CFP Selection - what exactly are the criteria the Committee uses? What are they, and what should they be?
The second release of the CFP Selection Committee rankings came out last night - and, there are a few threads on the teams, their ranking, etc.
Is it clear to anyone what the Selection Committee's criteria actually are?
The following list is the five factors the Selection Committee supposedly uses for the identification and selection of a team's ranking.
- Conference championships won,
- Strength of schedule,
- Head‐to‐head competition,
- Comparative outcomes of common opponents (without incentivizing margin of victory), and,
- Other relevant factors such as unavailability of key players and coaches that may have affected a team’s performance during the season or likely will affect its postseason performance.
Do you think the Selection Committee has actually been watching the teams play and using these criteria?
One thing I find ironic is - the first item - "Conference Championships won". Is this the current season? Does it matter who won their Conference Championship last year? Do they not consider prior years' records and performance?
Strength of schedule is also interesting. Yes - everyone (including myself) wishes that Michigan's OOC was stronger. This does, again - IMO, lead to a conference schedule perspective. B1G teams play 9 conference games, as do the Pac-12 & Big 12 - by definition, these are P5 teams. The SEC and ACC play 8 conference games, so - the SEC / ACC can schedule another P5 team - and, still only play 9 P5 games during the season - as the B1G, Big 12, and Pac-12 already do.
Which non P5 teams did Tennessee play this year: Ball State, Akron, and UT- Martin. How about Bama: Utah State, UL-Monroe, and Austin Peay. LSU: Southern, University of New Mexico, and UAB. Georgia: Samford, Kent State, and Georgia Tech (which is a P5 team). Ole Miss: Troy, Central Arkansas, and Tulsa. How does Michigan's OOC schedule compare to these? Last I checked, Michigan did not play an FCS opponent this season.
If LSU wins out - they would be the SEC Conference Champion - and, both Georgia and Tennessee would be one loss teams without a conference championship. So, which is the "first" SEC team in the CFP? Would it be LSU - a two loss team? Georgia - who would have lost the SEC CCG to LSU? Or, Tennessee, who neither won their division or conference?
My biggest issues, for a while, have been the subjectivity of the criteria and inconsistency of the Selection Committee. I'd welcome your perspectives / thoughts on this. I've viewed it as a beauty contest, and "do-overs" for select SEC teams.
Do you agree with the criteria? What would you remove, add, or change?
Thanks, and Go Blue!
November 9th, 2022 at 11:38 AM ^
Answer: whatever it takes to get two SEC schools into the playoffs. The methods change every year, but the end objective remains the same.
November 9th, 2022 at 11:48 AM ^
Selection criteria:
1. Do you play in the SEC?
2. If not, did you play anyone in the SEC?
3. If no, are you undefeated playing in either the ACC, B1G or PAC12
4. If no - disqualified
November 9th, 2022 at 12:21 PM ^
i know this is written as a joke/sarcasm, but i believe it to be 100% true until proven otherwise
November 9th, 2022 at 3:07 PM ^
With all due respect to Indy Pete's insight and wisdom, I think the criteria are actually more nuanced than that.
1. Alabama, until it's too blatant to choose them.
2. SEC
3. SEC
4. Other Southern school
5. Big Ten if absolutely forced to.
November 9th, 2022 at 12:06 PM ^
To be fair, the actual goal is to get three SEC teams in, but this is difficult to do without making the bias transparent.
November 9th, 2022 at 12:45 PM ^
If their goal is not to make bias transparent, they're doing a terrible job.
November 9th, 2022 at 12:51 PM ^
I just don't see that. The committee is made up of representation of conferences across the country. The Chairman is from the ACC. I don't know why the AD from Wyoming, Colorado, Kansas State, The Naval Academy or MICHIGAN....would have any interest in putting three SEC schools in the playoffs.
November 9th, 2022 at 1:13 PM ^
It is not simple. My conspiracy is they all scratch each other's back. Do you want me to schedule you in SEC so you can take money for a game? I am sure it will be subtle but I don't doubt backroom shenanigans are not going on.
November 9th, 2022 at 2:15 PM ^
Is this tongue in cheek? Its fun to joke about, but kinda of tin foil hat to think the committe members, which has B1G and B12 representatives, is ok with harming their conference regarding representatives on the biggest stage. And a portion of the members rotate every year, so that makes it even harder to collude like that.
Is there a bias? Probably so, but somehow OSU got in when they didn't win their conference.
