Week 12 College Football Playoff Update

[Football Frontier RSS Feed]

I had hope having somewhat of a playoff in college football would encourage athletic directors to schedule tougher non-conference competition. A committee could select teams based on a complete package of results. A committee could reward teams for going up … Continue reading →

More...

[Swaggs]

I like these sort of metrics, but I think losses appear to be too devalued here. I think granting a 3-loss Mississippi an automatic pass is the most stark example, but a 2-loss Michigan over several 1-loss teams is pretty questionable, as well (for example, an undefeated or 1-loss Clemson, 1-loss Alabama, Oklahoma, and Notre Dame). Or is Michigan losing OOC to a 3-loss Utah and beating Oregon State and BYU make them 160+ points better than a Baylor team with one less loss, a cupcake OOC schedule, and a similar number of quality wins (assuming both win out).

Getting big wins is important, but so are double digit or multiple losses when you only play 5-7 quality teams in a season. Again, I like the idea of achievement scores, but I think there should be a heavily weighted counter for losses and bad losses if we are using it to sort playoff contenders. Just something to to potentially look at going forward.

[tarcone]

And there is the SEC bias in the metric.

[Galaril]

Originally Posted by tarcone

And there is the SEC bias in the metric.

Yes for sure.

[CU Tiger]

What is Bama's signature win?

[Solecismic]

I would advocate taking the schedule as a whole rather than insisting on random concepts like the opportunity for a victory over a team that may or may not deserve a specific rating.

Every "judged" sport invariably includes some sort of degree of difficulty. The pro sports can roughly equalize that through scheduling, but college sports cannot.

If you skate a clean routine, but only try double-jumps while your opponents are trying several triples, they can fall a couple of times and still be ahead of you.

My program doesn't have a bias for any conference. The pre-season rating, which is based on several factors (not the polls, though), is completely removed by week 8. The SEC is simply performing better than other conferences, so those week-in, week-out games are tougher. 8-3 in the SEC is probably better than 9-2 just about anywhere else.

The key philosophy here is trying to balance who has the best team with who has accomplished the most and who has a legitimate claim on the concept of championship. These are all issues that a committee can consider. I can provide various tools for examining these issues.

So it's just my opinion that the playoff should exclude teams that did not win their conference. From there, I use the metric as a test to see the rough accomplishment of the conference champion.

Last year, it all worked very neatly and there was relatively little controversy. This year, who knows? Something like a 10-3 Mississippi (well, Alabama is probably going to destroy Auburn) could cause a lot of frustration.

[Suicane75]

I'm intrigued by the idea of Clemson being stiffed if their only loss is to a 1 loss North Carolina team who's won 11 straight games. Don't see both teams getting in in that scenario, not even sure they'd both deserve to be in, but it intrigues me non the less.

[Thomkal]

Sorry wrong college football thread-moved to right place

[CU Tiger]

Originally Posted by Solecismic

My program doesn't have a bias for any conference. The pre-season rating, which is based on several factors (not the polls, though), is completely removed by week 8. The SEC is simply performing better than other conferences, so those week-in, week-out games are tougher. 8-3 in the SEC is probably better than 9-2 just about anywhere else.
.

This is where we differ. Strongly.
There are probably ...say 20 nonsec teams that would win the east this year.
Name why the SEC is "simply performing better"

It's the losses to Toledo, Memphis and the Citadel, right?

It's the last second screw jobto let UF beat FAU.

Here is a fun what if.
What if FSU smokes Florida who then beats Bama....

[Solecismic]

Yes, we differ strongly. No conference can pass the gotcha test.

The SEC is ranked first in just about every system in the Massey comparison. And the ACC is ranked fifth in most. Which is a big part of the reason why a decent, but not great, North Carolina team has a 10-1 record (losing its toughest non-conference game - to 3-8 South Carolina). UNC hasn't even had to play Clemson, FSU or Louisville.

But if your question is whether there's any kind of bias, most computer systems (including mine) do not factor in reputation or any kind of an eye test.

[tarcone]

yes, Iowa has just a couple triple jumps. And is clean. And Alabama has a fall on a double jump. And only a couple triple jumps. In your analogy, who scores higher?
I guess that would be Alabama.

