4NCL North 2013-14

[David Robertson]

It might indeed be the case that Div 3N is (always has been?) a tougher environment than Div 3S. In the latter, the stronger teams get a free pass for the first two rounds, not to mention four match points, by playing some very weak opponents. Then the Swiss kicks in. But this year, in 3N, there won't be many push-overs, except perhaps in Rd 1. Won't stop the usual suspects from complaining of course.

[MartinCarpenter]

That probably depends on how strong all these new players coming in are. The teams haven't always been that strong in depth so with the board order rules as they are, you could see some of the second/third(!) teams ending up being relatively weak. Especially with them being quite new affairs.

Maybe some of them are going to run as split squads instead?

Certainly, with the main 'Bradford' team up in division 2 you can't see any of the Yorkshire teams being more than reasonable. Jorvik have too many WR (and Bradford) players to be otherwise.

[MartinCarpenter]

Well, looking at the team lists, anyone who qualifies for promotion - from either N or S - ahead of either Atticus or NE England will have earnt it. Depends quite who plays and how often of course but very strong squads.

Plenty of other decent teams in 4NCL N and maybe the odd softish team.

[David Robertson]

Plenty of other decent teams in 4NCL N and maybe the odd softish team.

I refer you to my post above

[Mick Norris]

Draw for first weekend now out too
http://www.4ncl.co.uk/1314_fixdiv3n.htm

Interesting to see the depth (or otherwise) of the squads too

[MartinCarpenter]

Is it just me who has that page showing blank?
(yes, just me needing to hard refresh it.).

They clearly haven't seeded the draw, which I suppose isn't insane with enough rounds for the swiss to sort things out reasonably. Cheddleton 2 seem to have done reasonably, Atticus' draw seems designed to make David's point for him

[David Robertson]

Frankly, I think it's a farce of a draw for all concerned, and arguably unfair when set against practice in 3S or any other Swiss event. One has to wonder sometimes what passes between the ears of the Arbiters involved.

And don't get me started on the allocation of teams to Pools in Div 2. The imbalance there seems to breach 4NCL protocols (as I have privately pointed out, but to no obvious effect - yet)

[IM Jack Rudd]

And don't get me started on the allocation of teams to Pools in Div 2. The imbalance there seems to breach 4NCL protocols (as I have privately pointed out, but to no obvious effect - yet)

I've just looked it up, and I agree with you: the standard allocation of teams would have one of KJCA and Poisoned Pawns being swapped with one of Bradford and Brown Jack. (Not that I'm entirely certain which pool is actually stronger as a result of this.)

[David Robertson]

And it really isn't that hard to get right, for heaven's sake

Relegated teams 5 & 8 to Pool X; teams 6 & 7 to Pool Y
Promoted teams 2 & 3 to Pool X; teams 1 & 4 to Pool Y
Surviving teams 1 & 4 to Pool X; teams 2 & 3 to Pool Y

Or very similar.

So what goes wrong when we don't get this simple arrangement?

[Roger de Coverly]

They clearly haven't seeded the draw

How can you tell? It looks plausible that there are (unpublished) seedings in use, but distorted by an apparent rule not to pair two teams from the same squad. Whilst seeding based on the top six nominated players is easy enough provided there aren't any dummy nominations, seeding the other team in the squad is more difficult. SofA look as if they have both teams in the top half. The round 2 draw appears to be paired as if all the round 1 matches are draws.

I'm not sure how the allocations to pools in divisions 1 and 2 are done. At the risk of introducing more distortions and the potential headache of withdrawing teams, would a mechanical rule based on the previous season's results work just as well? So for division 2a, you have teams 5 and 8 from the div 1 relegation pool, plus teams 5 and 8 from the div 2 promotion pool, then teams 2 and 3 from the div 2 relegation pool and teams 2 and 3 from division 3. Or something similar.

Are not the division 1 and 2 seedings currently done by arbiter's hunch?

Division 1 is always going to have the distortion of keeping the multi team squads apart. At least this year it's symmetric following the Grantham and Sharks merger.

[MartinCarpenter]

Well you've got to explain stuff like:
a) How Atticus (unarguably a top 2/3 seed on various measures) are playing the ~fourth strongest team of fourteen in the first round. Yes there are plenty of alternatives. Ferry 2 for starters!

b) How you've got Manchester Manticores 2 and Chedderton 3 (both definitely bottom seeds) playing in the second round

c) How MM1 can go from playing Ferry 2 in the first round (so surely top vs bottom) to playing Spirit in the second?

If this was seeded I'd love to know how, and would suggest it isn't the best way to do it! I guess the sort of saving grace is that a six(? or is it 8 before sharing?) round swiss with just fourteen teams might mostly balance this out.

