teams placed in the top eight slots. Results were consistent across

metrics)

 

 

With no SoS adjustment, tournament organizers need to run twelve 20

board rounds to have a 95% chance that the strongest team will place

in any one of the top eight slots. If we add an SoS adjustment,

tournament organizers can run nine 20 board rounds while still

achieving a 94.9% chance that the strongest team will place in any of

the top eight places. Tournament organizers can reduce the length of

the tournament by 25% without impacting the integrity of the results.

(In comparison, if the Tournament Organizers were to run an event with

nine 20 board rounds without any SoS adjustment, the accuracy of the

event would drop from 95% to 92.3%)

 

 

At this point in time, the primary value of this study is identify the

fact that significant improvements can be made to the traditional

Swiss Teams type format. Over time, we hope that it will be possible

to make more concrete recommendations regarding the best

implementation for an SoS correction as well as an executable that

could be used to optimize events formats based on time constraints.

 

 

Steve Willner was responsible for the original insight that an SoS

correction would have a impact the accuracy of the Swiss Team format.

All of the coding and simulation work (read this as the "real" work)

was done by Alex Ogan and Gerben Dirksen.

[TimG]

With no SoS adjustment, tournament organizers need to run twelve 20 board rounds to have a 95% chance that the strongest team will place in any one of the top eight slots. (Simply put, Swiss Teams are a real crap shoot)

That study uses a field size of 128 teams and does not make any mention of the differences in skill levels of the top teams. Are the top 10 team nearly equal? Top 16? How much better than the field was the top team? How much better than the 2nd or 8th best teams?

 

It would be interesting to hear how often a non-top 10% team won the 128 team simulated Swiss. AWM has claimed that a mediocre team can win a Swiss with a good ("lucky") last round match.

 

The biggest Swiss event I could find in the last year's worth of New England regional results was 61 tables.

 

I've played in a lot of small Swisses (that's all there are in Maine). My sense, purely anecdotal, is that the winners are almost always one of the top teams in the event -- very rarely does a mediocre team (relative to the field) end up on top. However, a mediocre team having a good day, perhaps bolstered by an easy last round draw, will often sneak into the overalls.

[hrothgar]

Hi Tim

 

The field size for that study was based on the big Swiss Teams events that take place in the Gold Coast congress down in Oz. I agree that field size could have a significant impact on the results. With this said and done, this thread started out discussing Swiss Pairs. All other things being equal, I’d expect that the field size for a Swiss Pairs event would average about double that of a Swiss Teams.

 

As I recall, the study assumed that the strength of the field was normally distributed; however, it’s been a while since I played with this (I keep meaning to reimplement this all in MATLAB). In the mean time, you’ll probably need to direct specific questions to Gerben…

[TimG]

The field size for that study was based on the big Swiss Teams events that take place in the Gold Coast congress down in Oz. I agree that field size could have a significant impact on the results. With this said and done, this thread started out discussing Swiss Pairs. All other things being equal, I’d expect that the field size for a Swiss Pairs event would average about double that of a Swiss Teams.

Swiss Pairs at an ACBL sectional, I believe. I would expect the field to be much smaller than the Gold Coast Congress, especially considering ACBL's propensity to flight events.

 

Perhaps Gerben is reading this thread...

[ArtK78]

Based on attendance at sectionals in our area, and including the fact that there will be single-session sectionally rated side games opposite the two-session Swiss Pairs, I would expect a turnout of between 10-20 tables for the Swiss Pairs.

[jdonn]

Despite that imps more random than mps and pairs more random than teams, I still don't find this form of scoring particularly random relative to what I want. Especially considering the inherent randomness in bridge anyway.

 

Besides there are other goals besides 'best team wins as much as possible'. For example the best players are probably most interested in playing the strongest opponents possible. The worst players are probably most interested in not getting creamed every round. The medium players probably enjoy a format where they have just a slight chance to win if the stars align. Even spectators, if there are any, are probably more interested in seeing very strong matchups decide the winner at the end of the event. So there is something for everyone in this format, aside from the fun since it essentially doesn't exist around these parts.