Swiss Pairs mis-matches

[Vampyr]

I am at a loss on how anybody can state that the results in rounds 1, 2, 3 and so on has no impact on later rounds if Match Point scores were converted to VP scores so long as the outcome of the event is decided by the accumulated sum of whatever kind of scores won in all rounds during the entire event.

 

Maybe you saw what I wrote above which is that you get your victory points at the end of a match, but the boards are now completely finished, and your good or bad boards in a completed match will not affect your score anymore. Do you play Swiss Teams in Norway, and use VPs? It is like that. You can go for 1700 against nothing, but after the match in question this will not be part of your score for the event. And you may win the match as well.

 

When raw percentages are used in Swiss Pairs, does the concept of winning or losing a match exist?

[Vampyr]

Anyway, our experience is that the final results based on Match Points calculated as specified in Law 78A extremely seldom extend outside the range 40% - 60%. So the alleged advantage by converting Match Points to Victory Points seems rather obscure to me.

 

I think you will find that in an eight-board match there is a greater range of scores.

[mycroft]

I still don't get it. If the boards are never played again, the 62% which becomes 18 VPs could be converted to 7.5 Final% (8 7-board matches) instead. "After the match the score will not affect anything ever again", again, no? Or the 38% that becomes 2 VPs became 4.5 Final% instead?

 

I can see an issue if your 55% on one board used for pairing turns into a 45% by the end of the night (or vice versa) - but that doesn't happen in Swiss Pairs, does it?

 

I understand Campboy's wish to limit the difference in the seeding rounds between 94% and 86% based on whether you got the truly horrible or merely the horrible in the two rounds before you're playing your kind of teams - and I can understand not discounting those rounds or the Swiss gambit will work even better - but I can't tell the difference between a matchpoint score that won't change after the round's over and a converted-to-VP score that won't change after the round's over - especially given that the potential value of the match is identical (which it isn't, even with duplicated boards, in an IMP event).

 

Obviously I'm missing something.

[Vampyr]

I still don't get it. If the boards are never played again, the 62% which becomes 18 VPs could be converted to 7.5 Final% (8 7-board matches) instead. "After the match the score will not affect anything ever again", again, no? Or the 38% that becomes 2 VPs became 4.5 Final% instead?

 

I can see an issue if your 55% on one board used for pairing turns into a 45% by the end of the night (or vice versa) - but that doesn't happen in Swiss Pairs, does it?

 

I understand Campboy's wish to limit the difference in the seeding rounds between 94% and 86% based on whether you got the truly horrible or merely the horrible in the two rounds before you're playing your kind of teams - and I can understand not discounting those rounds or the Swiss gambit will work even better - but I can't tell the difference between a matchpoint score that won't change after the round's over and a converted-to-VP score that won't change after the round's over - especially given that the potential value of the match is identical (which it isn't, even with duplicated boards, in an IMP event).

 

Obviously I'm missing something.

 

Well, suppose you get a zero on one board, but still win the match 20-0. For example. So that zero has no effect at all on your score.

 

Edit: Campboy plays in a country where there is only one random pairing; one-round-in-arrears scoring was scrapped by the EBU years ago.

[pran]

I still don't understand what is meant by "match" in a barometer style event for pairs.

 

But I can obviously clarify some of the questions asked:

 

Yes, we have Swiss events for Teams as well, in one of mine the setup was:

20 teams (of four) met in 9 sessions playing one match (against one other team) each session. Each match consisted of two rounds with 12 Boards each, and in all matches the same 12 boards were played simultaneously during the same round. The scoring was standard IMP converted to VP.

 

For each session after the first the teams were seated as follows:

Table 1: The currently leading team against the best placed team they had not yet met

Table 2: Same procedure applied to the teams not yet seated

Table 3: Same procedure again - and so on.

 

I believe this is "standard procedure" for all Swiss teams competitions?

 

A similar preocedure is followed for our Swiss for Pairs events, but we have no such concept as "matches" there.

 

A typical setup is:

 

50 pairs* meet in 16 rounds playing 4 Boards in each round (the same boards simultaneously at all tables!). Seatings are allocated randomly for the first two rounds.

Scoring in each round is standard matchpoint scoring with scores on each board ranging from +49 to -49

For each round after the second the pairs are seated as follows:

Table 1: The currently leading pair (after the last round scored) against the best placed pair they have not yet met

Table 2: Same procedure applied to the pairs not yet seated

Table 3: Same procedure again - and so on.