[Mick Norris]

I assume they expect the following pairings to be of teams winning round 1:
Spirit of Atticus A v Manchester Manticores 1
Jorvik v North East England 1
Cheddleton 2 v Bradford DCA Knights B

and these being teams losing in round 1:
Spirit of Atticus B v Bradford DCA Knights C
Manchester Manticores 2 v Cheddleton 3
Ferry 'Cross the Maroczy 2 v North East England 2

leaving Holmes Chapel v Ferry 'Cross the Maroczy 1 as downfloat v upfloat?

Maybe the strength of Cheddleton 1 has confused the estimated strength of Cheddleton 3, but there are 2 further weekends to iron out any mistakes

[David Robertson]

Martin is right - some explanation is required. Mick tries to help by 'guessing'. But this is utter nonsense. Why should any team have to second-guess the Arbiter's mind? And why is 3N being treated differently from 3S? At the end of the first weekend, the stronger teams in 3S could have hoovered up four easy points. Meanwhile, in 3N, the stronger teams are required to knock lumps off each other. Then, when the divisions merge after six rounds, the 3N leaders have to face the 3S front-runners without the benefit of those giveaway early points. The impact on squad planning must also be considered. Stronger squads in 3S know, via seeding, that they will have an easy first weekend. But in 3N, none of this applies. Here the Arbiters make it up as they go along, mucking up fair competition along the way. #completely useless

[Alex Holowczak]

Martin is right - some explanation is required.

Wearing my hat as a 4NCL Team Representative, I asked what the pairing rules were for Division Three, back in April 2012. I was fobbed off.

Undeterred, I asked again in September 2012, and was given an answer of sorts: "The general principles are to pair according to match points, with game points then average rating as the way of ranking the teams. Initially, seeding is by last season's finishng position unless we accept other evidence."

The e-mail continued to explain that the person who does the pairings "does the pairings with considerable care and I have found that he produces fair solutions."

I think this situation is thrust upon the arbiters for a few reasons:
(1) Drawing two rounds at once makes things difficult. If teams were prepared to wait until Saturday night to see who they play in the Sunday round, then you could write a pairing algorithm. I think Mick's point about guessing results is probably not too far away from what actually happens when creating the Sunday pairings.
(2) Some teams might register 12 players for a 6-man team. How do you seed properly, particularly with multiple teams? In round 1, this can be very difficult.
(3) Given the growth of 3N this season, there are a lot of new teams to guess the strength of for the first rounds, so you are pairing blind unless you look at player registrations.

If you compare this with Junior 4NCL, we have 5 rounds played over a weekend. I don't have problem (1) to deal with, and so I can use a pairing algorithm - specifically I treat teams as individuals and pair based on matchpoints (ignoring gamepoints) according to the Dutch pairing rules. I still have problem (2). If a team registers 25 players for 3 teams of 4, I treat players 1-4 for the seeding of the A team, 5-8 for the B team, and 9-12 for the C team. This dramatically inflates the average rating of the B and C team relative to other teams in the squad in practice, but I don't really see what I can do about that. The problem of seeding a n-player team when > n players can play in it is a problem in all team events, including the European Team Championship, where our open team is seeded 3 by taking the average rating of the top 4 boards. The reserve is ignored for seeding, which is probably why they re-sort by gamepoints before seeding when doing the pairings.

If you solve problem (3) by pairing after the registrations come in, and do your best with problem (2) given it will always be there, captains need to want to proactively get rid of problem (1) by removing the need to pair 2 rounds simultaneously for the first four weekends. That way you can argue about why the pairing algorithm that has been implemented is unfair, rather than argue about the arbiter's judgement of a fair set of pairings.

As a Team Representative, if a captain wants me to get in touch with the man who does the pairings to request an explanation, I'm happy to do so.

[Roger de Coverly]

(1) Drawing two rounds at once makes things difficult. If teams were prepared to wait until Saturday night to see who they play in the Sunday round, then you could write a pairing algorithm. I think Mick's point about guessing results is probably not too far away from what actually happens when creating the Sunday pairings.

If you can establish a seeding order ( point 2 below), you can pair as if it is an individual event. For the second round, just pair as if all first round matches are drawn. I thought that has been a method used in the past.

(2) Some teams might register 12 players for a 6-man team. How do you seed properly, particularly with multiple teams? In round 1, this can be very difficult.

There's the obvious problem of players being registered who aren't expected to play very much. For division three, using a rating cap of 2200 or 2300 might help. With multiple teams in a squad, there's no immediately obvious solution, particularly with the 80 point flexibility for overlaps. Perhaps just establish a well defined but arbitrary rule and see how it works out in practice. Initial seedings shouldn't make too much difference over time. The third division tie break is on Sum of Opponents rather than game points, so a gifted 6-0 result isn't that decisive.

There's an ancient approach which used to be employed for the British Championships. What you do is establish half the players/teams as seeds. This is so that seeds don't meet in the first round. After that you just pair randomly between the seeded half and the unseeded half. There's a potential accusation of bias, but use of pseudo random numbers would make the results reproducible.