November 9th, 2022 at 12:48 PM ^
The criteria:
1) media bias
2) personal bias
3) stupidity
November 9th, 2022 at 11:43 AM ^
There are no rules. Just money.
November 9th, 2022 at 11:46 AM ^
The Playoff is an "invite". They've said this multiple times. They'll pick who they want.
November 9th, 2022 at 11:47 AM ^
There are no conference champions yet. That gets factored in later.
November 9th, 2022 at 12:25 PM ^
This is true. And when the CFP states a criteria of "Championships won" it also considers division championships, though to a lesser degree than conference championships. So, for example, a team like Tennessee could theoretically drop one week when division titles are clinched, and then again the next week when conference championships are determined.
To respond to another item mentioned in the OP, this chart (open in new tab) shows the quality of competition among the top 12 CFP teams according to a composite of statistical rankings. Only 4 of the 12 have not yet played an FCS program (Alabama will play its FCS school on ChickenShit Saturday™). That means only 2 teams will have not played a bottom-tier team (i.e. a team in the yellow).
November 9th, 2022 at 12:37 PM ^
Thanks for posting this. It's a really interesting chart.
November 9th, 2022 at 12:43 PM ^
That chart was derived from this chart:
November 9th, 2022 at 3:13 PM ^
ChickenShit Saturday is an excellent coinage. That it comes so late in the season is infuriating.
November 9th, 2022 at 3:37 PM ^
h/t Ramzy @ 11W.
November 9th, 2022 at 11:49 AM ^
Great post with straight Facts!
I personally believe that one major l issue is and always will be, that ESPN plays a major factor in some of these picks with their deal with the SEC.
November 9th, 2022 at 11:57 AM ^
They use a spinny-wheel with 6 slots - 5 SEC teams and one rotating slot they use for Not-SEC teams one at a time.
Four quick spins, then names pulled from a fancy hat.
November 9th, 2022 at 11:57 AM ^
When the criteria is broken down and you factor in the closed door committee meetings it’s just a farce. The CFP format is trash in my opinion. As a B1G school id rather beat OSU and go to Pasadena. Like when I really think about it, it’s an ESPN funded and backed closed door meeting to decide who plays in the biggest games of the year. It has zero material determination on who is and is not the best team in the sport. Kind of conference isolationist of me, but with USC and UCLA joining, the only other teams I’d actually want to see in our conference are Oregon and Notre Dame. There really isn’t much outside of that interests me at the P5 level.
November 9th, 2022 at 2:04 PM ^
I mean saying it has zero determination on the best team isn't accurate. Georgia wasn't the best team last year? It may not be perfect but it's way better than anything that came before it. Split championships are lame, and BCS was terrible.
I respect your right to your opinion, but I continue to be surprised by those that say I don't care about national championships, just big ten championships.
November 9th, 2022 at 5:15 PM ^
Split championships are lame
Yes, they're lame when yours is one of the teams with the split championship ('47, '97). But otherwise who really cares?
"But we have to determine the 'best team in the country'"? I will guarantee you it's one of the teams that plays on Sundays.
Frankly I can't feel that what we've had the last twenty-five years has been more enjoyable or more satisfying or been better for college football than the old bowl game system with "mythical national championships", even when it did mean there were arguments to last the whole off-season.
November 9th, 2022 at 5:39 PM ^
So you'd rather have the coaches (or really SIDs) and AP voters deciding the best teams than settling it on the field? You could just say it's your record and who cares about rankings, but top 5/top 10 finishes matter a lot for recruiting.
I do think college football postseason has been better the last 25 years, but I respect that you feel differently.
November 9th, 2022 at 5:24 PM ^
I suppose that I didn't phrase it as good as I could have. The CFP has not added any meaningful value to choosing the best team in the country. Georgia and I'm sure any other national champ would have been picked by the computers for a national championship game the same with the old system. The methodology is inherently biased and isn't determined by the people who know football the best in the country. The committee provides no value to the college football as a sport, plain and simple. If there was a process that was in place to actually determine the best team in the land it'd be one thing, but it's a joke of a method for determining the most qualified times. Their framework is a joke, the conflicts of interest are obvious. The current system is more of a joke than the old AP polls or the BCS. The notion that two teams could both be worthy of sharing a title is fine, if they got matched up great, but leaving it to a TV network and a bunch of ADs who I'm sure are fine people but have no incentive to create the best on field matchups. It's just a trash system at this point I'd rather any of the other attempts at trying to determine who's the best team in the country.
