Tuesday, September 7, 2010

An Analysis of Pre-Season Rankings & BCS Bowl Game Participation

Editors Note:  A version of this post appeared at Playoff Pac here

The BCS likes to say that “at the beginning of the season, every bowl subdivision team starts out with an equal chance to become national champion.”1 When college football fans hear this claim, they probably experience a several-stage reaction: first laughter, then a pinch to ensure it’s not a dream; finally, they might double-check the quote to verify it was not being used in a sarcastic or humorous manner.

Tragically, though, this statement from Harvey Perlman, chairman of the BCS Presidential Oversight Committee, was all-too-serious when written in an Op-ed during the 2009 season.  Bill Hancock was also serious when he stated the following at the MAC Conference Media Day: “Everybody has the same chance to finish number one or number two.”2Of course, Playoff Pac believes that these claims are untrue for several reasons.    However, always willing to give ol’ Bill and Harvey the benefit of the doubt, I've decided to examine some numbers to see if the claim might have even a smidge of empirical validity.

I focused on the initial rankings in both major human polls, the Associated Press (AP) and Coaches’ Poll, for teams that have played in the BCS Title Game and the other BCS Games since their inception in 1998.  Specifically, I wanted to see where the teams who were selected to play for the BCS Title ranked in the preseason AP and Coaches’ Polls.  I did the same for the other BCS games with a particular focus on the initial rankings of teams who garnered at-large bids to BCS games.  Our goal was to track any correlation between initial poll rankings and which teams ultimately obtained BCS spots.  A more specific focus was on how likely it is an unranked team could earn a BCS at-large bid through on-field performance compared to schools which start with lofty preseason expectations.

First, I examined the 12 BCS Title Games.  The table below contains data for the initial season rankings from both human polls for the two teams that played for the BCS Title.
Lowest Rank

The mean values of 4.92 (AP) and 4.88 (Coaches’) indicate that a team that will play for the BCS Title starts out ranked in the top 5.  The other two numbers are even more revealing.  Amazingly, the lowest preseason rankings for any title game team were 19 (AP) and 20 (Coaches’) for Oklahoma in 2000.  In other words, if a team starts off the year outside of the top 20 in either poll, that team can pretty much forget the idea of being the national champions.  Also, the median value of 2 for both polls is telling, as it shows that half of the teams who played for the BCS title started the year ranked #1 or #2.  In addition, the only team in the last 6 seasons who played in a BCS title game after not starting in the top 5 was Florida in 2006 (#7-AP, #8-Coaches’). 

The 2000 Oklahoma team is an outlier in the sense that it was able to shoot up in the rankings because of 3 consecutive wins against in-conference top-11 opponents (Texas #11 10/7/00, Kansas State #2 10/14/00, Nebraska #1 10/28/00).  Thus, absent a scheduling quirk as shown above, if a team starts the year outside the top-15, its chances of becoming the national champion are seemingly nil.

What about the other BCS games though?  Surely with the recent success of teams from non-AQ conferences playing in and winning BCS games, the access should be much more open to lower-ranked or even unranked teams. However, the data indicates this may not be the case.

I compiled calculations to examine this issue by assigning any unranked team a ranking of “26” for mean and median calculation purposes.  Also, we only considered teams that earned at-large bids to BCS bowl games; I did not include teams who earned bids from winning one of the six Automatic Qualifying (AQ) conferences (because winning a conference is usually3 based solely on on-field performance). The results are shown below.
Lowest Rank

The mean values of 12.31 (AP) and 12.50 (Coaches’) indicate that teams that end up in BCS games, on average, start the year ranked fairly high.  This point is especially underscored by the median values: these values show, in both polls, that more than half of the teams who have earned BCS bids started the year in the top 9.4

While these overall numbers indicate that a high preseason ranking is essential, a few outliers buck the trend.  Seven teams, or 22% of the total 32 spots, have earned BCS bids after starting the season unranked (UR) in one or both human polls.  Two of these teams were from non-AQ conferences (Boise State 2006, Utah 2008).  Also, when looking at the last 4 years (2006-2009), when there were four at-large bids due to the creation of the fifth BCS game, at least half the teams who earned at-large bids started the season ranked 15 or lower in at least one of the polls.  Additionally, this streak actually extends back to 2004.5