[Solecismic]

If you want to play with the analogy, we could break it down into about 200 different kinds of jumps, each being a little more difficult. That's essentially what that particular metric does. And it's just one part of the overall picture.

In the meantime, you can reflect on just how much Iowa benefits from the Big Ten loading up the East and going to 14 teams. Drawing just Indiana and Maryland from the East (a combined 0-12 in the Big Ten until they played each other yesterday) was quite a break.

[Swaggs]

I think the tiers in this system "feel" a little closer to how I perceive the odds: The Spartans Shift Up: Updated College Football Playoff Predictions (Week 12) | FiveThirtyEight

I think the beef with the Frontier system is that the SEC and Big Ten champs are in, regardless of who wins it all. For the SEC, I think Alabama is the only team that should control its own destiny. Likewise the Big Ten, as Michigan State and Iowa control their own destiny and Ohio State probably does, as well (I can envision a situation similar to last season with Clemson and Bama in and 1-loss OSU, Oklahoma, and Notre Dame vying for two spots). I think it would be pretty appalling if a 3-loss Ole Miss were selected over a 1-loss Clemson, ND, or B12 or B10 team.

As an aside, I think it will be interesting to see how Oklahoma's loss to a .500ish Texas is perceived. I think it is a very similar blemish to Ohio State's home loss to a .500ish VPI last season. If Oklahoma misses out, despite a strong computer profile and a string of late season, high quality wins, the B12 will absolutely need to add a conference championship game.

[CU Tiger]

Originally Posted by Solecismic

Yes, we differ strongly. No conference can pass the gotcha test.

The SEC is ranked first in just about every system in the Massey comparison. And the ACC is ranked fifth in most. Which is a big part of the reason why a decent, but not great, North Carolina team has a 10-1 record (losing its toughest non-conference game - to 3-8 South Carolina). UNC hasn't even had to play Clemson, FSU or Louisville.

But if your question is whether there's any kind of bias, most computer systems (including mine) do not factor in reputation or any kind of an eye test.

Yep UNC lost to SC. A horrible, inexplicable loss. The worst loss OOC for the ACC this year. But UNC threw 3 endzone ints and lost by a score. Same logic used by Bama apologists. Plus it was the first game if the year, neutral site that was 80% SC fans blah, blah, blah.

Massey has repeatedly stated his system favors "traditional" teams (read no offense SEC teams).

This is a bad year to be making the SEC argument. Probably the worst year in 10 to be making that argument.

I'll ask again, can you name 1 quality win that Bama has?

And to clarify I've never argued the ACC is anywhere but 5th in the Power 5. But Clemson's 3rd string QB is the best QB in the SEC....

[Solecismic]

I'd say at Georgia and at Mississippi State - dominant performances on the road against solid teams. At home against LSU and the neutral-site win over Wisconsin. These are all quality parts of a resume.

I took a quick look back at the last bunch of seasons. I've had the SEC ranked first every year but 2011. The margin right now is on the higher end. I don't think they're having a bad year overall - just more cannibalization than usual.

I don't know about Massey's particular comments, but are we going to assume bias in every single algorithm out there? I guarantee you I do things differently from most, and I get the same results. It would help if you showed me Massey's exact quote.

[JonInMiddleGA]

Originally Posted by CU Tiger

But Clemson's 3rd string QB is the best QB in the SEC....

Umm, no.

I'm not entirely 100% convinced that your starter is even the best QB in the SEC. He's definitely in the conversation and in the group, but not an absolute. And I say that as someone who thinks Watson healthy and on a good day is one of the best in the country. But the guys at Arkansas & Miss State aren't slouches and would not be your #4 QB.

[digamma]

I think the issue is the SEC has been putrid out of conference this year, outside of the South Carolina upset over UNC and Alabama beating Wisconsin.

They've run up a great record over poor competition. I mean on Saturday, Florida and Georgia needed overtime to put away Florida Atlantic and Georgia Southern and the Citadel beat South Carolina. If those three results happen to Florida State, Georgia Tech and Clemson, people laugh at the ACC. And criticize them for scheduling them in November.

