[gnasher]

mchristie, on 2012-April-12, 04:27, said:

Actually, currently newcomers enter on a grade of 46.00. We monitor the average performance of the approximately 100 new players joining the system per week to see what their average strangth is. Over the last few months this has been around 45-46%. If it drops any lower, we'll reduce the newcomers' ranking to match. Zia has not yet played enough rated games to get a published grade, and you may feel it was lucky for his partners to be playing with such a skillful newcomer with an initial rating of 46, but it didn't seem right to make any exceptions. Should we?

Once you have integrated teams events into the calculation, that specific problem will probably go away. Zia must have played enough Gold Cup matches to make him into a proper citizen.

For the "strong newcomer" problem in general, I think it would be OK to make a subjective decision in a case where a new member had previously played in another country, and perhaps also where a player has reappeared after an absence of several years.

Alternatively, what about waiting until they'd played enough times to produce a reliable grade, then retrospectively applying that to the games they played earlier? Or are there technical obstacles to that?

[WellSpyder]

mchristie, on 2012-April-12, 04:27, said:

Actually, currently newcomers enter on a grade of 46.00. We monitor the average performance of the approximately 100 new players joining the system per week to see what their average strangth is. Over the last few months this has been around 45-46%. If it drops any lower, we'll reduce the newcomers' ranking to match.

The more I see of the scheme, the more impressed I am at how the issues involved have been tackled. Of course there are limitations, and it is possible a few people will focus rather obsessively on how to maximise their grade given how it works, but for most people I hope it is an interesting way of looking at how they are performing without being the end of the world if there appear to be anomolies in how some people are ranked relative to each other.

I do hope that eventually the decision to exclude teams of 4 bridge will be reconsidered, though, since a) for some players this is probably the majority of the bridge they play and b) I don't think conceptually there are major differences between how this could be done and how individual grading is done now in a pairs context.

[Jeremy69A]

Quote

I do hope that eventually the decision to exclude teams of 4 bridge will be reconsidered,

I don't think it is a decision to exclude but two problems. The first is the "one at a time" problem i.e. get it going first. I believe it is the intention to include teams. If and when this is done then in events where it is known who is playing who and there is a record e.g. via a bridgemate then all should be ok but for matches played at home in leagues or knock outs the system would not easily know who played whom.

[WellSpyder]

Jeremy69A, on 2012-April-12, 07:06, said:

I don't think it is a decision to exclude but two problems. The first is the "one at a time" problem i.e. get it going first.

I fully understand and accept this - hence "eventually" in my previous post.

Quote

I believe it is the intention to include teams. If and when this is done then in events where it is known who is playing who and there is a record e.g. via a bridgemate then all should be ok but for matches played at home in leagues or knock outs the system would not easily know who played whom.

I think there is a distinction between multiple teams events and straightforward teams of 4 matches. I believe there is an intention to include the former where it is known who plays who, using Butler or cross-imp scores. But my understanding is that there is currently no intention to include the latter, and that is what I intended to refer to. For these matches there is absolutely no need to know who plays who in terms of which pair plays which pair. All that matters is that such and such a team of 4 played such and such other team of 4 (maybe teams of 5 or 6 would need further consideration?), since obviously you cannot allocate a result to one table or the other.

[Jeremy69A]

Quote

I think there is a distinction between multiple teams events and straightforward teams of 4 matches. I believe there is an intention to include the former where it is known who plays who, using Butler or cross-imp scores. But my understanding is that there is currently no intention to include the latter, and that is what I intended to refer to. For these matches there is absolutely no need to know who plays who in terms of which pair plays which pair. All that matters is that such and such a team of 4 played such and such other team of 4 (maybe teams of 5 or 6 would need further consideration?), since obviously you cannot allocate a result to one table or the other.

You are right about multiple teams and Swiss events. I think it is desirable to include the head to head matches but there has to be some thought about how it gets recorded. If you organise a team competition and you want to get something as basic as who played and how many boards it is remarkably difficult! If something is to happen it can't rely on a load of manual data and amending data so this will need some thought first.

[gnasher]

Jeremy69A, on 2012-April-12, 07:20, said:

You are right about multiple teams and Swiss events. I think it is desirable to include the head to head matches but there has to be some thought about how it gets recorded. If you organise a team competition and you want to get something as basic as who played and how many boards it is remarkably difficult! If something is to happen it can't rely on a load of manual data and amending data so this will need some thought first.

