Tuesday, June 8, 2010

Balancing an Unbalanced Schedule -- A10 Conference Opponents 2011

Conference Match-ups, Agendas & Objectives
In the days before 14 and 16 team super conferences, a typical conference had no less than eight and no more than 10 members. The "regular conference" season consisted of each team playing every other conference member twice, once at home and once on the road (at the other member's home). These were often referred to as "balanced schedules" because (at least in theory) each team played every other team in the least and most favorable circumstances. As 11 and larger conferences became the norm schedule makers were faced with the challenge (or opportunity?) to develop schedules that met objectives other than "balance" (each team plays every other team twice, etc.). The schedules invariably involve play at least one game with every other team in the conference, and often two (mirror) games with one or more opponents. The mirror games provide the conference with opportunities to...
1. Develop attractive TV packages. Conference elites are matched to provide good TV story lines and interest beyond conference/team fan-bases. The 2006 Big East mirrors (the first season with 16 teams) matched Connecticut, Louisville and Villanova in a mini-round robin that was supposed to settle the #1, #2 and #3 seeds for the Big East Tournament. Louisville however lost three starters to injury and was hammered by their mirror opponents.
2. Preserve traditional rivalries. As Chris De Santis over at College Chalk Talk pointed out in his article "Win, lose or draw: A look at the A-10 pairings..." from last month, "... schools and coaches are not allowed to request partners. Schools are permitted to 'protect' one rival...", which allows Dayton and Xavier (cited by Chris) to renew their annual series, even as the conference's Big 5 Trio (La Salle, Saint Joseph's and Temple) typically have multiple home-away games, with the game that counts in the City Series (until recently) played in the "neutral" Palestra (Penn's home court).
3. Limit travel. The northeastern-based conferences (the Atlantic 10 & Big East in particular) have very large footprints. For the A10 Conference, which relies largely on D1 basketball to underwrite the varsity championships awarded by the conference in 21 separate (men's and women's) sports, stretches across eight states and the District of Columbia, from New England, south to Charlotte, North Carolina and west to St. Louis, Missouri, travel expenses are a considerable part of the Athletic Department budget. Biasing toward nearby opponents for mirror games can limit travel, especially during the winter months.

Atlantic 10 Announces Conference Opponents...Early?
Having covered the A10 for Rush the Court last season, a posted note over at Soft Pretzel Logic drew my attention. A bit early for schedule announcements perhaps, but I thought with more than a season's worth of comparative data, and a history of fair predictive capabilities using the Pythagorean Winning Percentage (among other stats), I wanted to see if applying a few quantitative measures might provide insight into the match-ups and how the regular season might play out. I also wanted to try out a variation of the Pythagorean Winning percentage that filtered for home games only and away games only.

The Mirrors -- Welcome to the A10 Alan Major
Reviewing the mirrors, it occurred to me that travel and distance may have been a criteria, since mirror games account for 37.5% of the conference schedule. Richmond has George Washington and Charlotte (and a Philadelphia-based team). Rhode Island has Massachusetts, a Philadelphia-based team and St. Bonaventure. Dayton has X, St. Louis and Duquesne, and so on. I averaged Pomeroy's adjusted efficiency ratings, and sorted from the highest Pythagorean winning percentage (most difficult -- on average -- opponents) to lowest (least difficult -- on average -- opponents) the match-ups sorted out in a decidedly "unbalanced" way...

Pythag WP
Charlotte0.826Same as 2009
St. Louis0.805Day & Duq
Xavier0.789Dayton only
Fordham0.781URI & Bona
Saint Joseph's0.773TU & UMass
Duquesne0.752Same as 2009
Dayton0.748Same as 2009
La Salle0.703Same as 2009
George Washington0.701Same as 2009
Richmond0.621Char & GWU
Massachusetts0.613Same as 2009
St. Bonaventure0.509Duq & FU
Rhode Island0.419FU & UMass
Temple0.384LSU & SJU