 

So for instance during round 4 we have just scored out round 3 and determined the seatings in round 5. Result slips are now handed out to each pair so that they learn their current score and where to go for round 5.

 

* We have had events with more than 200 pairs, see for instance:

http://bridge.no/var/ruter/html/9901/2014mixpar.htm (213 pairs, 30 rounds, 3 boards/round)

[Vampyr]

but I can't tell the difference between a matchpoint score that won't change after the round's over and a converted-to-VP score that won't change after the round's over - especially given that the potential value of the match is identical (which it isn't, even with duplicated boards, in an IMP event).

 

Obviously I'm missing something.

 

No, it is my very evident inability to explain it, but perhaps there is another way of looking at it.

 

Suppose Team A score 89% in their first match and Team B score 69%. They will both win their match 20-0 in terms of VPs, so they will be on the same score. If they are paired up in Match 2 it is as exact equals. (I don't know whether 69% is above the cutoff for 20VPs in a typical-length match, but obviously that detail is irrelevant).

[Finch]

No, it is my very evident inability to explain it, but perhaps there is another way of looking at it.

 

Suppose Team A score 89% in their first match and Team B score 69%. They will both win their match 20-0 in terms of VPs, so they will be on the same score. If they are paired up in Match 2 it is as exact equals. (I don't know whether 69% is above the cutoff for 20VPs in a typical-length match, but obviously that detail is irrelevant).

 

Yes. I don't understand why you think that is a good thing.

[gordontd]

Well, I mean that you get your VPs and the boards no longer have an effect on your scores, for example, you can get a bottom on a board and still have a big win, even 20-0. You can have two bottoms and a big win. Or a tie or whatever. Without VPs, those 0% boards will drag your score down for the rest of the event.

The boards contribute to the score for that round, and that score remains unchanged for the rest of the event (scoring corrections notwithstanding). That applies whether the score was in matchpoints, IMPs or VPs.

[gordontd]

What is the benefit of limiting blitzing in 6 board rounds that is different from the 2, 3, or 4 board rounds?

You are going to play a much smaller proportion of the field and one round will have a much larger effect on your overall score. I once played a local Swiss Pairs where, although it was scored in VPs, the scoring program showed our percentage. On the first round we were drawn against a very inexperienced pair indeed and scored 94%. I think it was better that it was converted to VPs, since it would have been an unreasonable advantage to have be benefit of a 94% score for 1/7 of the event.

[Vampyr]

Yes. I don't understand why you think that is a good thing.

 

Well, Gordon's comment just above is part of the reason, also that since VPs are cumulative your score can go up or down by a lot, which is perhaps a bit of a randomiser, but I find they it makes the event more exciting.

 

I guess it is partly psychological, too; if I have a bad board but still get 18 or 20 VPs (or a well-deserved 0) from a match, then I can move on to the next match more easily. I don't have to try to, nor can I, overcome the effect of the poor board.

 

Also, and this is important, with VPs you are scored against the pair you are playing, with whom you are supposedly evenly matched. If you are continually scored across the whole field then I think that the Swiss effect is entirely lost.

 

I guess you don't like Swiss Teams with VPs either, but these are also widely offered in the EBU so my opinion of these events is probably more prevalent than yours. Maybe your point of view is shared by many of the top players like yourself, but these are not the bulk of the EBU members.

 

I do think, however, that the straight percentages method might be quite fun for a qualifier. As a main event, I really feel it lacks something.

[mycroft]

Okay, I think I understand now. I'm not sure I'm happy that 31% of the matchpoints is worth nothing, but I've always had issues with that in team games as well.

 

It looks like there's two different games, and they both would be interesting.

 

Thank you for the "help me understand" - it helped me understand.

 

Re: one round in arrears - I guessed that (I know nowhere where they do that, at least any more), but it still takes a couple of rounds before you've shaken out the field. In Australia where they (used to?) seed the first round 1 vs 51, 2 vs 52...50 vs 100, effectively round 2 was "random draw 1-50, and random draw 51-100" as the first matches were almost all killers. In our club game, we seed the field differently (A with A, B with B, round 1); so now round 2 is "which A team that won 15-5 gets the C team that won their C match 15-5, and which gets the other A team that won 16-4?") - there are a lot of blitzes round 2 for some reason. Even if it's totally random seeding round 1, there are 4 kinds of matches:

 

• evenly matched good teams
  
• evenly matched weaker teams
  
• mild mismatches
  
• killer mismatches
  

 

and there are a bunch of round 1 medium winners of even A matches that catch medium winners of C matches (and, of course, medium losers of even A matches who catch medium losers of C matches) which tend to bring in predictable results. After a couple of rounds, 3 at the most, things start to run as much to form as they're likely to.

