Rivalry Weekend's BCS Impact

Although the BCS has been less than fair to Florida State throughout this season, if the Seminoles can knock of No. 4 Florida Saturday, there could be some major changes coming.

Florida State fans have complained about the Seminoles' low computer ratings since the BCS was released this year, as there is little doubt that the 10-1 ‘Noles are better than their average ranking of 17th in the BCS computers. Losses by previously undefeated Kansas State and Oregon this week only poured gasoline on the fire, as now the ‘Noles would seemingly have a shot at the title were it not for being so badly underrated in the computers. But savvy veterans of the BCS process have continued to preach patience, since the only BCS ratings that matter are those in the final week and things tend to correct themselves as the season goes on.

Without question, the current system is hopelessly flawed. The computer component of the BCS formula depends on computer formulas objectively measuring the data from the season and comparing each team's resume. The more data the computers have to analyze, the better "informed" and more accurate their rankings. But the politically-driven decision to eliminate margin of victory from consideration by the BCS computer formulas hurts the computers' ability to compare teams (since it effectively removes mountains of additional data the computers could otherwise access) and leads to less reliable results. All that said, the system is set and that's not changing. What matters now is that the end of year ratings have significantly more data factored into the formulas and are typically much more reliable than those earlier in the season.

In contrast to their Tallahassee rivals, the computers love Florida this year (current BCS computer average of #2), as the Gators have wins over 9-2 Texas A&M, 9-2 LSU, and 9-2 South Carolina, with their only loss coming against 10-1 Georgia. Without a doubt, a win over Florida would be the Seminoles' best in the eyes of the computers and would move them up a good bit in the computers. The real question at this point is how much—a question that could matter a great deal if Notre Dame should happen to lose at rival USC later Saturday afternoon.

The important thing to remember when thinking about the BCS computer formulas is that they all basically apply the transitive property to college football using sophisticated algorithms to compare each team based on the various "connections" established with each game. Consider the following example:

Florida wins at Texas A&M who then wins at Alabama.

If those are the only three data points available, every computer system will rank the teams as follows:


Texas A&M


Let's add another few data points:

Florida beats LSU who then loses at home to Alabama but wins at Texas A&M.

Obviously this makes things much more complicated, as three teams have now all beaten each other, with none winning over the other two. Thus the computers have to weigh which wins matter the most. In this example, the only clearly obvious ranking to the computers is to put Florida (wins over Texas A&M and LSU) at the top. The other three will be ranked slightly differently by the various systems depending on how the algorithms value home/away, when the game was played, etc.

This is why the computers love the Gators so much this year: wins over Texas A&M (beat Bama) and LSU (beat Texas A&M) are very highly valued while their win over USC (beat Georgia) helps mitigate their loss to Georgia.

Let's take a closer look at Florida State's case from the perspective of the computers:

Florida State lost to NC State who lost to Tennessee who is 0-7 in the SEC.
Florida State beat Clemson who beat Auburn and is 0-7 in the SEC

So in the absence of other connections, from the perspective of the computers, Florida State is worse than all seven SEC teams to whom Tennessee has lost but is better than Auburn.

This is why FSU currently ranks below six SEC schools in the computers, despite having a better record than those teams. But this weekend is where this can all change and change dramatically. Whereas NC State is regarded as worse than an 0-7 SEC team, Florida is 7-1 in the SEC, with the computers considering them as either the best or second-best team in the SEC. A win over Florida would work in much the same way for the ‘Noles that Florida's win over South Carolina presently does: it effectively offsets the team's single loss, since they beat the team that beat the team that beat them.

In Florida State's case, the effect of a win over Florida would be even greater because the effect of NC State's loss only indirectly affects FSU's position relative to the SEC, while a win over Florida is a direct win over a top SEC school. (Crudely put, the NC State game provides 1/3 of FSU's connection to the SEC while the Florida game provides 2/3 and is weighted more heavily.) Essentially, a win over Florida would likely vault FSU above every team Florida has beaten in the computers, a position they are currently kept from due to Tennessee's poor SEC performance and the ‘Noles' loss to NC State. Here's a simple way to summarize its effect:

Present situation:

Florida > LSU > Texas A&M > South Carolina > Tennessee > NC State > Florida State

If FSU beats UF

FSU > UF > LSU > TAMU > South Carolina > Tennessee > NCSU > FSU

Even if FSU's position on that scale based on two games were averaged out, the ‘Noles would be even with South Carolina. But since the UF game as a direct connection would receive more weight, the ‘Noles' final position would be higher on that scale.

But this weekend involves more than just one key game for the computers' opinion of FSU. Clemson's game against South Carolina (6-2 SEC) offers the opportunity to add yet another data point that gives that computers another chance to compare FSU to the SEC and the teams the SEC has beaten. So far, Clemson's win over Auburn has been next to worthless (though better than a loss) due to Auburn's complete meltdown. But a win over a South Carolina team that beat Georgia would vault the 10-1 Tigers (who do not have a loss against a team that lost to an 0-7 SEC team) close to Georgia in the computers—and would add tremendous impact to FSU's defeat of a Clemson team suddenly viewed by the computers as equal to a top SEC team.

If FSU beats UF and Clemson beats South Carolina

FSU > UF > LSU > TAMU > Clemson > South Carolina > Tennessee > NCSU > FSU

Should both FSU and Clemson win this weekend, expect both of their computer ratings to skyrocket, putting them right around and perhaps even above 10-1 Georgia (current CPU rank 6th) in the computers, simply because the NC State albatross would be deemphasized by the computers, who would then have two direct connections in favor of FSU and Clemson outweighing the single connection of NC State's loss to Tennessee and its indirect connection to the rest of the SEC.

My guess is that even without a very unlikely Georgia Tech victory over Georgia, Florida State's average computer rank would jump from 17th to about 8th with a win over Florida and well into the top 5 if Clemson also beat South Carolina. That still may not be enough to play for a national title (and certainly would not be without a Notre Dame loss), but it's at least a lot closer to where the ‘Noles would deserve to be ranked at 11-1.

Nole Digest Top Stories