How should the PAC 10 Divisions be drawn?
Rumors, originating from Colorado, stated that the PAC 10 was going to split into North-South divisions. The north would have the California Golden Bears, Stanford Cardinal, Oregon Ducks, Oregon St. Beavers, Washington Huskies, and Washington St. Cougars. The south would feature the Arizona Wildcats, Arizona St. Sun Devils, UCLA Bruins, USC Trojans, Colorado Buffaloes and Utah Utes.
My first impression of this news was that it was mighty generous to the two new teams in the league. USC and UCLA are considered prime scheduling due to the LA market and the recruiting wealth of southern California that all western schools want as big a piece of as they can carve.
Washington, Oregon, Washington State and Oregon State have been in the conference since its beginning. If they had approved such a plan I would have been amazed. I had sketched of an idea I was thinking of to promote a PAC 10 expansion (including Utah). Well, that is now history but I feel my sketch is now needed.
Travel costs are cut by grouping teams into regional groups. The four northern schools, four California schools and the two Arizona schools and two new schools are logical divisions I will call the North, California and South. The North and South have limited interest in playing each other, but a high interest in playing the California schools. The California schools also want to play each other every year.
NCAA rules require a conference to be split into two divisions to host a championship game. How do you take that mess and make division structure that works?
One big variable that is unknown is whether the design will have eight or nine games per year. Eight games allows home field advantage to be balanced and reduces the risk of teams eliminating fellow conference members from the top 25 rankings. Nine games allows the teams not locked together to play more frequently.
Currently the PAC 10 uses a nine game round robin schedule. Every current 12 team league uses an eight game schedule. The PAC 10 coaches recently discussed eliminating a round robin for the very reason listed above, but decided the value of a full round robin outweighed the cost of additional conference losses.
Plan A: Fixed Divisions
This plan acts like a control for this discussion. The PAC 10 has expanded to 12 and must make a schedule somehow. This is the most simplistic design available to build from. The best way to compare designs is by the fraction of years teams from each rivalry pair play each other.
Rivalry pairs are listed by the postal code for their state, LA for Los Angeles, NC for Northern California and NEW for Utah and Colorado. The numbers are the fraction of years a team from one rivalry pair would play a team from another rivalry pair. The top numbers are for eight game seasons and the bottom for a nine game season.
| LA | NC | WA | OR | AZ | New | |
| LA | 1 1 |
1/2 2/3 |
1/2 2/3 |
1/2 2/3 |
1 1 |
1 1 |
| NC | 1/2 2/3 |
1 1 |
1 1 |
1 1 |
1/2 2/3 |
1/2 2/3 |
| WA | 1/2 2/3 |
1 1 |
1 1 |
1 1 |
1/2 2/3 |
1/2 2/3 |
| OR | 1/2 2/3 |
1 1 |
1 1 |
1 1 |
1/2 2/3 |
1/2 2/3 |
| AZ | 1 1 |
1/2 2/3 |
1/2 2/3 |
1/2 2/3 |
1 1 |
1 1 |
| NEW | 1 1 |
1/2 2/3 |
1/2 2/3 |
1/2 2/3 |
1 1 |
1 1 |
Plan B: California Rotation
The key to this plan is that the NCAA rules do not state that the divisions have to be the same each year. The idea is that the California schools don't have to be divided once and for all between the two divisions. They can rotate between them on a four year cycle, with a home and away guaranteed with each team in each division every four years.
The California schools would also play each other permanently as crossover games, forming three groups I call California, North and South. Each school would play each team in their group every year.
With an eight game schedule each team has a remaining game that cycles among the crossover games. Since the crossovers are varying this becomes complicated. To make a home and away for this remaining game work one can set the lower bound of a cycle to 16.
With a nine game schedule this can be accomplished far more succinctly in an eight year cycle.
[Math edit: For the nine game schedule it is impossible to schedule 2/3 over eight years. This design cycles over 24 years. An eight game design with these proportions has been found with a 48 year cycle.]
| LA | NC | WA | OR | AZ | New | |
| LA | 1 1 |
1 1 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
| NC | 1 1 |
1 1 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
| WA | 5/8 2/3 |
5/8 2/3 |
1 1 |
1 1 |
5/8 2/3 |
5/8 2/3 |
| OR | 5/8 2/3 |
5/8 2/3 |
1 1 |
1 1 |
5/8 2/3 |
5/8 2/3 |
| AZ | 5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
1 1 |
1 1 |
| NEW | 5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
5/8 2/3 |
1 1 |
1 1 |
Significant gains are made relative to the non-division games of the fixed design. Some cost is absorbed in the total number of games with California teams that is needed to produce a full round robin for the California teams.
This plan is the most fair, but an eight team design is a scheduling nightmare.
Plan C: California Catering
Since the California schools are the heart of the conference, make them the heart of a division. Instead of rotating the California schools between divisions you would rotate other rivalry pairs into the California division. If the rotations are even this would have the same odds for the non California schools as the above designs.
But this plan does not have to be even. The new schools don't have to be rotated in. They would have a home and away with the California schools on a four year cycle.
The three state pairs would rotate, they would not play for two years, they would play but be in different divisions for two years and then they would play in the non-California division for two years.