Pomeroy's conference-wide average for the A10 last season was 0.7114, so 0.673 suggests the schedule maker came fairly close to balancing the mirrors conference-wide, but the standard deviation (0.139...) from most difficult (Charlotte) to easiest (Temple),suggests that travel may have taken priority over balance. Temple's Fran Dunphy will no doubt include this schedule maker on his Christmas card list for 2011, as all of Temple's mirror opponents, Fordham, La Salle and Saint Joseph's, are not only very close but, having logged a collective conference record of 9-39, struggled in conference play. I have no doubt (regression to the mean...) that Fordham will be good enough to win at least one conference game next season. I suspect it won't be one of those games with Temple. Perhaps the scheduler believes the 55 year old City Series rivalry between La Salle, Saint Joseph's and Temple will make their games with Temple more competitive than the Hawks' and Explorer's collective 9-23 record suggest. The large number of all-repeat mirrors is surprising -- six conference teams (43% of the conference members) will play the same three opponents as 2009. If the bias is distance, the repeats make sense. Tannenwald's lamented "...Still, I wonder if someone in Newport News could have figured out a way to get all the Big 5 teams to play each other twice." could be solved by the three Big 5 Brethren without the active intervention of the schedule maker if each used their single "protected opponent" option (per Chris DiSano) in a round robin fashion. La Salle "protects" Saint Joseph's; Saint Joseph's "protects" Temple; Temple "protects" La Salle.

Home & Away Opponents -- Different Strategies & Circumstances
An elite conference team would most likely want to host their single meeting with other elite teams, and take the weakest conference opponents on the road if possible. For those teams a higher Pythagorean WP for home opponents would make sense. A team struggling to break the conference upper division however would most likely prefer to have it most "beatable" opponents at home and largely concede games with the elite teams by taking them on the road. An "activist" scheduler might want to use home-road game pairings to "rebalance" the schedule from the mirror games, and give a truer picture of the relative strengths/weaknesses of the members. Or the conference may simply flip the home-away opponents from the previous season (home opponents become road opponents, etc.)...

George Washington0.4390.639
La Salle0.3090.706
Rhode Island0.4360.710
Saint Joseph's0.4060.551
St. Bonaventure0.3690.739
St. Louis0.2330.538

I used the raw efficiencies to calculate a Pythagorean winning percentage for home and away games. Charlotte will host Dayton, Fordham, La Salle, Saint Joseph's and Massachusetts next season, in addition to their mirrors. Of that group, only Dayton had a Pythagorean Winning Percentage above 0.500 for it's conference road games last season. The 49'ers road slate by contrast, will be fairly challenging, as they face Duquesne, Rhode Island, St. Bonaventure, St. Louis and Temple (in addition to their mirror), all of whom, excluding St. Bonaventure, had a >0.500 Pythagorean winning percentage in their home conference games. Massachusetts drew the most challenging set of guests (Dayton, Duquesne, George Washington, Richmond and Temple -- 0.518), though Duquesne (Charlotte, George Washington, Rhode Island, Temple and Xavier -- 0.499) and Temple (Charlotte, Rhode Island, Richmond, St. Bonaventure and St. Louis -- 0.463) It does not appear the schedule maker simply flip home and road games however, as only Dayton, Duquesne and La Salle have no repeat home/road opponents. Every other team has at least one (Rhode Island has four, George Washington, St. Louis and Xavier have three) repeat home/away opponent. For the teams that will compete for top seeds in Atlantic City next March (Dayton? Richmond? St. Louis? Temple? Xavier?) the strong guest/weak host match-ups are not there, though for teams that may celebrate a middle seed in Atlantic City (Fordham? La Salle? St. Bonaventure?) the weak guest/strong host match-ups appear to be in place.

Limitations, Exclusions & Exceptions...
The conference received three bids to the NCAA Tournament, two bid to the NIT (both Rhode Island and Dayton advanced to the Final Four, Dayton won) and three bids to the CBI (St. Louis advanced to the Finals), a good (but not great...) season relative to the conference's history. Regression to the mean suggests that Temple, Xavier and Richmond will not be quite as good, nor Fordham and La Salle quite as bad. Sequence matters too, La Salle was better in December, when Ruben Guillandeaux, Yves Mekongo Mbala and Kimmani Barrett were healthy, than in February.

Thanks to Jonathan Tannenwald over at Soft Pretzel Logic for his post on the A10's home/away opponents for next season late last month. Jon's interest, largely all matters Big 5, writes for Philly.com and blogs about Big 5 and Ivy League sports.

No comments: