Road to C-USA title goes through Rice
CHARLESTON, W.Va. -- A sudden blip of support in the two polls associated with the Bowl Championship Series appeared to swing Marshall's attempt to host the Conference USA championship game, but the Thundering Herd will travel to Rice for Saturday's game.
In a clear result of the Herd's 59-28 win over East Carolina on Friday, Marshall received votes in the USA Today Coaches Poll and the Harris poll, each of which constitutes one-third of the BCS formula. Six computer indexes, such as the Sagarin ratings, make up the other third.
According to Conference USA officials, Rice's strength in the computer rankings outweighed Marshall's advantage in the polls. But Marshall supporters are questioning whether the application of the BCS formula was correct.
Marshall athletic director Mike Hamrick was not available for comment Sunday night. In a prepared statement, he said, "We're certainly disappointed that we will not be able to host Saturday's Conference USA championship in front of our wonderful fans.
"With that said, we are extremely excited for the opportunity to compete for a league title and look forward to a great game with Rice University."
Marshall and Rice, both 9-3 overall and 7-1 in league play, won their respective divisions to advance to the championship game, which is played at the site of the team with the best record. The Thundering Herd and Owls did not meet, so the head-to-head tiebreaker could not be used.
Per the league's description in its weekly releases, the tie is then broken by the team with the "highest BCS ranking."
The BCS formula's treatment of the polls is straightforward: Take the number of points earned in the poll and divide it by the number of possible points.
Marshall received 13 points in the coaches' poll, where there are 62 voters and a maximum of 1,550 points. That gives MU 0.0084 points for the BCS formula.
The Herd received 10 points in the Harris poll, where 105 voted for a maximum of 2,625 points. That earned the Herd a smaller slice, 0.0038.
By the BCS's guidelines, Marshall would have to average in the top 25 to even register in that category.
The methodology listed on the BCS's web site states: "Points will be assigned in inverse order of ranking from 1-25. A team's highest and lowest computer ranking will be discarded in calculating its computer rankings average. The four remaining computer scores will be averaged and the total will be calculated as a percentage of 100."
In fact, two teams ranked in Sunday's BCS top 25 registered a zero in the computer rankings, Louisville and Fresno State. So would Marshall and Rice, in that stated protocol.
Total up the three categories and divide by three, and you have Marshall with a BCS score of 0.0041. Rice, which did not receive votes in either poll, scores a zero.
On that count, the noon game Saturday would be at Joan C. Edwards Stadium in Huntington.
But not so fast, say the powers that be at C-USA.
Instead of basing the computer indexes on the BCS's strict top-25 level, C-USA decided to award values based on a 1-through-125 scale. After dropping the highest and lowest rankings, the inverse totals were tallied and divided by 500, or four times 125.
Rice had the advantage here, which was Marshall's fear before it registered votes in the "human polls." The Owls' computer rankings were 49, 42, 43, 59, 54 and 44; the Herd's rankings were 63, 41, 56, 66, 56, 54.
So after dropping the highest and lowest and taking the inverse values, Rice had 314 points, 0.628 after dividing by 500. Marshall had 275 points, 0.55 after dividing by 500.
Using that system, Marshall's 0.1874 didn't measure up to Rice's 0.2093. With that, Saturday's game will be at Rice Stadium in Houston.
That didn't much please Thundering Herd fans in the Twitter universe and elsewhere, who posed several questions Sunday night.
The obvious one was the modification of the BCS formula. C-USA's announcement Sunday declared that "the final analysis" was confirmed by Bill Hancock, BCS executive director.
Muddying those waters are 1-125 rankings published by CBS Sportsline, which stuck to the non-modified BCS formula and ranked Marshall 33rd, with Rice 48th. Earlier Sunday, numbers-cruncher Jerry Palm published projected standings, which pretty much were confirmed after the official 1-through-25 numbers rolled out.
Palm's involvement was part of C-USA's release:
"Please note that rankings like Jerry Palm's do not factor any computer poll values for teams not receiving votes and therefore are not reliable for this purpose. The Conference has been in contact with Jerry who fully agrees with this analysis."
The release quoted Palm thusly: "I worked with C-USA today on an extended BCS formula for breaking the tie between Marshall and Rice for home-field advantage in the conference championship game.
"It doesn't match the formula I use for my ratings on CBSSports.com, which serve a different purpose, but I believe it was absolutely appropriate and fair."
The weight of the computer scores poses another question. Rice's 0.628 score on that third of the formula falls between the scores for Baylor and Michigan State on the top-25 board, with Marshall's 0.55 higher than Oregon. It was the difference between those seemingly large scores that worked in Rice's favor.
Finally, the question of when the procedure was set and whether officials at Marshall and Rice were notified in advance may or may not be answered. Hamrick did not return a call by the Gazette for further comment, and the Conference USA phone number was not functioning.
Reach Doug Smock at 304-348-5130, firstname.lastname@example.org or follow him at twitter.com/dougsmock.