[Vampyr]

I recently played in a Multiple Teams event where there were 30 boards in play, of which each team played 24. I realise that organisers do not always have a sufficient number of boards to prevent this (although I think the organisers in question should have known, because they knew how many teams were going to be competing).

Anyway, I was wondering two things:

1. What proportion of boards- or opponents-not-in-common makes a competition worthless (by some useful mathematical criteria)?

2. Would VPs help a bit with evening out the scores, and are VPs sensible at all in matches with four or fewer boards?

[barmar]

I'm not sure this is appropriate for the Laws forums, as the format of tournaments is not addressed at all in the Laws.

FYI, most Swiss Team and Knockout events in ACBL do not have any boards in common among different matches; the hands aren't even pre-duplicated, they're shuffle-and-play. The main exceptions are the late rounds in national events such as the Spingold and Vanderbilt KOs. Even the national Swiss Team tournaments do not use common boards; the only teams that play pre-duplicated boards are the leaders, because they may be on Vugraph.

[Vampyr]

barmar, on 2013-October-21, 17:56, said:

I'm not sure this is appropriate for the Laws forums, as the format of tournaments is not addressed at all in the Laws.

I know. Perhaps there should be a forum for movements and the like. "Offline Bridge" seems too broad.

Quote

FYI, most Swiss Team and Knockout events in ACBL do not have any boards in common among different matches; the hands aren't even pre-duplicated, they're shuffle-and-play. The main exceptions are the late rounds in national events such as the Spingold and Vanderbilt KOs. Even the national Swiss Team tournaments do not use common boards; the only teams that play pre-duplicated boards are the leaders, because they may be on Vugraph.

I know this, but it does not answer my question. (In the EBU, the early and middle rounds of most knockout competitions are played privately, so they don't use duplicated boards either. I think that this is mainly OK for knockouts.)

[gordontd]

Vampyr, on 2013-October-21, 17:24, said:

I recently played in a Multiple Teams event where there were 30 boards in play, of which each team played 24. I realise that organisers do not always have a sufficient number of boards to prevent this (although I think the organisers in question should have known, because they knew how many teams were going to be competing).

You seem to misunderstand a lot, to the extent that I wonder whether you are looking for things to complain about.

Clearly we did not know how many teams were competing, since we had to expand the number of sections in play at the start of the event.

Secondly, I wonder what you would have done differently with as many boards available as you like? Remember the number of boards to be played was fixed, as advertised, at 24. 25 teams to play 24 boards - your choice.

Thirdly, in multiple teams events it is always the case that there is at least one more board-set in play than the number of teams. In this case the teams playing three-board rounds had one more set in play than the minimum possible, and those playing two-board sets had four more boards in play than the minimum - not anything that anyone else would consider the least bit problematic - and nor did they.

[PhilKing]

Vampyr, on 2013-October-21, 17:24, said:

Anyway, I was wondering two things:

1. What proportion of boards- or opponents-not-in-common makes a competition worthless (by some useful mathematical criteria)?

The answer could be that 100% is fine. The equation would involve not just the proportion of boards, but the number played (for example, if every team played one million boards in a marathon round-robin, luck would clearly even out even if different boards were played in each match).

For smaller numbers, the nature of the boards themselves becomes a critical variable factor. Say the boards missed were all swing hands favouring accurate bidding judgment or play, that would make a difference. You can't pluck "24.6%" out of a hat - the answer to your question is "nobody knows".

With every side playing 24 boards, the result would rarely be "worthless" even if no teams played the same boards in different matches, but obviously some teams will have the "advantage" of swingier boards. Think of a Swiss in the old days with boards dealt at the table.

[gnasher]

Another consideration is the number of teams that qualify for the next round/win prizes/get masterpoints. For example, if it's qualifier for a final with a fixed number of teams, the benefit of having swingier boards increases as the number of qualifiers reduces.

