Playoff Epihany

Judging the Judges

December 5, 2014

Two days away from Selection Sunday (football variety) is as good a time as any to to give a no-whining review of the Committee's choices to date.

D1 Playoff version 1 is likely to be pretty uncontroversial, despite the best efforts of the Talking Heads to create some. To wit,

Baylor-TCU "head to head". Give me a break, winning a 119-point game by about the home field advantage at home does not by itself prove anything to me. Even the most ardent "head-to-head uber alles" proponent has an agenda if they consider that win "decisive."
Taking head-to-head to its logical conclusion, if you compare head-to-head-to-head... TCU is better than 114 other teams, Baylor is better than 103 (same as Georgia and Utah, see Division 1A Win Path Summary, and if you spend enough time with the teamlinks from that page you also get a feel for what "body of work" might mean if given a precise definition.)
"Good loss." This provided much comic relief on the talk shows, but it's only funny outside of context, which they didn't explain well enough for most people. I would have explained it as the "worst loss comparison." I have no problem with the concept that to compare teams with the same number of losses, a team that lost to #8 might be better than a team that lost to #30 (especially if the former has wins against teams better than #30.)
It's not so much that the loss to #8 was good, it is that #8's loss to #30 was not-so-good.
"Game control." Awkwardly explained in an untimely manner, but I actually think this is a legitimate criterion. On a predictions-game board that used to be around one of the players had a rating system that included "the quarter in which the winner took a two-score lead and never led by less afterward." I thought then (and think now) that is a very reasonable way to reward solid winners without having to account for garbage time blowouts in MOV-based systems.
I especially like that by referencing it the committee acknowledges that points (actual and "style") do matter. They always did, and one of the worst mistakes the BCS made was to try to pretend they didn't.
CFP Committee "poll." I said back at the beginning the whole "top 25" idea was a bad one. It turned out to be moreso than we thought. If they want to give us a weekly progress report to talk about (and sell ads to us) they really, really, really should publish something like this.
Just publish a list of teams being considered by any committee member for their individual top 4 along with the number of members considering each team.
My version based upon the computer rankings only includes actual top-4 "votes" by the computers, but the humans would be allowed to list more than four each. Heck, each human might list 25 teams they're "considering for top four", but the publication of a "top 25" is disingenuous. And if they use "top 25" votes to seed the field, they'll only get it right by pure luck.
Florida State. I think they have the Seminoles about right. Being undefeated is sexy and all, but if there are 10 teams that would be favored to go undefeated against a particular schedule then actually doing so warrants a top-10 rating, not necessarily a top two. Thanks Be to Western Kentucky for sparing us Marshall's inevitable repetition of 2007 Hawai'i (more than half of 1A would be favored over every one of Marshall's opponents to date, including WKU.)
Almost every way I calculate the consensus of the 117 objective rankings published by Dr. Massey I get the same top four the 12 humans came up with, in the same order. Love the conspiracy-theorists argument that "they're just trying to set up the most profitable matchups." They may well be, but if so they accidentally got the seeding right anyway.
By the way, the methods that don't match the committee's ranking do not have FSU as one of the "best four." 2014 may be a year that the BCS would not have gotten right.

How hard is it?

I had a playoff epiphany the other day while pondering that last bullet item. The general consensus is that the BCS' objective was to get a matchup between "the two best teams", and the only year where there was some doubt that the process worked was 2004, when there was no unambiguous way to determine which two of the top three were "the best two." While it's really cool that for the moment the Committee's top 4 probably are measurably the best four teams in the division...

...for the Playoff Committee's success factor our criteria shouldn't be "did they pick the four best teams?" It should be the same as it was for the BCS: "Are the two best teams in the field?"

Think about it. The great benefit of the playoff is that with four spots to fill it will be almost impossible to create a field that doesn't include the two best. If you can't decide between second and third -best (the only problem the BCS ever had, and only had once in 16 chances) it doesn't matter because you can choose 'em both. Since the BCS came into existance there never really was a three-team race for the #2 spot, so the fourth playoff spot just provides insurance that we get the top two in the playoff.

OK, you might be able to make an argument that, say, #5 is really better than #4 (or even #3) but if you can't make an argument that they should be #1 or #2, all of the rest of us have no gripe and you're just a homer. No controversy except that manufactured to sell papers (internet ads.)

So enough with "the committee has a hard job" stuff. Their only job is to make sure the "two best" teams are included, and they get to pick four!.

Long live the "best 2" + 2 format!