***And I agree with the Tiger...the ACC is a mess this year. The argument isn't about that.

[Solecismic]

No argument about Saturday. But other than that, the SEC has been the top conference. Obviously, if Mississippi ends up somehow winning the SEC, that's a political problem because there are three losses, including one at Memphis. I think this is an overreaction to the two overtime games.

Why would just about every computer ranking have the SEC on top if the non-conference results weren't solid? It's not even a close battle between first and second.

Why is it OK to schedule revenue-generators in September, but not in November? Ideally, the power conferences close ranks and all non-conference games are competitive. Though I suppose there would be a bidding war for scheduling Kansas.

With the opinion piece, I'm putting forth the idea that teams that don't win their conference should not be considered for the playoff. But that is just an opinion. Silver says the committee will consider non-winners, and Ohio State is actually in a excellent position to take advantage.

[CU Tiger]

Originally Posted by JonInMiddleGA

Umm, no.

I'm not entirely 100% convinced that your starter is even the best QB in the SEC. He's definitely in the conversation and in the group, but not an absolute. And I say that as someone who thinks Watson healthy and on a good day is one of the best in the country. But the guys at Arkansas & Miss State aren't slouches and would not be your #4 QB.

Jon, I'm talking about Chad Kelly.
He was the 3rd string Clemson QB before being ran out of town...

[CU Tiger]

Originally Posted by Solecismic

No argument about Saturday. But other than that, the SEC has been the top conference.

The #3 team in the MAC and the #7 team in the Big 12 BOTH beat Arkansas. This wasn't a one time fluke. From there Arkansas started in conference play and has racked up quite an impressive string of "Top 25 wins"

[Solecismic]

Yes, we differ strongly. No conference can pass the gotcha test.

The SEC is ranked first in just about every system in the Massey comparison. And the ACC is ranked fifth in most.

[tarcone]

Yes, I see no SEC bias in any of the arguments you have posted ( *eyeroll* ). Wow. You should sell this to ESecPN. With this and the FPI, the SEC would compare favorably to the NFL.

[Solecismic]

That was a persuasive argument. I appreciate the time you put into that analysis.

[CU Tiger]

As opposed to you just reposting the same line from last night?

Look you write a great video game that Ive got many hours of enjoyment from. But you condescending, omniscient and authoritative tone on a subjective conversation is tiring.

Hard to have a discussion when one side brings facts and the other restates the rankings of a meaningless poll.

[Solecismic]

You repeated your argument, I repeated mine in response. We disagree.

We do agree this is tiring.

[Swaggs]

I think a conference can be the strongest, but not have one of the four best teams (or seasons). That is the only problem that I have, due to this part of the article:

Summary

With the Big Ten and SEC winners a lock for the playoff, that leaves two spots. I would assign them in this order:

Clemson (at 13-0)
Oklahoma (at 11-1)
Stanford (at 11-2)/Notre Dame (at 11-1) winner
Clemson (at 12-1, as ACC champion)
Big XII winner (Oklahoma State (at 11-1) or Baylor (at 11-1))
North Carolina (at 12-1 or 11-2, as ACC champion)

The way this list is set up, it can’t be exhausted.

I think the SEC is still the strongest conference overall, but Alabama is the only slam dunk playoff team in my mind. If Florida loses to FSU and wins the SEC or Ole Miss wins the SEC, they should need a whole lot of help to be considered.

Ditto Michigan. They somehow have an .800 score with two pretty good wins (both at home against BYU and Northwestern -- two teams that are in the 20-45 range) and two losses to the two best teams they have played (@ Utah and @ Michigan State). Now, if they beat Ohio State and Iowa, they will have a strong resume, but what is the metric that makes them an .800 (with those wins), but makes Baylor a .638 (w/ one loss to Oklahoma and wins over Oklahoma State and TCU) or Oklahoma State a .678 (with one loss to Baylor and wins over Oklahoma and TCU)? Same with Mississippi and their .800 (with a win over Alabama and Mississippi State, but three losses).

