Question for everyone. In a game against a highly ranked team, do you think it is better to win big, or win a close competitive game? The obvious answer is that winning big is better. But this weekend's results got me thinking.
CFP rankings aren't out yet so I'll use the AP for this discussion. OSU and Wisconsin both won by large margins. And as a result, Minnesota dropped 6 spots and Michigan dropped 7 spots. Then you have Auburn that won a close game against Alabama. Alabama then dropped only 4 spots.
This might not seem like a big deal, but let's consider the drops of ranked teams after they lost to OSU and LSU.
Auburn dropped 2 spots, Florida dropped 2 spots, and Alabama dropped 2 spots.
Wisconsin dropped 5 spots, Penn State dropped 3 spots, and Michigan dropped 7 spots.
The discrepancy is likely a combination of SEC bias and the fact that OSU generally won big in those games (average margin of 24 points), while LSU won close "competitive" games (average margin of 7 points).
Now, as we near the final CFP rankings, LSU's wins are looked upon more favorably as a result of said discrepancy. Because OSU is nearly ranked number 1, they almost didn't gain anything from winning big. Where LSU effectively gained because the teams they beat dropped less and are therefore viewed as better teams all because the games were close.
Hopefully the CFP committee takes all of this into account, and based on what we've seen so far from this year's committee, I think they will. This also highlights the importance of the committee using multiple criteria and including the "eye test". While the talking heads want to only use data and national perception when it fits their narrative. Never understood why when the CFP chair gives a reason for a ranking, the talking heads act like that is the only reason and not a combination of all the data points and ideas. "Make up your mind, is it the eye test or analytics? Be consistent." Idiots. Rant over.
As an aside, I almost wish the committee would release a more analog rating tied to the rankings. Stick an arbitrary number with each team showing how strongly the team is viewed. The polls and rankings are too binary. If teams 10-15 are really close, figure out some way to show it so comparisons are taken in the correct context.