Wednesday, December 3

Do You Know Who's Picking Your National Championship?

We're on the record supporting a twelve team playoff, but since college football still doesn't have one, we turn our focus to the BCS. Every college football fan understands that the BCS rankings rule the roost, but most couldn't name half of the computer polls that annually can play a tremendous role in the selection of a national championship match-up. All but one of the polls are shrouded in mystery regarding their selection process and more often than not, they produce some questionable results. With the systems being "secret" there exists no checks and balances to ensure proper results. It's with this skeptical eye that we took a look at each of the polls that comprise the BCS and quickly found many glaring flaws. More after the break...

The Harris Interactive Poll is, “[c]omprised of former players, coaches, administrators and current and former members of the media who have committed to submit rankings for the top 25 college football teams each week. These panelists were randomly selected from among more than 300 nominations submitted by the 11 Division I-A Conference offices and the independent schools. The panel has been designed to be a statistically valid representation of all 11 Division I-A Conferences and independent institutions.”

Individual Harris poll voters have found themselves routinely criticized at the the end of the season (the only time votes are made public). This week, a Harris voter, Pat Quinn, was quoted as saying, “I think Alabama and Penn State will probably play for the national championship. They're the only two undefeated teams, aren't they?” Harris Interactive attempted to apply a standard practice that could counter error or bias, but were denied by the conference commissioners (and the Notre Dame AD). The Harris Interactive poll accounts for one-third of the BCS standings. A full list of voters can be found here.

USA Today Coaches Poll – Composed of 61 D-1 coaches, the poll accounts for one-third of the BCS poll. Of the 61 head coaches, eight or 13% of the voting pool will not be coaching their team next year. Tommy Bowden, Joe Glenn, Ty Willingham, Jeff Genyk, Sylvester Croom, Phil Fulmer, Rocky Long, and most recently Hal Mumme have all been fired or resigned. Hal Mumme, no longer the coach at New Mexico State, was the coach that last season voted Hawaii number 1 heading into the bowl season. A point of intrigue with the poll, Oklahoma edged Texas by a mere one point with two first place votes to none for the Longhorns. The two votes for Oklahoma in the top spot appear to be from Mike Leach and Sooners head coach, Bob Stoops EDIT:Stoops had a vote last season, but does not have one this year (HT: CGB). Mack Brown also has a vote, but chose Alabama at the top spot. Had Mack been self-serving, the teams would've been tied in the poll. Of note, Urban Meyer does have a vote, potentially explaining the one top vote the Gators got. A list of voting coaches and a conference breakdown can be found at the end.

The final third of the BCS rankings are composed of six computer rankings, with a team's highest and lowest ranking being thrown out before computing their ranking.The six computer ranking providers are Anderson & Hester, Richard Billingsley, Colley Matrix, Kenneth Massey, Jeff Sagarin and Peter Wolfe. After issues with teams with weak schedules having an advantage, each ranking now has a strength of schedule element.

Anderson & Hester – They boast the last original and unchanged computer ranking used in the BCS. They also claim to have the most accurate strength of schedule and conference strength rankings. The strength of schedule component makes their rankings even more curious. They have Utah ranked at number 2, behind only Alabama. The top two schools have SOS rankings of 56 and 61, while number 3 Texas and 4 Oklahoma have the fifth and seventh ranked schedules. Despite having such a pronounced advantage in strength of schedule, Utah earned the nod solely for being undefeated. Looking at Texas and Oklahoma's schedule, the lone ranked team that Texas faced that Oklahoma did not was Missouri, and they played no team of note in non-conference games. Oklahoma, on the other hand played TCU and Cincinnati. Another flaw appears when looking at the conference rankings. They have the ACC being the top conference, with the Big 12 second, the SEC third, and the Big East fourth. Most glaringly the rankings place the SEC closer to the Big East and the Big Ten than either the Big 12 or the ACC. Another telling sign of the flaws is the ranking of Oklahoma State. They're 9-3, with all of their losses coming to teams that were at one point in the season ranked number one. They find themselves ranked 19 and behind the likes of Georgia Tech and Michigan State who have the same records, but without facing that level of competition. They also find themselves behind Missouri, who has the same record and who OK State defeated in Missouri.

