[cromar]

I am a member of two rural clubs, each of which meets once per week during the winter months.
In club A the championship is decided by taking the average of all percentage scores, provided a minimum number of attendances (usually 75% of total possible).

In club B, the champions are those with the highest average of the best x scores, typically best 10 or best 12.

As I see it, a drawback common to both methods is that no allowance is made for the strength of the opposition. For example, a pair beating 19 other pairs with a score of, say, 58% is rated inferior to a pair beating only 11 other pairs (on a bad weather night) with, say, 60%.

Furthermore, the rationale behind method B leaves me baffled. I know of no other sport/pastime where one is permitted to disregard a poor score.
This method can produce results like this:
Pair A: 57, 58, 56, 52, 58, 57, 56, 52, 59, 57, 53, 49, 50, 52, 58. Overall average: 54.9%, Best 10: 56.9%
Pair B: 60, 50, 59, 65, 41, 46, 56, 46, 59, 59, 43, 40, 60, 58, 57. Overall average: 53.3%, Best 10: 58.3%

Is Pair A really inferior to Pair B?

How is the championship decided in other clubs of this type? I would be grateful for members’ comments.

[Bbradley62]

cromar, on 2011-April-21, 10:47, said:

I know of no other sport/pastime where one is permitted to disregard a poor score.

Professional tennis rankings are based on this sort of method. Some specific tournaments are required to be included in the rankings of the top players (the four Grand Slam tournaments, etc.) but others are on a "best x" method, where playing more allows you more opportunities to have good results and disregard the bad results.

[Bbradley62]

cromar, on 2011-April-21, 10:47, said:

As I see it, a drawback common to both methods is that no allowance is made for the strength of the opposition. For example, a pair beating 19 other pairs with a score of, say, 58% is rated inferior to a pair beating only 11 other pairs (on a bad weather night) with, say, 60%.

Are you suggesting that a 20-pair field is better than a 12-pair field? Is it not possible that the weaker players stay home on bad weather nights and the field is actually stronger, so 60% on that night is more difficult than when everyone shows up?

[Bbradley62]

One advantage of the Club B system is that the leading pair is less likely to stay home in order to protect their lead. In the Club A system, showing up on the last night and having a bad game could cost the leading pair the championship.

[paulg]

I am also a member of a rural club that runs its competitions from September to May. Bad weather is also a factor during the winter.

The club pairs championship is run over ten evenings in the season, with the best six percentage scores counting. Reasons for this format include the problem with people missing evenings due to weather or ill-health, but I think the club committee also like to keep the competition competitive for as long as possible. It also means that players are not penalised for attending the club - certainly I've known pairs not play at other clubs because they could only worsen their score in the club competition, and this hits the club financially.

The club has looked at other scoring methods, such as masterpoint totals or placing points, to cope with the variable size of field. However it stuck with percentages for the club championship, I think to let pairs 'excel', but does use placing points for aggregate competitions.

To be honest most people do not care that much and it probably makes little difference which method is used. But any method that means it is better to stay at home than play at the club is really bad.

[NickRW]

cromar, on 2011-April-21, 10:47, said:

As I see it, a drawback common to both methods is that no allowance is made for the strength of the opposition.

That is true - however, to make allowance for strength of the field, you have to have a method of quantifying that. Such methods are possible, but involve work - and after you've done it you will probably have members baffled at where their (adjusted) score came from. Not worth the hassle in other words for club competition.

One thing where you might want to do that for is if you're, say, elected to pick a team for an inter-club team of 8 or similar and want to pick your 4 best pairs - but if you bother with that then I suggest you keep your number crunching in a private spreadsheet. Even if you do that you may identify a pairing of players X and Y for your, say, 3rd best pair - then player Y has to cry off due to family commitments, but X feels already invited and wants to play with someone you regard as unsuitable and even though your 5th best substitute pair doesn't include X. S**t happens to even the best plans.

Quote

Furthermore, the rationale behind method B leaves me baffled. I know of no other sport/pastime where one is permitted to disregard a poor score.

The rationale is, as other people have pointed out, that it encourages maximum attendance and your score can't get worse.

Nick

[mycroft]

Bbradley62, on 2011-April-21, 10:59, said:

Are you suggesting that a 20-pair field is better than a 12-pair field? Is it not possible that the weaker players stay home on bad weather nights and the field is actually stronger, so 60% on that night is more difficult than when everyone shows up?

No, but the law of large numbers applies. You want to win a SimPairs? Find the smallest legal weakest game you can find. If you're a 60% player in that field, it is *much* more likely that you're going to get a 75% in a 5-table game than in an 18 (of course, it's *much* more likely that you're going to get a sub-50% game, too; that's how variance works. But if you're in a "win or go home" frame of mind, 66% is the same as 40%).

So, no suggestion that a 20-pair field is better than a 12-pair field; just that a 65% game in a 20-pair field is harder to get than a 65% game in a 12-pair field (especially if top is 9 on the first and 5 on the second - "just" better than average is 52% for 5-/9, but 60% for 3/5).

