Division 3 2012-13

[Christopher Kreuzer]

The final cross-table for Division 3 is here:

http://www.4ncl.livechess.co.uk/div3_20 ... s-rd11.htm

I think the top four are promoted, which would be Hackney, Cambridge University 2, Bradford DCA Knights A, and Brown Jack. Three of those teams (the ones that came 2nd to 4th) were relegated from division 2 last year, so are going straight back up (are Hackney a new team?). The other team that was relegated from division 2 last year was FCA Solutions 1, which finished 12th this year.

Can anyone explain how SOS (sum of opponent's scores) is calculated for teams that withdrew before completing all 11 rounds, and for the teams that played those teams before they withdrew?

[MikeTasker]

Hackney have been playing for several years, but managed to keep a more stable side this year.
The table does not seem to be ordered by game points,according to it Guildford 3 have more GP
than Brown Jack and should come 4th and be promoted.

[Christopher Kreuzer]

The 2012-13 rules are here:

http://www.4ncl.co.uk/1213_rules.htm

13.3 says "End of season team placings in the Division 3 combined division will be decided in the first instance on match point totals. If at the end of the season teams in the Division 3 combined division are tied on match points, the tie shall be broken in favour of the team having “(a) the highest sum of opponents’ match points (b) the greatest number of game points (c) the highest sum of opponents’ game points (d) the toss of a coin. "

Confusingly, the cross-table doesn't seem to show tie-break (a), though it does show tie-breaks (b) and (c). I still don't know how you calculate those tie-breaks for teams that have withdrawn, or where triangular matches took place.

[MikeTasker]

OK, I assumed it was game points. Yes it is unclear what happens if a team(s) used to make the
sum of opponents match points has withdrawn, maybe the average of the opponents match points with the withdrawn teams removed, very confusing!

[Sean Hewitt]

Sum of Opponent's [Match Point] Scores is the first tie break. Simply take the match points of each of the 11 opponents and add them up. This ranks highest the team that played the 'tougher' field. If a team does not play in a round, this simply means that they gain no match points that weekend. The tie break therefore pays no attention to whether a team plays (or not), or whether they participate in a triangular match.

[Joey Stewart]

I still maintain that instead of using sum of opponents scores as a tiebreak as opposed to game points /match points, you might very well just skip straight to the coin flip which is equally as random and probably would be seen as fairer.

[Roger de Coverly]

I still maintain that instead of using sum of opponents scores as a tiebreak as opposed to game points /match points

Match points comes first, so the team winning or drawing the most matches will win.

The point of using SOS as the next tie break is that it favours teams that have been amongst the leaders all season. So a team that has a series of 3-3 draws against strong opposition will be favoured against one that loses a couple and then hits weaker teams with 6-0 or 5-1.

In an all play all, using game points as a tie break is rational, in a Swiss it doesn't work nearly as well as witnessed by its downgrading or abandonment in major team events such as the Olympiads.

[Sean Hewitt]

I still maintain that instead of using sum of opponents scores as a tiebreak as opposed to game points /match points, you might very well just skip straight to the coin flip which is equally as random and probably would be seen as fairer.

Take the following example. Team A starts well and is constantly at the top of the division. Week after week they get good results, winning 3.5-2.5 against other top sides, conceding the odd draw and losing a match at the end. They finish on 16 points.

Team B loses its first two games. Paired with much weaker opponents, it regularly wins 5-1 whilst avoiding all the top teams. In round 8 it loses again, getting easier slightly easier pairings in the last weekend where they manage to win all 3 matches (including, of course, a top match in round 11) to also finish on 16 points.

Which teams performance most warrants promotion? Which team would have the better SOS? Which team would have the better Game Points score?

[Joey Stewart]

A good consistent team should be getting through on match points anyway, so the inconsistent occasional high scorers will mostly not get through but if they do manage to come joint first then they deserve it.

What I mean is that both sides have EARNED every game and match point and their fate is in their own hands - when you start introducing the lottery of opponents scores, then it becomes totally reliant on the luck of the pairings. Its a good system to sort out tiebreaks after that but I do think that promotion should be primarily decided on achievement.

It is a fairer system then the splitting of the pools in the top divisions, but that is another story....

[Sean Hewitt]

when you start introducing the lottery of opponents scores, then it becomes totally reliant on the luck of the pairings.

If the pairings were random then I would agree with you, but they are not. Pairings are seeded, based on the number of match points each team has. So the team that has the higher SOS has played, overall, a tougher bunch of opponents. To my mind, that is more meritorious than thrashing a number of weaker teams. Others may think differently.

[Greg Breed]

when you start introducing the lottery of opponents scores, then it becomes totally reliant on the luck of the pairings.

If the pairings were random then I would agree with you, but they are not. Pairings are seeded, based on the number of match points each team has. So the team that has the higher SOS has played, overall, a tougher bunch of opponents. To my mind, that is more meritorious than thrashing a number of weaker teams. Others may think differently.

I agree with Sean, SOS is much fairer than game points alone.

[Andrew Bak]

If the pairings were random then I would agree with you, but they are not. Pairings are seeded, based on the number of match points each team has. So the team that has the higher SOS has played, overall, a tougher bunch of opponents. To my mind, that is more meritorious than thrashing a number of weaker teams.

I agree with this entirely. Speaking as captain of a Div 3 North team, we were fully aware that we had played fixtures against Aigburth and Manchester who were towards the bottom of the table which meant that if we were in a tie on match points for the last promotion place, we would likely lose out due to SOS.

Thankfully we ended up playing a lot of the top teams, losing against the two that finished above us and beating all the others, knocking them out of the promotion spots so tiebreaks were only an issue for the final promotion spot.

[Alex Holowczak]

I have a good example of the perils of gamepoints, even in an APA.

In the Wolverhampton League this season, St. George's and Bushbury tied on 15 matchpoints. Bushbury had 42 gamepoints, including 3 6-0 defaults, and St. George's had 35 gamepoints, and lost one match 6-0 by default (against Bushbury!).

A few years ago, a club won the League on gamepoints after a series of defaults, so the rule was changed to make it fairer. It meant that a team could not include wins by default in its gamepoints total. Tied teams couldn't include results against those teams either in calculating their gamepoints total.

This meant that Bushbury were reduced to 24 gamepoints, and St. George's ended up with 25.5 gamepoints, thus winning the Wolverhampton League on tie-break.

What makes this a nonsense is that, if the fixtures were a bit different, St. George's might have played Bushbury last. They would have worked out going into the match that they'd be ahead on tie-break. Therefore, to win the League, they could just deliberately default the match to Bushbury, and the gamepoints wouldn't count. So St. George's would win the League.

It's not often that I stick up for the rules of the Birmingham League, but their rule is that tied teams (for a position that affects promotion/relegation) have a playoff; either a match or jamboree. That's probably the fairest way of settling things.

I guess my point here is that tie-breaks aren't easy things to create!

[MartinCarpenter]

Surely not a play off for teams who have played each other during the season? That seems like a recipe for organisational trouble. Organising teams post the end of the season isn't easy. Head to head result and then game points if tied seems fine then.

[Alex Holowczak]

Surely not a play off for teams who have played each other during the season? That seems like a recipe for organisational trouble. Organising teams post the end of the season isn't easy. Head to head result and then game points if tied seems fine then.

The Birmingham League has successfully organised end of season playoffs for 116 years. According to someone who has done the numbers, there's no higher a default rate in those than there is in regular season league matches. This is, of course, impractical in the 4NCL - but I was talking about local league tie-breaks really.

However, I would be happy with a tie-break of the result of the head-to-head game(s) in the regular season.