[NickRW]

Siegmund, on Feb 26 2010, 07:32 PM, said:

IS there any defensible reason for teams of 6 being standard for a 4-player game?

Even in our local county league where matches are privately arranged single sessions of 24 boards taking place over (mainly) the colder months, most team captains prefer to have a "squad" of 5 or 6 players to call on (if only for the fact that not everyone will be available for all possible dates.)

As for the bigger events, people generally seem to be of the opinion that they don't want stamina to be a big part of the picture. Your assertion about stamina being an important part in other (team) sports is only partially true as most of them have fairly free rules about substitutions these days.

I just think you have a valid, but minority view.

Nick

[Cascade]

nigel_k, on Feb 27 2010, 09:32 AM, said:

Cricket is the only team game I can think of at the moment where you can't use substitutes.

Cricket has substitute fielders. It is only batters and bowlers that cannot be substituted in.

And there is provision for a runner if the batter sustains an injury so that he cannot run.

[mycroft]

The only thing I would say that hasn't been said is in response to one of Fred's comments about the Spingold/Vanderbilt:

"People tend to play considerably more slowly in big events. Sometimes the bridge doesn't end until 2AM or so"

There's a solution to that. In WC events, and outside the ACBL, the solution has been implemented, and seems to work (with grumbles from two teams).

In NIH-world, the response is "Bridge should be decided by the players, not the directors/clock". But that's just replacing one bridge skill (ability to play at your best while enduring 6 hours of bridge every session) with another (ability to play as close as possible to your "unclocked" bridge skill in 8.5 minutes/board).

I'm very glad the suggestion to restrict the number of boards played per match in the Vanderbilt/Spingold from 64 to 52 failed. It would have accomplished the same outcome (a session=4 hours), but with a *lot* higher increase in the luck factor than getting people to play to time.

But I know I'm in the minority, at least around here. Also, to be clear, I am *not* ragging on Fred - he's just point out what happens, not saying it's the right thing (or that it's not).

[Fluffy]

I would love to play 4 team events representing my country, after all we have troubles finding a decent 4th player, finding 6 is a hard task, so somehow I beleive this helps the countries with not so many top players.

But that is just a different event, playing 4 at a team you will probably end up playing less boards per tournament.

Also it wouldn't be so hard to play teams of 10-12 players, with 4 tables competing with the same boards instead of just 2. Teams of 6 are more popular, but that doesn't mean that they must be the only option.

[Cyberyeti]

The national event where the English/Welsh county representative sides play each other is teams of 8 with a max of 12 players. 4 groups of 9 or 10 with 2 from each qualifying for a separate final. This is played over a weekend with a late finish on the Saturday night.

As one of the smaller counties that qualifies very occasionally, we find we tend to play better as a team of 8 than with more. I know I find that if I sit out in the middle of a session, I don't get my concentration back immediately (so if I have any control I sit out at the beginning or end of sessions).

[TimG]

fred, on Feb 26 2010, 08:22 PM, said:

In a few tournaments things have worked out so that I played either all the boards or only half the boards. I can deal with either of these scenarios if necessary without complaining, but in general I prefer to play 3/4 of the boards.

Is it that you prefer to play 3/4 of the boards or that you prefer to play 48 boards?

If a long event such as the Spingold were scheduled for 48 boards per day, would you still want to play 3/4 of the boards? What about a shorter multi-day team event like the Reisinger?

Would you prefer the two-day and three-day NABC Pairs event to entail just 48 boards per day?

[fred]

TimG, on Mar 6 2010, 01:49 PM, said:

fred, on Feb 26 2010, 08:22 PM, said:

In a few tournaments things have worked out so that I played either all the boards or only half the boards. I can deal with either of these scenarios if necessary without complaining, but in general I prefer to play 3/4 of the boards.

Is it that you prefer to play 3/4 of the boards or that you prefer to play 48 boards?

If a long event such as the Spingold were scheduled for 48 boards per day, would you still want to play 3/4 of the boards?

I don't know. I have been playing 64-board matches in these events forever. Usually I play 3 segments but on occasion I have played either 2 segments or all 4 segments. Like Goldilocks, for me 2 is not enough, 4 is too much, and 3 is just right. I have no idea how I would feel about this issue if the matches were only 48 boards long (though I am quite sure I would not be happy to see this change).

Quote

What about a shorter multi-day team event like the Reisinger?

The typical pattern for a 6-person team with a sponsor is that one of the pro pairs will play all the boards on day 1 and the other pro pair will play all the boards on day 2. If the team plays well enough to make it to the 3rd day (not easy), usually the pros themselves will decide which of their pairs plays both sessions on that day. Typically this decision will be based on sensible factors like who is playing and feeling better, but sometimes factors like seniority and ego get in the way.