I don't think you need any more information than for a pairs event.

When you enter a pairs event, you and your partner are treated as a single contestant, and each of you gets equal credit or discredit for the result. Therefore all you need to know is (a) who the players were in each partnership and (b) what the overall score was.

When you enter a teams event, you and your teammates should also be treated as a single contestant. Therefore all you need to know is (a) who the players were in each team and (b) what the overall score was.

In a cross-IMPed pairs, each player's rank for the session is:
SGa = Ga + (6.48i/x + SOpp – Ga/2 – Gb/2)

In a head-to-head teams match, each player's rank should be:
SGa = Ga + (6.48i/x + SOpp – Ga/4 – Gb/4 – Gc/4 – Gd/4)

where
Ga = your rank
G[b-d] = partner's/teammates' rank
SOpp = strength of opposition
i = net IMPs
x = no of boards played.

[AlexJonson]

I guess the issue is 'head to head' teams of four where some teams have six players (or n), so you have to think about the consequences of that.

When I was young, I would have said teams of 4 should = 4 players, but that is probably impractical.

I doubt that it's acceptable with more than 4 without knowing who played and when, to just equalise all the results.

[Cthulhu D]

gnasher, on 2012-April-12, 04:45, said:

For the "strong newcomer" problem in general, I think it would be OK to make a subjective decision in a case where a new member had previously played in another country, and perhaps also where a player has reappeared after an absence of several years.

Alternatively, what about waiting until they'd played enough times to produce a reliable grade, then retrospectively applying that to the games they played earlier? Or are there technical obstacles to that?

Yeah if a superstar turns up you can probably do it by hand, the alternative is the chess solution. You make established players grades more resistant to change and new players very flexible on the basis that you have confidence that your established ratings are right. This also protects established players from damage because the adjustment from losing to someone is pretty low. This also protects established players who turn up and have a disaster session for external reasons.

[Jeremy69A]

You may be able to find a way to assess a grade from a locally played teams match but you also have to deal with the input of data into the system. For all pairs and events it is done automatically. Bridgemates in every home?

[gnasher]

Jeremy69A, on 2012-April-13, 02:25, said:

You may be able to find a way to assess a grade from a locally played teams match but you also have to deal with the input of data into the system. For all pairs and events it is done automatically. Bridgemates in every home?

When I report a the result of a Crockfords match, I send an email to the EBU telling them:
- Who won.
- Who played on the winning side (assuming that it's a team of more than four).
Somehow this information makes its way into the Masterpoint system, so somebody at the EBU must type it onto a computer.

If they were using this information for grading, they would need to know, in addition:
- The winning score
- Who played on the losing side

So yes, it would create a bit more work for the EBU staff, but it's not an order-of-magnitude difference from what is currently done. Alternatively, you could provide an online form for reporting match results.

[gnasher]

Actually, thinking about this a bit more, with a head-to-head knockout match I don't see any need to know the IMP difference.

The objective is to win, so the players should be rated on whether they achieved that objective. Once you start rewarding them for winning heavily or penalising them for losing heavily, you create conflicting objectives.

When a team is behind, we want them to try to turn the match around. We really don't want to see them playing conservatively so as to limit the damage to their grade, like some mediocre cricketer batting for his average.

[Jeremy69A]

Quote

When a team is behind, we want them to try to turn the match around. We really don't want to see them playing conservatively so as to limit the damage to their grade, like some mediocre cricketer batting for his average.

I don't think it is the role of the grading system or its inventors to have a view about how contestants play. It will be for the contestants to decided what they think is best. If you do decide on imps etc rather than straight result you also have to think about the effect of concessions.
If you play at the club on a normal pairs night and are doing mediocrely with a few rounds to go will you try to win by edging a few over the slips or will you play for your NGS rating? I guess in those clubs with ladders this problem has had to be resolved by contestants before.

Quote

Actually, thinking about this a bit more, with a head-to-head knockout match I don't see any need to know the IMP difference.

That would simplify it.