After a full review of the numbers, the recent BCS success by teams such as Boise State (2006, 2009) and Utah (2008) has shown that a preseason ranking may not be prevent your favorite team from making a BCS Bowl.  However, the fact remains that every team does not start out with an equal chance to become national champion or to even play in a BCS bowl game.  Rather, a team’s preseason ranking still has a much larger correlation in determining which teams get selected to play in BCS games, especially the BCS Title game.  Half of the teams who have played for a BCS Championship started the year ranked either #1 or #2 in both human polls.  These numbers show that at least as to the hundreds of teams not named Texas or Florida, the BCS is fixed before the first snap.  These numbers refute the fallacy that “every bowl subdivision team starts out with an equal chance to become national champion.”  

Editor’s Note: The 1999 Coaches Poll rankings from week 1 were used as the Pre-Season Rankings were unavailable and not listed in the ESPN College Football Encyclopedia


1.  http://www.washingtontimes.com/news/2009/nov/22/perlman-deciding-ncaa-football-championship/?page=2
2.  http://www.themorningsun.com/articles/2010/08/02/sports/srv0000008988157.txt
3. *Texas in 2008 finished tied with Oklahoma & Texas Tech in the Big 12 South division but Oklahoma got selected for the Big 12 Championship game due to its higher BCS ranking). 
4.  Note: Oklahoma (2001) and Nebraska (2003) are included in both this set of calculations and the previous one since they each played in a BCS title game despite not having won their conference and thus, had not earned an automatic bid.  If they are removed from the calculation the Means do not change significantly, 12.97 and 13.17, but the Median values drop to 12.5 for each poll. 
5.  A couple of points need to be made regarding our categorization of at-large bids.  First, Notre Dame, which has special BCS privileges, is responsible for 3 of the 32 at-large bids, including 2 of the 7 UR bids. This is important because Notre Dame has a special set of rules built into the BCS.  The current rules, effective starting in 2006, mandate that Notre Dame earn an automatic berth if it is ranked in the top-8 of the final BCS standings.  Thus, while I considered it an at-large bid for the purposes of this study, it does not need face the same level of difficulty starting the season ranked outside the top-20 or even unranked that any other team does.

Second, the BCS adopted a rule beginning in 2006, which stated that any conference champion from a non-AQ conference would earn an automatic BCS game bid if the team finished in the top-12 of the BCS standings or in the top-16 while being ranked ahead of at least one of the AQ conference champions.  However, only one such bid is reserved for teams which meet this criterion as shown when Boise State was forced to sweat out an at-large bid in 2009.  Thus, while I considered Utah (2008) and TCU (2009) as at-large bids for purposes of this study, they each technically earned automatic bids to BCS games.

Third, while it appears that Notre Dame’s criteria may be more difficult (top-8) than non-AQ teams (top-12), one should remember that it does not have to win a conference and also has natural advantages which make it much more desirable to BCS officials than your average mid-major school. 

Game Results Compiled from

Poll Data compiled from the following sourcesESPN College Football Encyclopedia (1998, 1999 Coaches Poll)
http://www.infoplease.com/ipsa/A0774952.html (1998 AP Poll)
http://sportsillustrated.cnn.com/football/college/news/1999/08/14/ap_poll_ap/   (1999 AP Poll)
http://www.nationalchamps.net/NCAA/polls/AP_Poll.htm (2000)
http://www.nationalchamps.net/NCAA/polls/Coaches_poll.htm (2000)
http://www.nationalchamps.net/NCAA/polls/AP_Poll_2001_Preseason.htm (2001)
http://www.nationalchamps.net/NCAA/polls/Coaches_Poll_2001_Preseason.htm (2001)
2002 – 2009 AP and Coaches Polls - http://espn.go.com/college-football/rankings

No comments:

Post a Comment