I have made it to the 3rd day maybe five times. I have generally not cared that much if I played 1 or 2 sessions - for me just getting to the final is nice. I suppose, if all else was equal, I would prefer to play the entire final, but since I have been fortunate in having strong teammates over the years it is not that big a deal for me to sit on the bench during the first session of the final (the sponsor-pair almost always plays the first session so that is when one of the pro pairs has to sit out). If I am to sit out there is a usually a good reason - most likely that the other pro pair is playing better or is feeling less tired than my pair.

Quote

Would you prefer the two-day and three-day NABC Pairs event to entail just 48 boards per day?

No - I think the current length (52 boards) is fine.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[whereagles]

In a world-class chess match contenders usually play for 3-4 hours a day. In bridge you might get to play twice as much daily. Chess is much more stressing, though. But bridge at high levels isn't exactly relaxing either

[JanM]

Just one quick comment about the number of boards played each day. In the World Championships, they play 48 boards a day, not 64, yet we still find that at the end of the tournament the play often isn't as good as it was at the beginning. Playing all day every day for anywhere from 6 days (Vanderbilt) to 2 weeks (Bermuda Bowl) is very tiring. Add in the possibility of someone being sick or an emergency arising so someone can't play, and it's REALLY hard for a 4 person team to do well.

Also, although part of the reason the Vanderbilt & Spingold end at 1 or 2 am is that there are sometimes people who play too slowly, part of the reason is just that it takes longer with screens. Instead of the normal 7.5 minutes per board without screens, these events are usually scheduled at 9 minutes per board. That means a 16 board quarter takes about 2-1/2 hours, the half takes over 5 hours, because you need a little break between quarters to compare and maybe regroup. Add in a 2 hour dinner break (in the Senior KO in San Diego, the dinner break was only 1 hour and as far as I could tell no one liked that) and you have a 12 hour day. So if play starts at 1PM, the normal finishing time is 1AM. That means (at least for those of us who can't go instantly from high adrenaline, high focus to sleeping) that the players are getting to sleep at the earliest at 2AM. It really is a grueling schedule for the pairs who play both the final segment and the first one the next day.

As a Vugraph operator, I "play" all 4 segments in a day. I don't have to think. I don't have any pressure about trying to win the match. And I'm usually exhausted at the end of the day. Is it any wonder that the players, even if they play "only" 1/2 or 3/4 of the boards are tired?

[FrancesHinden]

There's no need for a 48-board day to be a 12-hour day. The playing times in European (i.e. EBL) events seem much better to me than those in English ones. San Remo was typically 48-50 boards per day played as one session from about 10.30am to 2.30 pm and then a second from about 3.30pm to 7.30pm. Not a huge break for a late lunch, but enough time to have a rest. The early finish lets you have enough time to have a good dinner and wind down. English events typically finish at about midnight and it takes me a while after that to want to go to sleep.

But back to the original topic, you have to allow for the fact that a lot of bridge - even right at the top levels - is still played by amateurs. That means fitting in lots of boards per day because people don't have the time to miss too much work to play bridge. The hardest bridge I've ever played was the semi-finals and final of the Gold Cup, at 64 boards a day for two days (we played throughout). I was totally exhausted at the end of that and couldn't have done any more. The big English events are typically 60-odd boards a day (the Spring 4s, the Premier League, the Brighton teams) for between 2 and 4 days.

I've happily played 100 boards over a weekend in multiple teams or pairs events such as the Pachabo or the Corwen but any event with frequent moves and/or scoring breaks always seems less intense than e.g. the 20-board sets of the premier league.

[JanM]

FrancesHinden, on Mar 7 2010, 12:51 PM, said:

There's no need for a 48-board day to be a 12-hour day.

ACBL Knockouts are 64 boards, not 48. One of the advantages of playing only 48 boards a day, as the WBF does, is that the day doesn't have to be so long. But the disadvantage is that either the event isn't as formful because not so many boards are played, or there have to be a lot more days.

[xcurt]

mycroft, on Mar 5 2010, 05:05 PM, said:

I'm very glad the suggestion to restrict the number of boards played per match in the Vanderbilt/Spingold from 64 to 52 failed. It would have accomplished the same outcome (a session=4 hours), but with a *lot* higher increase in the luck factor than getting people to play to time.

Alert mathematically incorrect. p(better team is leading after N boards) is proportional to log(N).

[hrothgar]

xcurt, on Mar 8 2010, 06:18 AM, said:

mycroft, on Mar 5 2010, 05:05 PM, said:

I'm very glad the suggestion to restrict the number of boards played per match in the Vanderbilt/Spingold from 64 to 52 failed.  It would have accomplished the same outcome (a session=4 hours), but with a *lot* higher increase in the luck factor than getting people to play to time.

