Swiss Pairs

[jdonn]

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.

And I would add as a question, what makes the team in second with a round to go any more deserving of coming in second place if they lose the last round to the team in first than the team in third is if they win the last round against the team in fourth (by enough)? Would it be better if the team in second played the team in second to last and earned the victory only if they smashed them by more than the team in first smashed the team in last? I would barely even call that event bridge.

 

Btw you are assuming the field looks like this with a round to go:

 

1. Best team

2. Second best team

3. Mediocre team

4. Mediocre team

5. Mediocre team

....

 

Please tell me you see quite a few large problems with making those assumption...

[wank]

lol the swiss pairs at the english summer congress can get 250 tables in a good year. i don't know why it's not so popular in america.

 

i don't collect them myself but mugs love these events because the ebu awards a tiny fraction of a premium master point for every match won or drawn.

 

from what i've seen, americans love masterpoints even more than english mugs - i've seen amercans describe themselves as 1000 point players or whatever, whereas most people here have no idea what their masterpoint rank is, just that they want more points- if the acbl want to make these events more popular, crank out the masterpoint awards.

[gnasher]

Given the obsession with masterpoints amongst ordinary players in the ACBL, this format ought to be very popular.

[TimG]

Given the obsession with masterpoints amongst ordinary players in the ACBL, this format ought to be very popular.

It requires duplicated boards, doesn't it? The ACBL has been very slow to switch to machine duplicated boards; the last sectional I played in we hand duplicated boards and that was just a couple months ago.

 

In Swiss team events, the boards are typically hand shuffled and dealt prior to each match.

[mrdct]

In Swiss team events, the boards are typically hand shuffled and dealt prior to each match.

It would be at least 20 years since I last saw a swiss teams event with hand shuffled boards.

[Siegmund]

There was one place that Swiss Pairs was commonly used in the US - in the "homestyle" games, non-duplicate games where you played 4-board matches against another pair and converted the result to VPs after each round.

 

They were a failed experiment to draw rubber bridge players into the club. They are still legal, and still award masterpoints at the usual club game rate, plus .06 per match won. They are somewhat less serious than the average game of Go Fish.

 

Myself, I would be delighted to never see a Swiss movement anywhere, pairs or teams, bridge or backgammon, for the rest of my life.

[MFA]

I like the format. It's cool that the pairs in contention battle it out face to face eventually. Of course one can't expect all pairs to finish in exact order of bridge skills, but that's the case in any short tournament. In swiss pairs it is not all just a competition in slaughtering the bad pairs the most.

 

Icelandair Open is a nice, annual tournament with this format.

[gnasher]

In Iceland and Norway they play a Swiss movement with 3- or 4-board rounds, whereas in England (and now Delaware) there are a smaller number of 8-board rounds.

 

One of BBO's competitors also used to run its pairs events on a Swiss basis, with two-board rounds. I don't know if they still do that.

 

There are advantages to both approaches: playing lots of short rounds means that by the end of the event all the pairs in contention will have played much the same set of opponents; long rounds increase the importance of the match between the two leading pairs on each round.

 

I think I prefer the Nordic style, as long as there are enough good pairs to go around. That might just be the novelty value, though.

[Vampyr]

In Swiss team events, the boards are typically hand shuffled and dealt prior to each match.

It would be at least 20 years since I last saw a swiss teams event with hand shuffled boards.

In Europe, however, they are quite common. And they use Victory Points there as well. I have often wondered about the use of VPs in an event where the boards are hand-shuffled. My feeling is that the randomness added by the shuffled boards makes this inappropriate, and that win/loss should be used in such events. (The alternative would be a complicated VP scale involving factoring).

[WellSpyder]

Btw you are assuming the field looks like this with a round to go:

 

1. Best team

2. Second best team

3. Mediocre team

4. Mediocre team

5. Mediocre team

....

I know, let's just cancel the final round and then we will have teams in the desired order!

 

Slightly more seriously, I don't think the argument is that the teams WILL be in that order, just that it is the most likely order. But of course if that is the most likely order after n-1 rounds, it is also the most likely order after n rounds.

[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)

 

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%)

This doesn't sound to me like a crap shoot. At least not any more than most other forms of bridge. In a 64 table MP pairs, I wouldn't expect the best pair in the field to always be in the top 8; there are upsets every year in the Spingold and Vanderbilt.