[mycroft]

And sure enough, I look over last night at the next table, where The Round 2 Match featured two 19/20 VP teams: I know Masterpoints Mean Nothing, but as I said to our table: "So, we have 28 thousand masterpoints vs... 28". Okay, I might have been understating for effect, but I know it was less than 80 for the team of 4, because the highest MP rating was 19.

 

I'll let you guess the result of that match :-)

[Vampyr]

Okay, I think I understand now. I'm not sure I'm happy that 31% of the matchpoints is worth nothing, but I've always had issues with that in team games as well.

 

It looks like there's two different games, and they both would be interesting.

 

Thank you for the "help me understand" - it helped me understand.

 

Re: one round in arrears - I guessed that (I know nowhere where they do that, at least any more), but it still takes a couple of rounds before you've shaken out the field. In Australia where they (used to?) seed the first round 1 vs 51, 2 vs 52...50 vs 100, effectively round 2 was "random draw 1-50, and random draw 51-100" as the first matches were almost all killers. In our club game, we seed the field differently (A with A, B with B, round 1); so now round 2 is "which A team that won 15-5 gets the C team that won their C match 15-5, and which gets the other A team that won 16-4?") - there are a lot of blitzes round 2 for some reason. Even if it's totally random seeding round 1, there are 4 kinds of matches:

 

• evenly matched good teams
  
• evenly matched weaker teams
  
• mild mismatches
  
• killer mismatches
  

 

and there are a bunch of round 1 medium winners of even A matches that catch medium winners of C matches (and, of course, medium losers of even A matches who catch medium losers of C matches) which tend to bring in predictable results. After a couple of rounds, 3 at the most, things start to run as much to form as they're likely to.

 

Yeah, one round in arrears is rubbish. But anyway, I think that you might have better results if you didn't seed the field. We don't here and there don't seem to be huge mismatches in round 2. Or you could separate the different levels into their own fields. As far as the VPs go, well some people like them and some don't. There are plenty of events for both types of people.

 

Of course, there are places where VPs are used where the same boards are not played at all the tables, and I don't know if anyone likes these!

 

EDIT: By the way, when the points won are limited to 20, the team that lost 20-0 is still in the event.

[TimG]

Might it be the case that total MP is fine for events with short rounds but that VP is better for events with longer rounds?

 

Does anyone know of an online reference for Swiss Pair events? I've done some googling tonight, but can't find much of anything.

 

Most of the items I have found talk about pairing being staggered one round. Isn't this a thing of the past with the use of Bridge Mates?

[gordontd]

Might it be the case that total MP is fine for events with short rounds but that VP is better for events with longer rounds?

I think that's true. The advantage of total MP is that each board has equal value. In a match converted to VPs the later boards in the match will have greater or lesser value depending on how the match has gone thus far. The advantage of VPs is that a very mis-matched first round doesn't carry over a disproportionate effect.

Does anyone know of an online reference for Swiss Pair events? I've done some googling tonight, but can't find much of anything.

Did you find the basic articles on David Stevenson's website?

 

Most of the items I have found talk about pairing being staggered one round. Isn't this a thing of the past with the use of Bridge Mates?

Yes, it has largely been supplanted by the use of Bridgemates.

[pran]

Might it be the case that total MP is fine for events with short rounds but that VP is better for events with longer rounds?

 

Does anyone know of an online reference for Swiss Pair events? I've done some googling tonight, but can't find much of anything.

 

Most of the items I have found talk about pairing being staggered one round. Isn't this a thing of the past with the use of Bridge Mates?

To answer your last question:

Staggering one round allows the next round seating to be known immediately at the end of each round while not staggering imposes an extra delay between rounds, even with Bridgemates. This delay per round easily amounts to the time otherwise allowing another board per round.

[gordontd]

By the way, for anyone who was interested in the original topic of this thread, we did end up adding a regulation for mis-matches in Swiss Pairs scored by matchpoints - WB 3.5.3

[gordontd]

To answer your last question:

Staggering one round allows the next round seating to be known immediately at the end of each round while not staggering imposes an extra delay between rounds, even with Bridgemates. This delay per round easily amounts to the time otherwise allowing another board per round.

My experience is that the additional time is much less than this. However, when the rounds are short and there are a lot of them, the downsides of round-in-arrears are reduced.

[pran]

My experience is that the additional time is much less than this. However, when the rounds are short and there are a lot of them, the downsides of round-in-arrears are reduced.