As for "throwing out bad scores", that's fairly common in ranking schemes. In addition to the tennis ranking (which, I guess, is "good" to throw out the odd low result), the most obvious one is the (USGF) golf handicap, which is based on the best 10 of the last 20 rounds, but is used (in handicapped events) against all rounds (so it's usually "bad" for the player to throw out the result; except that a low handicap as a status symbol/ego boost is probably better than a "fairer" handicap to most golfers).

[MickyB]

Playing in a small sim pairs field might increase the chance of you scoring 75% locally, but it will have very little effect on the variance of your score nationally.

[Free]

At our club it's method B. It's not so strange as you think. The main reason for this is that many people don't play all events, while the club wants to promote people to come and play as often as possible. Once you've played 10 events, you can only improve your current score. If you count all results (method A), then you can ruin your score if you have a bad day. So it might pay off not to come at all.

This doesn't mean that the pair winning in method B is actually the best pair, but it fits in the philosophy of club championships.

Imo to avoid the problems with your example on method B, one should also count the number of events that were played. Obviously if 2 pairs have played the same amount of deals, their average over all events should count. But you also need a way to compare 2 pairs that have played a different amount of deals.

Perhaps you should use the average number of matchpoints won, instead of percentages. This would also include the size of the field, because you can gain more matchpoints in bigger fields. Still, I don't know what the best method is to compare 2 pairs with a different amount of events played, without punishing someone for playing too much.

[mycroft]

MickyB, on 2011-April-21, 13:50, said:

Playing in a small sim pairs field might increase the chance of you scoring 75% locally, but it will have very little effect on the variance of your score nationally.

My apologies, Micky, you are correct. I of course, meant things like the ACBL-wide pairs events, where it actually does make a difference - because it's only the local score that matters, it is not scored across 10 000 top.

Having said that, my point still stands as to the difference small fields can do to variance and big games.

Having said *that*, I really don't think it matters in the end, unless the reward for "championship pair" is such that it's worth trying to game the system - in which case there are much easier ways to game the system.

[Quantumcat]

In Australia we use the CARDS rating system. Everybody who is a member of a club supporting this system has a rating usually between 40 and 60. In every session, the field's strength is calculated and each pair has an expected finishing position and score. If they do better than this, they get a plus to their rating (usually between +0.01 and +0.5), and a negative to their rating if they do worse. The stronger member of the pair gets a smaller plus than the weaker member, or a bigger minus if they do poorly. People only get a rating if they have played over 13 times in 12 months, so it is reasonably accurate even when you only first get your rating.
This system adjusts for your partner and the field strength. For instance, if you play with a world-class expert and finish first with 60%, you will probably get a minus. If however you play with the club simpleton and manage to wrest the decisions off him and finish 3rd with a score of 55%, you will probably get a plus. If the room is full of old ladies who can't count to 13, you will probably need 70% just to avoid a minus, while if it is a national pairs event with world class players, you could get a hefty plus just for finishing in the top half of the field.
In any case, while you are debating the merits of overall average or best ten scores, the above knocks both approaches out of the field :-) Not every club is signed up to the system, and many of our top players haven't played enough games that use the rating system to be placed where they should be. It may take a few more years before the top ten rated players really are the top ten Australians.
The official website is here: http://www.bridgecentral.com/

[gnasher]

Is there an explanation anywhere of how the CARDS system actually works?

[paulg]

CARDS looks like a handicap system, certainly a lot more sophisticated than the systems that are currently used at my club.

However, although there are many events where handicaps are used, the club did not want such a system for the championship pairs.

[gwnn]

gnasher, on 2011-April-26, 04:24, said:

Is there an explanation anywhere of how the CARDS system actually works?

yes it's a little annoying how they write all sorts of stuff except the formulas.

[gnasher]

This is how it's done in the online Barbu world:

http://www.barbu.co....explanation.htm

The method described there has a number of disdvantages:
- Only a small number of people understand it.
- Even those who do understand it are unsure how fair it is.
- Some of the people who don't understand it are nevertheless certain that it's unfair.
- The same people believe that it does various things that it doesn't do.
- The same people think that it's possible to make it both fairer and simpler, but are unable to explain how.

[FrancesHinden]

Back to the original question.

To decide what the best method is, you need to decide on your objective.

If the objective is to identify the best pair in the club, you should play some qualifying sessions, then a semi-final, then an all-play-all final over a number of weeks (ideally with some seeding/swiss elements to try and keep the in-contention-pairs playing each other through out).

But the objective is almost certainly not to identify the best pair in the club. Rather, it is to run a competition over a number of weeks that encourages people to to attend the club and also means that one of the better pairs will win. Hence as others have said, any method of scoring that gives people an incentive not to play is bad.

My local club runs various 'ladder' events (where your score is the average of your best n scores out of m evenings). Their pairs championship is a straight one evening qualifier and two evening final.