Either way requires a cycle of twelve years.
| LA | NC | WA | OR | AZ | New | |
| LA | 1 1 |
1 1 |
2/3 5/6 |
2/3 5/6 |
2/3 5/6 |
1/2 1/2 |
| NC | 1 1 |
1 1 |
2/3 5/6 |
2/3 5/6 |
2/3 5/6 |
1/2 1/2 |
| WA | 2/3 5/6 |
2/3 5/6 |
1 1 |
2/3 2/3 |
2/3 2/3 |
5/6 1 |
| OR | 2/3 5/6 |
2/3 5/6 |
2/3 2/3 |
1 1 |
2/3 2/3 |
5/6 1 |
| AZ | 2/3 5/6 |
2/3 5/6 |
2/3 2/3 |
2/3 2/3 |
1 1 |
5/6 1 |
| NEW | 1/2 1/2 |
1/2 1/2 |
5/6 1 |
5/6 1 |
5/6 1 |
1 1 |
The state rivalry pairs must play the new teams nearly every year (every year in a nine team design) for this to work. It is interesting that the number of games between state rivalry pairs does not change for the eight and nine team designs.
The hardest part of this design might be to convince liberal leaning university presidents to accept that not all teams are equal.
This design also expands nicely to 14 teams treated like the two new teams. This brings the percentages of games with the California teams to those of the fixed design that opened this thread.
0 recs |
12 comments
|
Comments
Wow…you just blew my mind.
I was thinking of something more simple. Keep the N/S divisions as projected, and keep the 8 conference games (to reduce the guaranteed losses in the conference to maximize bowl eligibility). You play your 5 in-division games (with your designated natural rival) and 3 out of division games, 1 of which would be a fixed, none rotating rivalry game. This is to placate the Cali schools who have a long history of playing each other. They could even swap after two years if they are concerned about UCLA not playing Furd, or USC not playing Cal, etc.
This way, the NW schools get a guaranteed game in CA every year, and 1 in SoCal every other year. With 12 teams, its really the best that can be expected, and overall the best for the conference.
by Hockey Beard in SLC on Jun 18, 2010 3:20 AM EDT reply actions
Both the alternates presented
should allow a home and away game with a team in California every year, and a game in S. Cal on average five times every four years.
And they both preserve the round robin of the four California Schools
BCS Evolution -- Punctuating the Equilibrium - twitter
all i know is i want as a utah fan both usc and colorado in our division.
I'm all about covering the spread and moneylines. Craig Thompson just might end up being a savvy guy. Hell is just a word, the reality is much much worse." Event Horizon". Now remember, when things look bad and it looks like you're not gonna make it, then you gotta get mean, I mean plumb, mad dog mean. cause if you lose your head and you give up then you neither live or win. That's just the way it is. "The Outlaw Josey Wales". To protect the sheep you gotta catch the wolf, and it takes a wolf to catch a wolf. "Training Day". Rashad will beat Rua b/c it's a bad matchup for him. 2/1 odds. Josh Koscheck will dethrone Gsp and win the welterweight title. 2.9/1 odds on that too. Peace.
by wolfmanshowlforever on Jun 18, 2010 3:45 PM EDT up reply actions
Everyone wants USC in thier division
Them and UCLA. As a Utah fan I would rather have UCLA than USC. That is the rub.
BCS Evolution -- Punctuating the Equilibrium - twitter
I'm not worried
As a Utah fan I know I want to have Colorado in our division. We are the natural rivals; however, we haven’t played in a long time. We need to play every year and have it count towards winning the division.
After that I am torn about divisions. I have so many ties to the west coast I am just glad to be joining the PAC 10. I will be happy with any divisional split.
Here's an idea
And I admit it hasn’t been thought through all the way, so I’m sure somebody will spot the flaws that I haven’t. The point in having a Conference championship game is to (hopefully) put the two top teams against each other right? What if we had a seeding system that changed the division every two (or other #) years, based on the average record of individual schools (or rivalry pairs collectively?) over a predetermined set of time (five years maybe?). It seems this geography issue is getting highly contentious so let’s just do away with it entirely.
So based on a seeding system you would have:
Division A:
- team
- team
- team
- team
- team
- team
Division B:
- team
- team
- team
- team
- team
- team
Say what you mean, and say it mean. - Clint Ruin
Not sure what just happened
But for some reason that whole thing just got reformatted so it makes no sense.
Division A would have #‘s: 1,4,6,8,10,12
Division B would have #’s: 2,3,5,7,9,11
Say what you mean, and say it mean. - Clint Ruin
Seeding System
I like the idea of a seeding system. It eliminates the Big 12 problem of having a cupcake division and a Texas division. It gives the conference a true champion with a strong schedule that should perform well at Bowl games and outside the conference. It mixes up the geographies for fan and team travel. This requires more consideration. Does anyone know if any other conference, either currently or in the past, uses the seeding system?
Every conference has used fixed divisions
except for the WAC from 1996-1999, prior to the formation of the MWC. They used a design where every two years the eight middle teams would swap from the Pacific to the Mountain division.
BCS Evolution -- Punctuating the Equilibrium - twitter
I entertained putting rivalry winners in one division
this would reduce the value of the championship game, as one team would be heavily favored every year but would maximize the number of regular season games between the top team.
BCS Evolution -- Punctuating the Equilibrium - twitter
There has been much discussion about this to creat a “zipper” sstem, where all of the natural rivals are split into opposite divisions. This would allow each team one guaranteed game against the LA schools each season. The rivals are then also assigned as protected rivals.
This still breaks up the rivalries with the Northern and Southern California schools.
It will be interesting to see if the North-South California rivalries will survive the desire of every other team to optimize their games with USC and UCLA.
We are really talking a difference of one game every eight years between the zipper design and the ideas I propose.
That seems a fair price for preserving the four California round robins and the four NW school round robins.
The PAC 10 has my write up examining all the plans I have seen and providing a concrete implementation of my design.
BCS Evolution -- Punctuating the Equilibrium - twitter