I don't so much have a problem with those two teams being in the conversation, but when the metric puts Michigan and Mississippi closer to Alabama than to one-loss Baylor, UNC, or Oklahoma State, I think it needs examined. I don't' begrudge Jim and applaud him on the effort. The numbers are the numbers and I don't think he is shaping them to have bias, but I think losses should largely negate good wins and think that could be done more so here.

With all that said, I will be shocked if the SEC or Big Ten are ever left out of the playoff.

*all those metrics and scenarios are considered best case, with the teams respectively winning out.

[digamma]

OK, found the data I was looking for.

The SEC is 38-5 against non-Power 5 teams this season.

They are 6-2 vs. Power 5 plus Independents.

The Big 12 also has only played 8 games against the other Power 5 teams plus Independents.

The Pac 10 has played 10.

ACC (helped by the Notre Dame relationship) has played 19.

The Big 10 has played 21.

The ACC and SEC numbers will go up this weekend as there are four games between the two conferences.

I'm not arguing with what the computers spit out, but it's clear that some conferences are better at figuring out the scheduling mechanics (and have been for some time) than others. 29-2 vs. the Sun Belt, Conference USA and FCS without playing much of anyone else gets the numbers spinning.

[Solecismic]

Originally Posted by Swaggs

I think a conference can be the strongest, but not have one of the four best teams (or seasons). That is the only problem that I have, due to this part of the article:



I think the SEC is still the strongest conference overall, but Alabama is the only slam dunk playoff team in my mind. If Florida loses to FSU and wins the SEC or Ole Miss wins the SEC, they should need a whole lot of help to be considered.

Ditto Michigan. They somehow have an .800 score with two pretty good wins (both at home against BYU and Northwestern -- two teams that are in the 20-45 range) and two losses to the two best teams they have played (@ Utah and @ Michigan State). Now, if they beat Ohio State and Iowa, they will have a strong resume, but what is the metric that makes them an .800 (with those wins), but makes Baylor a .638 (w/ one loss to Oklahoma and wins over Oklahoma State and TCU) or Oklahoma State a .678 (with one loss to Baylor and wins over Oklahoma and TCU)? Same with Mississippi and their .800 (with a win over Alabama and Mississippi State, but three losses).

I don't so much have a problem with those two teams being in the conversation, but when the metric puts Michigan and Mississippi closer to Alabama than to one-loss Baylor, UNC, or Oklahoma State, I think it needs examined. I don't' begrudge Jim and applaud him on the effort. The numbers are the numbers and I don't think he is shaping them to have bias, but I think losses should largely negate good wins and think that could be done more so here.

With all that said, I will be shocked if the SEC or Big Ten are ever left out of the playoff.

*all those metrics and scenarios are considered best case, with the teams respectively winning out.

I think the biggest question about a playoff of this type is whether it should include the top four teams, regardless of conference. Or if it should be a continuation of the conference championship process.

Last year, the committee got lucky and didn't have to go into that philosophy. That may well happen again this season with Alabama/Florida, Iowa/Michigan State, Oklahoma, Notre Dame, Clemson. If these teams win out, no one will have too serious of a complaint about their credentials. Except the supporters of the team left out if all five win out.

The metric in question has flaws. It's intended for use as one piece of a complex puzzle. Accumulated accomplishment is a very specific concept, and is not intended to answer the best team question. Instead, I use it to answer the question, "is the sum total of this schedule representative of a team that should have the opportunity to play for the national championship?"

From watching a lot of games and from analyzing a lot of results, my conclusion is that the eye test is not a good one. Average teams look flawless against weak competition. One matchup really changes the eye test. While scores are very useful and improve accuracy, you have to be careful not to rely on them too much as they are somewhat situational.

I've never been a big fan of the four-team playoff concept. Sometimes compromise makes a problem worse. The pure bowl system has its flaws - most notably with disputed champions. The 16- or 24-team playoff extends the season uncomfortably.

This hybrid will, soon enough, force the committee to decide between a very strong team that didn't win a conference and a two-loss team that did. Even now, the Big Ten could present difficult inclusion issues. This could happen (5-10% chance, most likely):

Michigan 11-2 (2-1 HtH in group)
Iowa 12-1 (0-1)
Michigan State 10-2 (2-0)
Ohio State 10-2 (0-2)

All four teams would be ranked maybe as tightly as 5-10. You could argue that none of the four deserve it, but picking between them will be difficult.