Alert mathematically incorrect. p(better team is leading after N boards) is proportional to log(N).

LOVE to see the derivation of this one...

[xcurt]

Sorry Richard I stated it too quickly.

Loosely...

Regardless of the distribution of IMP swings, when you add many of them you get a normal distribution. Hence the exact expression is p(Z<N^0.5), where Z is some normal incorporating the standard deviation in IMPs/bd and the difference in ability between the teams. Taking the log of both sides and putting all the constants on the left you get, for some constants k1 and k2, p(k1+k2x < log[N]) where x is taken from some uniform distribution.

The point I was trying to make is that the chances of an upset in a 52-board match just aren't that much greater than they are in a 64-board match.

[mike777]

xcurt, on Mar 7 2010, 10:41 PM, said:

Sorry Richard I stated it too quickly.

Loosely...

Regardless of the distribution of IMP swings, when you add many of them you get a normal distribution.  Hence the exact expression is p(Z<N^0.5), where Z is some normal incorporating the standard deviation in IMPs/bd and the difference in ability between the teams.  Taking the log of both sides and putting all the constants on the left you get, for some constants k1 and k2, p(k1+k2x < log[N]) where x is taken from some uniform distribution.

The point I was trying to make is that the chances of an upset in a 52-board match just aren't that much greater than they are in a 64-board match.

When you say not that much greater do you mean something close, very close to zero....

I understand zero.....not that much greater is a bit fuzzy to me.

[mycroft]

xcurt, on Mar 7 2010, 09:18 PM, said:

mycroft, on Mar 5 2010, 05:05 PM, said:

I'm very glad the suggestion to restrict the number of boards played per match in the Vanderbilt/Spingold from 64 to 52 failed.  It would have accomplished the same outcome (a session=4 hours), but with a *lot* higher increase in the luck factor than getting people to play to time.

Alert mathematically incorrect. p(better team is leading after N boards) is proportional to log(N).

Alert: absolute value of the difference is irrelevant to the mathematics.

While I will admit that I'm SWAGging completely, and have no idea where the comparison is, the mathematics in my statement is:

p(U(52)) - p(U(64)) > p(U(64 in 8h40) - p(U(64 in 11h) (+breaks and dinner in both cases)

Now this could both be small numbers, and I have no idea what they are, and I don't play at anywhere near that level. But my guess is that I'm right, even if the numbers are small. And it does have the advantage that the ACBL-centric teams get used to playing to time for the WBF events...

[xcurt]

mycroft, on Mar 8 2010, 12:08 PM, said:

...

If you're thinking that maybe the fatigue factor is less important than the reduction in luck from more boards...

I'm saying the reduction in luck isn't that great, independent of fatigue.

It sounds like Fred is saying the fatigue difference from 48->64 boards is significant enough that you get a big edge by only having to play 48 any given day.

[whereagles]

xcurt, on Mar 8 2010, 03:18 AM, said:

p(better team is leading after N boards) is proportional to log(N).

oh? I would think it's something like a binomial or so.

[Siegmund]

If we're assuming normally distributed imps on each board, then the z-score of the teams is proportional to sqrt(# boards).

Maybe xcurt is trying to say that if the z-score is proportional to that, then the probability of that specific outcome will be e^-(z^2/2) -> e^-(k*#boards). But I think it's more likely he was trying to say something else entirely and it got garbled.

[rogerclee]

xcurt, I don't understand your logic, but I understand mine.

Suppose team 1 is better than team 2 by a measure k (imps/bd), they play n boards, and the standard deviation of imps/bd for every board is s. Assume each board is an independent event.

Then the distribution of imps/bd for team 1 in a match of n boards should behave normally with mean k and standard deviation s/sqrt(n).

It is not hard to write the probability of team 1 beating team 2 from this, it is

P(n, k, s) = 1/2 + 1/2*erf(k * sqrt(n) / (s*sqrt(2)))

So if this function grew as log(n) then it should have derivative proportional to 1/n, but this function has derivative proportional to

exp(-nk^2/2s^2) * 1/sqrt(n)

Looking at a couple plots it seems like the model 1/2 + a * sqrt(n) does a much better job than the model 1/2 + b * log(n+1) at modeling the behavior of the actual probability function. 1/2 + b * log(n+1) tends to overestimate the probability of winning for small values and underestimates it for large values.

In addition I ran the numbers of a team with a 0.5 IMP/bd advantage (realistic for something like the early rounds of a Spingold/Vanderbilt match), assuming s= 5.6, over 64 and 52 boards. Team 1 has a 76.2% chance of winning over 64 boards but a 74.0% chance of winning over 52 boards.