November 9th, 2022 at 5:49 PM ^
I appreciate your response and clarification! I do think that the committee who deliberate for hours are more knowledgeable than AP voters or coaches who just have time to check the box scores. And both groups have plenty of conflicts of interest themselves, especially the latter.
If there are improvements, let's push for them (more teams, different members, addition of computer rankings) but going from playoffs back to just bowls feel like moving in the wrong direction.
November 9th, 2022 at 3:42 PM ^
That NC trophy shines bright when recruiting.
November 9th, 2022 at 11:58 AM ^
Note that record is not mentioned, like at all. This is so they can rank 2-loss SEC teams higher than 1-loss or undefeated teams from other conferences. Hell, why not put a 6-loss SEC team in?
/s but still strange that something like "overall record or performance" is not the first thing listed. I suppose it's implied in the "conference championship" piece, but still.
November 9th, 2022 at 11:58 AM ^
I would honestly prefer a return to a BCS-style ranking system using a composite of advanced stats rankings than a committee.
The committee is inherently biased and can't even usually back up their decisions with coherent/consistent arguments. It's a joke. Let computers do it, and then we can quibble over the algorithms but if there's a composite it'd be hard to argue that various advanced stat ranking systems were collectively biased.
November 9th, 2022 at 12:07 PM ^
The BCS formula wasn't perfect ...... but one thing it was: it was 100% transparent. Everyone knew how the computers/polls/etc were weighted and even the algorithms for the 6 computer polls used were publicly available.
Now it's 12 people deciding things behind closed doors.
That's not a step backwards. That's 1000 steps backwards.
I'm 100% with you - go back to the formula, ditch the committee. The only losers in this case are the committee members who don't get a paid junket to DFW (to eat steaks and stay in 5-star hotels) every week in November.
November 9th, 2022 at 2:58 PM ^
I am behind this as well. It will shift the importance to polls which has a higher number of people. It will be better but not sure it will satisfy everyone. Looking at ESPN for example, we know there are people there who wouldn't be objective. But it is a bigger pool and more often than not they are probably going to be better...
November 9th, 2022 at 3:07 PM ^
For real. I wanna be in the room where it happens, the room where it happens.
(Are Hamilton jokes still topical?) ¯\_(ツ)_/¯
November 9th, 2022 at 5:24 PM ^
we can quibble over the algorithms
IIRC everyone did and hated "the computers" in large part because the algorithms had the possibility of being gamed (hence the line about ignoring margin of victory, I would suspect), and they ignored the all-important (!) "eye test."
Polls are subject to even more manipulation: coaches, fans, (and even media) have every incentive to upvote their own teams and opponents and downvote others.
November 9th, 2022 at 12:00 PM ^
you forgot a step
0. Rearrange the below to best benefit the SEC
November 9th, 2022 at 12:03 PM ^
To your question about SoS you're way too in the weeds about this. There are metrics that encompass the totality of non-conf and conference play and I think the committee uses these. Take for instance the SoS of the top 5 teams:
ESPN SoS:
UGA - 47th
OSU - 53rd
UM - 74th
TCU - 68th
Tenn - 2nd
So the committee is largely ranking the undefeated teams by SoS which makes sense. They're all 9-0, so then look at who beat the best teams?
And they deviate a little from this with Michigan and TCU. They're saying, eh, we don't quite buy TCU because they won a lot of close games and several of those teams were missing QBs when they played.
There's no getting around Michigan's awful schedule right now. It's not just OOC. The Big Ten is down and Iowa, MSU, others have not helped M's SoS. Michigan has played only one ranked team and only three with a winning record. That doesn't compare favorably.
Regardless, I think they're in great shape to possibly stay in the top 4 with a close loss to OSU. We should be happy that it appears the committee is watching the games which is why we're ahead of TCU.
November 9th, 2022 at 12:23 PM ^
Even on SoS, there is a philosophical argument:
- If we consider SoS as "how much better did this team do versus how an elite team would do?" --- Georgia has a tougher SoS than Ohio State. Georgia has played 2 teams that are clearly Top 10 in the various advanced metrics (Tenn, Oregon), while Ohio State has played 1 that is borderline Top 10 (PSU) and that's it.
- If we consider SoS as "how much better did this team do versus how an average team would do?" --- Ohio State has a tougher SoS than Georgia. Ohio State has played more teams that are clearly in the Top 50 of the various advanced metrics (ND, PSU, Wisky who SP+ and FEI both love, Iowa) than Georgia has (only Tenn, Oregon and Florida).