I'm sure it will be possible to sort something out but the devil will be in the detail to make it as meaningful as possible without creating a significant extra workload and therefore cost

[mchristie]

Yes, we are slowly working towards using teams events, for events such as SwissTeams where the boards are played multiple times. They will be graded by looking at the performance of individual pairs.
For individual team of four matches, yes there are problems about data collection which could probably be overcome with a lot of effort, but the biggest issue is that there is only one result and it affects equally all eight players. That makes each board statistically worth only 1/4 of a dupliacte pairs board, so a 48 board match would be given as much weight as a 12 board pairs session. But..., the biggest problem is "systematic errors". The NGS suffers a bit from this, if a player is over or undergraded, this affects their partner's grade, and leads to people complaining about the system. For head to head team matches, you grade is susceptible to errors in any of the other seven players. I suspect that the inclusion of such events would simply make individual grades less reliable.

[gnasher]

mchristie, on 2012-April-13, 03:10, said:

the biggest issue is that there is only one result and it affects equally all eight players. That makes each board statistically worth only 1/4 of a dupliacte pairs board, so a 48 board match would be given as much weight as a 12 board pairs session.

Surely the ratio is 2:1, not 4:1? A pairs result affects two players, and a teams result affects four players, so a 48-board match is worth a 24-board pairs session.

[mchristie]

gnasher, on 2012-April-13, 03:24, said:

Surely the ratio is 2:1, not 4:1? A pairs result affects two players, and a teams result affects four players, so a 48-board match is worth a 24-board pairs session.

That's what I thought originally, but in a head-to-head match the four players at THE ONLY OTHER table have an equal impact on the result, so such a result affects eight players, thereby contributing a second 2:1 factor. Once a board is played at several other tables the effect of the other tables' players rapidly diminishes. I think it's (n/n-1) if the board is played n times.

[gnasher]

Jeremy69A, on 2012-April-13, 03:10, said:

I don't think it is the role of the grading system or its inventors to have a view about how contestants play. It will be for the contestants to decided what they think is best. If you do decide on imps etc rather than straight result you also have to think about the effect of concessions.

Yes, concessions are another reason to grade people (in knockout matches) solely on the basis of whether they won or lost. Maybe it's worthwhile to compare it with chess: in chess they don't grade players on the basis of how many moves they took to win, or how many pieces up the winner was at the end - all that matters is the final result.

For similar reasons, I would grade Swiss matches according to the number of VPs won or lost and the strength of the opponents, ignoring the results on the individual boards.

It sounds as though I've entered this discussion rather too late, though. And of course Mike and others have put much more thought into this than I have.

This post has been edited by gnasher: 2012-April-13, 10:04

[Vampyr]

A quick question: In Swiss teams, Bridgemate II's give the names that they expect to appear at the table. Can they be programmed to switch the names to the ones that actually appeared at the table? What if there are more than 4 players on a team or the partnerships have swapped?

And if all this is possible, can the players be relied upon to make the changes (and make them accurately)?

[Oof Arted]

Vampyr, on 2012-April-13, 09:05, said:

A quick question: In Swiss teams, Bridgemate II's give the names that they expect to appear at the table. Can they be programmed to switch the names to the ones that actually appeared at the table? What if there are more than 4 players on a team or the partnerships have swapped?

And if all this is possible, can the players be relied upon to make the changes (and make them accurately)?

If they know who they are possibly , but dont hold your breath

[weejonnie]

mchristie, on 2012-April-12, 04:27, said:

Actually, currently newcomers enter on a grade of 46.00. We monitor the average performance of the approximately 100 new players joining the system per week to see what their average strangth is. Over the last few months this has been around 45-46%. If it drops any lower, we'll reduce the newcomers' ranking to match. Zia has not yet played enough rated games to get a published grade, and you may feel it was lucky for his partners to be playing with such a skillful newcomer with an initial rating of 46, but it didn't seem right to make any exceptions. Should we?

So this explains why the average grade for the NGS is below 49.

(One point - if we play in the same direction as a pair - is the SOpp the strength of the pair or the average strength of the two players?

I have a grade of 62.12 and a partnership grade of 59.02 with my usual partner.

Yet his grading is 54.87 making an average grade 58.495)

[Vampyr]

Oof Arted, on 2012-April-13, 09:14, said:

If they know who they are possibly , but dont hold your breath

I think this is about right. So I really don't know how teams events can be included, unless a separate rating is created for teams. It does not seem like it will be very accurate. But maybe the brains behind the scheme will come up with something.