I figure 7 minutes per board. Instead of moving directly to their next tables all players must wait for the information on their new seatings to be distributed and understood before they can move. How long does that take according to your experience?

 

(My experience is that players who have completed a round early often leave the table for various reasons, do you have some kind of regulation requiring all players to remain patiently at their tables until they are told where to go next?)

[gordontd]

I figure 7 minutes per board. Instead of moving directly to their next tables all players must wait for the information on their new seatings to be distributed and understood before they can move. How long does that take according to your experience?

 

(My experience is that players who have completed a round early often leave the table for various reasons, do you have some kind of regulation requiring all players to remain patiently at their tables until they are told where to go next?)

Your formulation is that a three-board round would take 28 minutes with current-round assignments, whereas I would expect it to be about 25 minutes.

 

We often have a monitor displaying the assignments for those who have left their tables before the end of the round.

[pran]

Your formulation is that a three-board round would take 28 minutes with current-round assignments, whereas I would expect it to be about 25 minutes.

 

We often have a monitor displaying the assignments for those who have left their tables before the end of the round.

With medium experienced players we allocate 7 minutes/board plus 2 minutes for the round shift, or 23 minutes/round with 3 boards.

 

Our scoring program offers a facility to perform the draws for next round once 95% of the scores in the current round have been entered, thus avoiding the staggering. To my knowledge this facility is little used (if any at all) here and our players apparently have no problem with the way we do things.

 

Our experience and simulation has confirmed that staggering has little impact provided the number of rounds is at least 8.

[TimG]

Did you find the basic articles on David Stevenson's website?

Yes, I did find two articles on David's website. They were helpful to get a basic understanding, but I am now looking for something more.

 

A few questions:

 

1) I am in the US where players are used to playing 2 or 3 boards per round (24-28 in a session) in club play, often in a Mitchell movement. I think that players (especially at the club level) enjoy the social aspect of frequently switching opponents. I think they would resist a change to Swiss Pairs if they played against only 4 opponents in a session. So, I am particularly interested in how efficiently Swiss Pairs can be run if there are something like 8 rounds of 3 boards each. Does the pairing time add significant time to the event?

 

2) For multiple session events, let's just focus on two sessions for simplicity, is there typically a qualifying session and a finals/consolation session?

 

3) I imagine that including a match element to the scoring helps to retain interest late in an event even for those that are apparently out of the running. But, there must still be some inclination to drop from an event if a pair is doing poorly. Does this something that is acceptable? I can remember times many years ago where it was routine for Swiss Team participants to drop from an event with a couple matches yet to be played. I have not seen this recently, but I wonder if this might be the case when there is not a "complete movement" that must be finished.

 

4) Is there "stratification" in Swiss Pairs? It is the way of the game in the US today.

 

I understand I'm hijacking a thread, I hope no one minds.

[RMB1]

A few questions:

1. With our hardware/software, generating and displaying the assignments/pairings is not a factor. But current-round assigning does not allow for tables to finish late and the pairs catch up in the next round. Round-in-arrears for 3-board rounds copes better with slow tables.
   
2. We usually play through, but there are lots of possibilities for splitting into sections for later sessions.
   
3. Giving up mid-session is "frowned upon" - there can be post-event disciplinary sanctions. We allow pairs to withdraw between sessions and do have stand-by pairs at the start of a session who can make-up a half table.
   
4. We do stratify master-point awards and prizes for overall ranking. For long rounds (6,7,8 boards) we award master-points per "match" won: these are not stratified.
   

I understand I'm hijacking a thread, I hope no one minds.

GordonTD has an answer to the OQ (original question)

[TimG]

1. We do stratify master-point awards and prizes for overall ranking. For long rounds (6,7,8 boards) we award master-points per "match" won: these are not stratified

No masterpoints for shorter matches? Seems reasonable, but also awarding masterpoints for shorter matches (meaning more people win masterpoints) also seems like a possible draw.

[WellSpyder]

I am in the US where players are used to playing 2 or 3 boards per round (24-28 in a session) in club play, often in a Mitchell movement. I think that players (especially at the club level) enjoy the social aspect of frequently switching opponents. I think they would resist a change to Swiss Pairs if they played against only 4 opponents in a session.

I agree people are likely to resist change. My perspective, though, when we play Swiss Pairs once a month at the club, is that in many ways it feels more sociable playing 8 boards against the same opponents, giving time for more interaction and a decent contest, rather than just a random brief interlude in an overall evening of chopping and changing.