[gnasher]

gordontd, on 2013-October-22, 01:37, said:

Secondly, I wonder what you would have done differently with as many boards available as you like? Remember the number of boards to be played was fixed, as advertised, at 24. 25 teams to play 24 boards - your choice.

Couldn't you have 24 one-board rounds, with one set of 25 boards in play?

[gordontd]

gnasher, on 2013-October-22, 04:50, said:

Couldn't you have 24 one-board rounds, with one set of 25 boards in play?

Damn! How could I have overlooked that?

[MickyB]

Vampyr, on 2013-October-21, 17:24, said:

1. What proportion of boards- or opponents-not-in-common makes a competition worthless (by some useful mathematical criteria)?

As Philking has said, there is no answer to this question. It is not like a pairs event, for which you might be able to draw a conclusion that one movement leads to a result that has twice as much significance as another movement.

[helene_t]

Vampyr, on 2013-October-21, 17:24, said:

1. What proportion of boards- or opponents-not-in-common makes a competition worthless (by some useful mathematical criteria)?

I don't think that 24/30 is much of an issue. In theory an above-average team could be disadvantaged if the six boards they didn't play happened to be "swingy" but this is only true to the extent that swingy-ness means that the skill-to-imp coefficient is higher. In county leagues we compare sums of VP scores between teams that have played different boards. Having an overlap of at least 18/24 between any two teams already improves a lot on this.

Quote

2. Would VPs help a bit with evening out the scores, and are VPs sensible at all in matches with four or fewer boards?

Using the new VP scale without rounding effects it is probably not so bad to use VPs for short matches, and in a very heterogenous field you have the advantage that the head-to-head matches between the strongest teams become more influential relative to the high imp scores obtained by bunny bashing. But I would expect that in general, raw IMPs give better statistical power. Assuming that imps are normal distributed with a mean that is equal to the skill difference between the two teams and constant SD, the best you can do is to simply to add them up. Assuming a thicker-tailed distribution of the IMPs, the VP scale might be better.

[NickRW]

helene_t, on 2013-October-22, 05:43, said:

Assuming a thicker-tailed distribution of the IMPs, the VP scale might be better.

Well, given the NGS ratings seem to be "thick tailed" then indeed perhaps VPs are better.

[campboy]

I definitely think it is right to cap the margin of victory at some plausibly-attainable* amount in any situation where it matters. Whether you do that by some sort of VP scale or simply cutting it off at some maximum is a matter of taste.

*ie the trivial upper bound of 24*number of boards doesn't count, even though it is finite

[Vampyr]

gordontd, on 2013-October-22, 01:37, said:

You seem to misunderstand a lot, to the extent that I wonder whether you are looking for things to complain about.

Who's complaining? I just asked a question. And if I misunderstand, I am grateful when people clear it up for me.

Quote

Clearly we did not know how many teams were competing, since we had to expand the number of sections in play at the start of the event.

I hadn't thought about people who showed up only for the teams and hadn't pre-registered.

Quote

Secondly, I wonder what you would have done differently with as many boards available as you like? Remember the number of boards to be played was fixed, as advertised, at 24. 25 teams to play 24 boards - your choice.

I don't know, but I was surprised as the EBU rarely have excess boards in play.

[gordontd]

Vampyr, on 2013-October-22, 16:16, said:

I don't know, but I was surprised as the EBU rarely have excess boards in play.

That's true, and it's something I try to ensure as far as possible. Perhaps you've been spoiled by me?

But sometimes it's unavoidable, and it certainly doesn't mean the teams are each playing in a different competition.

[cherdano]

I can sympathize with the OP. When I look through the hand records after a club evening, the boards I didn't play are always the ones where I would have done great.

[Vampyr]

gordontd, on 2013-October-23, 01:49, said:

That's true, and it's something I try to ensure as far as possible. Perhaps you've been spoiled by me?

There is no doubt that this is true. I find EBU competitions superior to those in any of the other dozen or so countries I have played in.