Jeff Sagarin – The computer poll with the most name recognition currently has Oklahoma, Texas, and Texas Tech as the top three. According to Sagarin's poll, three of the top ten teams have not defeated a top ten team this season: Boise State, and the two SEC heavy hitters Alabama and Florida. The top four SOS rankings all come from the ACC: Virginia, Duke, NC State, and Wake Forest. Texas and Oklahoma are the only two top ten teams with SOS rankings in the top 20 (12 and 17, respectively). Consensus number one in the human polls, Alabama's SOS (73) is only three spots higher than TCU's (76). The biggest gripe against Sagarin's rankings this year: FCS top team James Madison is currently ranked 24th.

Richard Billingsley – Billingsley implements a starting ranking for teams, that can impact their rankings similar to the human polls. Billingsley, a self-proclaimed Oklahoma fan, did start the season with the Sooners at a higher ranking than the Longhorns. The only computer other than Anderson & Hester with Alabama at the top spot, followed by Oklahoma and Florida. Texas comes in at four, despite a tougher schedule and their one loss coming on the road to a top team, as opposed to Florida who lost at home to an unranked team. The glaring errors appear in the 9 and 12 spots. Number 9 is Ohio State, one spot above Penn State despite the Nittany Lions win in Columbus. At 12, Billingsley currently has BYU. They're ranked above TCU, who has the same record and destroyed BYU 32-7 in October. He also has conference rankings posted with the Big 12 at number one. Curiously, though he has the SEC at 2 and the ACC all the way down at five, behind both the Big and Pac 10. What's so curious about the ACC's ranking, they have 11 teams ranked in the top 53. Conversely, half of the SEC is ranked 55 or lower. Clearly, this system is heavily biased towards conferences with an elite team. The individual SOS rankings are also cause for concern. He understandably has Baylor at 1 having faced all three of the Big 12 south teams, but 2, 3, and 4 all come from the SEC: Mississippi St, Georgia, and Florida. This is largely because Billingsley gives credit for the opponent's ranking “at the time”. This led to Florida getting credit for beating three top ten teams this season. Their wins over LSU, South Carolina, Hawaii and Georgia all appeared to be more impressive at the time. A ranking like this does not provide an accurate assessment of a team's schedule.

Colley's Matrix – The only open-source computer ranking has Texas ranked number one, followed by Oklahoma and Alabama. Texas's SOS of 4, seems high compared to Oklahoma's 13 which was discussed before. Alabama's SOS of 82 is by far the highest of any top 10 BCS conference school. Colley has the Big 12 as the top conference, followed by the ACC, who barely edged out the SEC. Interestingly, the Pac-10 is far closer to the MAC than to the Mountain West. As a fan, these rankings appear to most closely resemble my sensibilities. It would have Texas in the Big XII championship and assuming victory facing the winner of the SEC championship.

Massey Ratings – Oklahoma, Texas Tech, and Texas serve as the top three, but Utah comes in at four ahead of both Alabama and Florida. Most concerning about this poll is the fluctuation. Teams regularly jump multiple places (up or down) even at this late time. Georgia Tech and Boston College jumped 7 places, Boise St and Utah 3 spots into the top 10, but Florida and Florida State both remained unmoved after their game. Georgia was number 7 until their loss to Georgia Tech. Florida sits at 6 and Georgia at 7 weeks after their game? Sigh, these are the people picking our championship game.

Dr. Peter Wolfe – The Big 12 triumvirate reigns with Oklahoma at the top. Alabama and Utah round out the top 5. With Boise State finishing ahead of Florida. Another poll that seems to have difficulty dealing with undefeated teams, regardless of quality of opponents faced. These rankings have Florida closer to BYU than Texas Tech. OF note, at 16 and 17 Georgia and Georgia Tech with the same record, and Georgia Tech having just defeated Georgia.