Who has the better SoS? I'd go with Georgia (philosophical argument #1 above). Beating a higher quantity of top teams means more than beating a higher quantity of more average teams. But others may rank Ohio State's SoS better, and I could see where they're coming from.
If I personally ranked SoS among the 4 undefeated teams to this point, I'd go UGA (clearly #1), TCU/Ohio State, Michigan.
November 9th, 2022 at 3:39 PM ^
I reject the suggestion that it matters whether you're using the expectation of an elite team vs an average team. Statistically and directionally it doesn't matter except maybe at the extreme margins.
Even still, Missouri, South Carolina and Auburn are all in the top 55 of FPI, they're all top 53 in Sagarin, and all top 51 in FEI. And given that those teams along with Iowa, Wisky, MSU are all in the fat part of curve, there's not a huge difference in win expectancy against any of those teams for an average team.
November 9th, 2022 at 4:23 PM ^
But isn't the KEY question that rankings should answer is how to do the elite teams differentiate versus each other?
If there are a couple of undefeated teams and I'm comparing them, I absolutely want to know (and will value the team which is the first of these):
- Which of them played a schedule that included more teams from the right side of the tail (the harder to beat teams).
- Which of them played a schedule that included more teams from the middle of the distribution curve (the muddled middle).
November 10th, 2022 at 11:49 AM ^
Absolutely and that's essentially what ESPN's SoS does. And if the probabilities say it's more difficult, it's more difficult. It sounds like you're proposing an override to the probabilities to weigh the very tough games more than the probabilities already weigh those games?
If you use a methodology that says "how likely would it be for some baseline team to have a certain record against a certain schedule" which is the ESPN methodology (and seems like the best methodology IMO), I don't think it matters if the baseline team used in the analysis is elite or average. Either way should yield the same relative results, except on the extreme margins.
Regardless, it would never make sense to use an average team as your baseline so if that's what you're suggesting, I agree. When you're comparing the 3rd, 4th, 5th, 6th best teams, you're talking about very good to elite teams. ESPN uses the "average top 25 team" which isn't quite that good, but it's a very good team and almost certainly yields the same results directionally.
November 9th, 2022 at 12:39 PM ^
There's no getting around Michigan's awful schedule right now. It's not just OOC. The Big Ten is down and Iowa, MSU, others have not helped M's SoS. Michigan has played only one ranked team and only three with a winning record. That doesn't compare favorably.
I don't think this gets talked about enough. Just like this is M's easiest schedule in decades, this is also probably the weakest the B1G has been in decades. You have OSU and M as clearly top five teams, PSU is probably a top 10-20 team (difficult to know because outside of M and OSU, they haven't played anyone better than middling), and then a whole pile of mediocrity (or worse).
If you're taking a wholistic view of conference strength that isn't overly influenced by the best and worst teams (something like looking at the middle 50%), you could argue that the B1G is fighting with the ACC for weakest P5 conference...
November 10th, 2022 at 12:27 PM ^
Exactly. That said, Michigan's SoS will improve dramatically with the final three games, one against ranked Illinois and one on the road against OSU.
Michigan's remaining schedule is the 5th toughest in the country. So Michigan's SoS will improve but won't be ahead of any team that we might be compared to with one loss except maybe Clemson.
November 9th, 2022 at 12:06 PM ^
Quit trying to use logic and reason when discussing a beauty pagent.
November 9th, 2022 at 12:15 PM ^
Stig,
Let's just skip the interview process, the evening gown, etc. - and move right to the politically incorrect Swimsuit Competition.
November 9th, 2022 at 12:25 PM ^
Is this where the Song Girls come in, because they're the only reason I can see USC in the top 10...
November 9th, 2022 at 12:39 PM ^
Va,
Those USC Song Girls might be better at defense than the student athletes wearing helmets on Saturdays...
November 9th, 2022 at 5:36 PM ^
Once we get them to learn the difference between USC and the other team, sure!
November 9th, 2022 at 2:35 PM ^
While we're at it, if we're going politically incorrect, why don't we ditch the final Q&A ("I'm hoping for world peace") with a pole dance competition.
November 9th, 2022 at 5:38 PM ^
You convinced me. Ha!
November 9th, 2022 at 12:09 PM ^
Right now its about TV ratings, there is absolutely no reason USC should be higher than UCLA. Bama isnt a top 10 right now, and LSU has 2 losses...