[awm]

Here's an experiment which will perhaps illustrate the issue.

 

Take the results from a real swiss event (pairs, teams, whatever).

 

Consider two teams which finished close to each other in the standings. We probably only care about teams that finished close to the top, so consider two such teams (say 1st and 2nd place).

 

For each of these teams, we can compute the "strength of schedule" that they faced, in the simple way proposed. All this means is, you take all the opponents faced by each team and sum up their final victory points. To remove the effect of facing each other (if they did play each other) I suggest ignoring the teams in question. So:

 

For team one, compute the total of the final VPs of all teams they played except for team two.

For team two, compute the total of the final VPs of all teams they played except for team one.

 

My claim is that if team one has a substantially weaker strength of schedule according to this formulation, then something has gone wrong. The team that placed second can easily be a better team, yet they had to play against much stronger opposition and therefore couldn't rack up as many VPs.

 

I suggest that this situation (1st place team has substantially weaker schedule than 2nd place team) is not at all rare in a swiss event. This issue doesn't exist in "round robin" type formats.

 

With that said, I agree that sometimes "random events" are fun (they give more people the opportunity to win or place high) and there is some benefit to people getting to play teams of "comparable level" as much as possible. There is absolutely nothing wrong with running a swiss pairs and I'd probably play in one occasionally. Variety is certainly good.

 

On the other hand, I think there is a problem if events which are "less random" (i.e. more skill) are being consistently replaced with events which are "more random." I would not like to see all the MP pair events (or even a substantial majority) replaced by swiss IMP pairs (as seems to be the case in some regions).

[barmar]

In Swiss team events, the boards are typically hand shuffled and dealt prior to each match.

It would be at least 20 years since I last saw a swiss teams event with hand shuffled boards.

In Europe, however, they are quite common. And they use Victory Points there as well. I have often wondered about the use of VPs in an event where the boards are hand-shuffled. My feeling is that the randomness added by the shuffled boards makes this inappropriate, and that win/loss should be used in such events. (The alternative would be a complicated VP scale involving factoring).

It's also still common in the US.

 

There are some clubs that use preduplicated boards in Swiss teams (or small clubs that play a full round robin). But I've never seen them in tournaments. Maybe the final day of the North American Swiss at the NABC uses preduplicated boards.

[Vampyr]

I would not like to see all the MP pair events (or even a substantial majority) replaced by swiss IMP pairs (as seems to be the case in some regions).

I would not like to have any major pairs events to be changed to IMP pairs, whether Swiss or not.

 

I don't think that adjusting results based on some sort of "Swiss points" system is a good idea. For one thing, it seems that you would have to make the adjustment after every round instead of just at the end, so the scoring system would be opaque and not comprehensible to the players.

 

Another problem is that early in the event there will be many pairs on the same scores, so there will be a lot of randomness in whom you are drawn against, so the method seems rather unfair.

 

One year in the Brighton Swiss Teams, we started the Saturday morning with 10 VPs on a 30 average, because one of our team was a bit the worse for wear on the Friday evening. On Saturday we sliced through the field like a knife through butter, and ended up in the secondary final.

 

Would it have been "fairer" for us to miss out, since our Swiss points would not have been very many? Would it have made it a "better" event? What do these words mean?

[paulg]

I like the format. It's cool that the pairs in contention battle it out face to face eventually. Of course one can't expect all pairs to finish in exact order of bridge skills, but that's the case in any short tournament. In swiss pairs it is not all just a competition in slaughtering the bad pairs the most.

 

Icelandair Open is a nice, annual tournament with this format.

The Icelandair, in common with one of the Scottish tournaments, does not convert a match to victory points but you keep your matchpoint score - barometer (swiss) pairs is a better description of this event.

 

I think it is a great format and better than Swiss Pairs converted to VPs.

[hrothgar]

Would it have been "fairer" for us to miss out, since our Swiss points would not have been very many? Would it have made it a "better" event? What do these words mean?

I don't recall anyone using the word "fairer" to describe a Strength of Schedule adjustment; not did anyone use the expression better.

 

I made a very specific claim: Incorporating a Strength of Schedule adjustment increases the accuracy of a Swiss Teams format.

 

If you don't really care which team places where, don't use one. If, on the other hand, you think that its valuable that a tournament produces an accurate ranking of the skill level of the various teams...

[hotShot]

The swiss format is a theoretical equivalent of a sorting algorithm, that will bring everything into the correct order within log2( Number of objects ) steps.

 

To work properly it is necessary that the comparison function can give a true answer about which object is "larger" than the other. Unfortunately the remaining randomness of bridge distorts that comparison function.

On a small number of boards, the better team/pair will not necessarily win.

 

The algorithm is designed to use only wins and losses. The amount of a win should not mean anything. So using IMPs, VP or MPs for more than to determine the winner, leads to an additional distortion of the result.

 

To make a good swiss you should only count the wins and losses to determine the ranks. If everybody is playing duplicated boards, one could use the MPs/IMPs as a tiebreaker for those who have the same number of wins, but it will reintroduce a little more randomness to the final result.

[hrothgar]

To make a good swiss you should only count the wins and losses to determine the ranks. If everybody is playing duplicated boards, one could use the MPs/IMPs as a tiebreaker for those who have the same number of wins, but it will reintroduce a little more randomness to the final result.

"Good" is defined how?

 

Also, do you have any kind of analysis that backs up this assertion? (It seems very counter intuitive to me)

 

I can certainly believe that it would be better to use BAM scoring on individual boards. However, I don't think that the same logic should carry over to entire matches.

[helene_t]

The swiss format is a theoretical equivalent of a sorting algorithm, that will bring everything into the correct order within log2( Number of objects ) steps.

[.....]

The algorithm is designed to use only wins and losses. The amount of a win should not mean anything. So using IMPs, VP or MPs for more than to determine the winner, leads to an additional distortion of the result.

 

To make a good swiss you should only count the wins and losses to determine the ranks. [....]

It's not the idea of Swiss that it should resemble a sorting algorithm.

 

The idea is that having two top pairs play each other provides more information about their relative strength than having them both playing against weaker pairs. Also, if each table has a NS pair and an EW pair of approximately the same strength, the strength of the EW line and the NS line will be similar so the comparison of an NS score to an EW score is fair.

 

I am pretty sure that using IMPs is more efficient than counting wins/losses.

[mycroft]

Swiss works wonderfully as a sorting algorithm - but only for the extreme ends. Not perfectly accurately, true, but for "on the day".

 

If you run N rounds for 2^N pairs (in WIN/LOSS), you'll get a top team and a bottom team. Top team wins. You can't say with any certainty at all who's second, however. Overswiss a round, and you get #2. Another round, you get 3 and 4.

 

However, if you leave the sorted-out pairs in the swissing, you muck up the results (especially if you enforce a no-playback rule). So you need to remove the winner and last place from the game after N rounds, and the next pair after N+1, ... Nobody's going to go for that.

 

So we overswiss and leave the polluting pairs in the mix; the "fair" movement is damaged. It's damaged more (in a purely theoretical sense) by VPs (although that removes the "single-elimination" feel of a pure W/L swiss), more yet by non-duplicated boards, and so on...all of which works to the advantage of the pairs that aren't the best. You know, N-1/N% of the paying customers?

 

Curling bonspiels do a couple of interesting formats - either N games guaranteed (for 3: A (winners), B (lost first match), C (lost second match, irrelevant of what happened in 1), producing an A winner, a B winner, and a C winner) or N-ple knockout (where when you lose the first match, you drop into B, when you lose in B, you drop into C.) Notice how both of these require the A winner to play fewer (frequently much fewer) games than the C winner (in curling, this is considered a plus from the A winner's perspective - more drinking time).

 

(Note: qualifiers, unlike bonspiels, tend to work as a full-RR into Page playoffs: 1v2, 3v4; winner of 3v4 plays loser of 1v2 for the right to play the 1v2 winner in the final. To determine a "best team", this actually works very well; you have to have the "win the big games and against the fishes" skill to lead the RR because being 1 or 2 is a distinct advantage (two wins v three, can lose the first one if you come back to beat them on the revenge), but you still have to win two or three big games to win the event. It only works with prequalifiers, though, as the size of the RR has to be limited).

 

But all of that theory is irrelevant in Real Life. In Real Life, people like Swiss events because they like playing against people of their level (and swissing does do a good job of "bucketing" skill levels fairly quickly), and because of match awards. Swiss pairs has the added benefit that you don't have to find teammates you're willing to play with all day - I'm surprised it hasn't just exploded here (I guess BKOs will still award more monsterpoints, and the one-session swisses, well you have a team already; and Sunday will never change. Oh well). Also, the fact that it's a poor method of determining the winner is a plus for the 3rd-6th (or so) teams, who have a good chance of winning (and a good chance of being 10th, but still).

[Gerben42]

I think anyone who has played Swiss Teams knows that it's extremely random. I still feel sorry for Fluffy for his Estoril tournament, for example.

 

One of my positive Swiss experiences was the mixed teams in Pula. I ended up 2nd although I think my team wasn't remotely among the best 10. But we got easy opponents and some luck in the last round.

 

HotShot has shown the decisive point. In a swiss tournament, you cannot have VPs for it to be fair.

 

Consider two teams which finished close to each other in the standings. We probably only care about teams that finished close to the top, so consider two such teams (say 1st and 2nd place).

 

For each of these teams, we can compute the "strength of schedule" that they faced, in the simple way proposed. All this means is, you take all the opponents faced by each team and sum up their final victory points. To remove the effect of facing each other (if they did play each other) I suggest ignoring the teams in question. So:

 

For team one, compute the total of the final VPs of all teams they played except for team two.

For team two, compute the total of the final VPs of all teams they played except for team one.

 

My claim is that if team one has a substantially weaker strength of schedule according to this formulation, then something has gone wrong. The team that placed second can easily be a better team, yet they had to play against much stronger opposition and therefore couldn't rack up as many VPs.

 

I've done that for the Estoril transnationals and the result was quite shocking.

[Finch]

Short (one day, so up to 50 board) Swiss pairs events in the format played in England i.e.

- 7x7 or 8x6 or 6x8 matches

- matchpoints converted to VPs

- assignments one round in arrears

 

are very very random. Just look at all the 1-day county events and see how lots of different pairs win them, including pairs who you don't see doing well more than once, or in any other type of event.

 

Longer events (2-3 day ones) and/or not converting to VPs and using current match assignments are much better.

 

It's a pity that the EBU has reduced the amount of current-match assignmenting it does.

[gordontd]

It's a pity that the EBU has reduced the amount of current-match assignmenting it does.

Assigning? :P

[Mbodell]

I think Swiss pairs are a fun sounding format, and I certainly enjoyed them online (MP or IMP, it doesn't really matter either can be the basis of swiss pairs).

 

In the ACBL there are three obvious problems:

 

1. They are unfamiliar. So you are less likely to see them.

 

2. They require pre-duplicated board sets to play more or less barometer style. This should be done for all events IMO (for fairness and event integrity - certainly more important than cell phone bans), but it is a problem for organizers in the ACBL tradition.

 

3. I've heard the master point awards are off in that the master point awards are based on number of tables more so than number of players so you will not award as many points in a 400 person swiss pair than you do for a 400 person swiss team. This is problematic if your goal as event organizer is to maximize the number of master points that are awarded.

 

But really, I think it is good that different locations are offering different events.

 

In terms of making swiss teams more fair and accounting for SoS, that's what the ?Zeligmo? (I forget the spelling) method does where it uses the swiss results recorded to then "simulate" what would happen if a complete round robin were to be played. The results of that calculated full round robin are then used to compare the team. I'm sure someone here remembers the article in question which studied one of the big European team competitions.

[Vampyr]

Short (one day, so up to 50 board) Swiss pairs events in the format played in England i.e.

- 7x7 or 8x6 or 6x8 matches

- matchpoints converted to VPs

- assignments one round in arrears

 

are very very random.  Just look at all the 1-day county events and see how lots of different pairs win them, including pairs who you don't see doing well more than once, or in any other type of event.

 

But... by adding all of these one-day county events it's not as if the EBU are eliminating the longer ones. And the "lots of different pairs winning" cannot be a drawback in terms of getting bums on chairs. Top players like you, Frances, need to remember that there has to be events in which the little guys can enjoy themselves and hope to win, or there will be no events at all.

 

Longer events (2-3 day ones) and/or not converting to VPs and using current match assignments are much better.

 

It's a pity that the EBU has reduced the amount of current-match assignmenting it does.

It is a shame. And why is it so difficult? They could post the assignments around the room as they do in Swiss teams, or just project them onto a wall.