Most of the people behind the computer rankings will acknowledge that they're just a piece of the puzzle and not a fail safe solution, but there are far too many glaring weaknesses in the majority of these rankings. They seem to be in the ballpark, but fall well short particularly when trying to gauge undefeated teams. If we assume they're 95% accurate, that's far from acceptable. If a doctor performs surgery and it's 95% successful, it probably wasn't successful. If that seems too extreme, if you go to a drive-thru window and they get your order 95% correct is that acceptable? These computer ranking by and large have major flaws and no other billion dollar business would accept such shoddy work out of people working on the companies golden goose. Again, the only logical move is to a twelve team playoff. Annually, the BCS fails some conference and some team. This year is setting up to be no different.

For more on the computers, this recent article gives background and additional insight.

USA TODAY Coaches Poll voters:
Tommy Bowden, Clemson
Joe Glenn, Wyoming
Ty Willingham, Washington
Jeff Genyk, Eastern Michigan
Sylvester Croom, Miss. St.
Phil Fulmer, Tennessee
Rocky Long, New Mexico
Hal Mumme, New Mexico St

ACC (6)
Frank Beamer, Virginia Tech
Bobby Bowden, FSU
Tommy Bowden, Clemson *
Butch Davis, UNC
Jim Grobe, Wake
Tom O'Brien, NC St

Big 10 (7)
Jim Tressel, Ohio State
Bret Bielema, Wisconsin
Mark Dantonio, Mich. St
Bill Lynch, Indiana
Rich Rodriguez, Michigan
Joe Tiller, Purdue
Ron Zook, Illinois

Big 12 (7)
Art Briles, Baylor
Mack Brown, Texas
Gene Chizik, Iowa St.
Dan Hawkins, Colorado
Mike Leach, Tx Tech
Bo Pelini, Nebraska
Gary Pinkel, Missouri

Big East (4)
Greg Schiano, Rutgers
Randy Edsall, Uconn
Steve Kragthorpe, Louisville
Jim Leavitt, S. Florida

Conf. USA (6)
Neil Callaway, UAB
Todd Graham, Tulsa
George O'Leary, Central Florida
Mike Price, Texas-El Paso
Mark Snyder, Marshall
Bob Toledo, Tulane

MAC (6)
Jeff Genyk, Eastern Michigan*
Turner Gill, Buffalo
Butch Jones, Central Michigan
Doug Martin, Kent State
Shane Montgomery, Miami (Ohio)
Frank Solich, Ohio

Mountain West (4)
Joe Glenn, Wyoming*
Rocky Long, New Mexico*
Gary Patterson, TCU
Kyle Whittingham, Utah

Pac-10 (5)
Mike Bellotti, Oregon
Rick Neuheisel, UCLA
Mike Riley, Oregon St
Jeff Tedford, California
Tyrone Willingham, Washington*

SEC (7)
Steve Spurrier, S. Carolina
Sylvester Croom, Miss. St.*
Phil Fulmer, Tennessee*
Urban Meyer, Florida
Les Miles, LSU
Mark Richt, Georgia
Tommy Tuberville, Auburn

Sun Belt (4)
Mario Cristobal, Florida International
Todd Dodge, North Texas
Howard Schnellenberger, Florida Atlantic
Rick Stockstill, Middle Tennessee

WAC (4)
Pat Hill, Fresno State
Hal Mumme, New Mexico State*
Chris Petersen, Boise State
Dick Tomey, San Jose State

Charlie Weis, Notre Dame

1 comment:

Kelly said...

Great piece man. This clearly illustrates the deficiencies in the BCS. I love the fact that this is the method that determines who plays for the national championship. I think we'd be better off just picking two schools out of a hat.