As long as college teams play widely variant schedules, there's a risk of serious controversy.

In the NFL, the schedule length and common games ensure a competitiveness that we can understand. So when 12 out of 32 teams reach the playoffs, we're used to a system where a wild-card or a 4-seed may actually win the Super Bowl.

In college, there's a lot more variance. Still, every ranking system has error. Massey computes a percentage of instances where a lower-ranked team has defeated a higher-ranked team. This ranges from 12% of all results to 19%. Mine comes in at 14%. The error percentage is much higher when only considering the top 25.

So, when I look at a schedule as a whole, I think it's very important to note the difference between, let's say, a victory over the 40th-best team and a victory over the 80th-best team. Your chance of an upset loss is much higher against the 40th-best team.

That's the long answer to the Baylor/Oklahoma State question. The numbers are objective, but the interpretation is not - and I can certainly understand where two people who have a solid knowledge of college football can have a very strong disagreement as to who deserves to be in the playoffs.

[BishopMVP]

Originally Posted by Solecismic

With the opinion piece, I'm putting forth the idea that teams that don't win their conference should not be considered for the playoff. But that is just an opinion. Silver says the committee will consider non-winners, and Ohio State is actually in a excellent position to take advantage.

So in a scenario like 2012 in the ACC you would support an 8-5 Georgia Tech team (if they had beaten FSU in the title game) over 11-1 Clemson (I'm giving them that OOC win vs South Carolina) that beat them easily head to head?

In a scenario like 2011, you wouldn't support an 11-1 Alabama team that only lost in OT to unquestioned #1 LSU for one of the other 3 playoff spots?

In a 3-way tie with 1 conference loss for the Big 12 crown, you think the committee should only consider whichever team officially "won" the conference, even if that team went 0-3 OOC and the "2nd place" team is sitting there at 11-1?

Sorry, but I can't see any reason to throw out 25-33% of the season when we're already dealing with such small sample sizes.

[digamma]

Originally Posted by digamma

OK, found the data I was looking for.

The SEC is 38-5 against non-Power 5 teams this season.

They are 6-2 vs. Power 5 plus Independents.

The Big 12 also has only played 8 games against the other Power 5 teams plus Independents.

The Pac 10 has played 10.

ACC (helped by the Notre Dame relationship) has played 19.

The Big 10 has played 21.

The ACC and SEC numbers will go up this weekend as there are four games between the two conferences.

I'm not arguing with what the computers spit out, but it's clear that some conferences are better at figuring out the scheduling mechanics (and have been for some time) than others. 29-2 vs. the Sun Belt, Conference USA and FCS without playing much of anyone else gets the numbers spinning.

Updating

SEC 45-10 overall, 7-5 vs. P5/Ind, 38-5 vs. non-P5
Big 10 43-13 overall, 12-9 vs. P5/Ind, 31-4 vs. non-P5
Pac 10 29-8, 6-5 vs. P5/Ind, 23-3 vs. non-P5
Big 12 23-7, 4-4 vs. P5/Ind, 19-3 vs. non-P5
ACC 37-19, 9-14 vs. P5/Ind, 28-5 vs. non-P5

[Atocep]

Originally Posted by digamma

The Big 12 also has only played 8 games against the other Power 5 teams plus Independents.

Largely because of Baylor, KState, and Oklahoma State's scheduling philosophies.

At least with Baylor's 2nd loss we don't have to deal with Briles accusing the Big 12 playoff rep of not understanding Texas well enough to get them into the playoff.

[Breeze]

Originally Posted by digamma

ACC 37-19, 9-14 vs. P5/Ind, 28-5 vs. non-P5

The 9-14 P5/Ind record is significantly impacted by the 1-5 record against Notre Dame...who is only 2 conference games away from having an ACC schedule. If you counted Notre Dame as an ACC school, I think the P5/Ind record becomes 10 - 10... and the non-P5 becomes 31 - 5...

just for discussion purposes...

[Swaggs]

I think it is also worth noting that the Pac 12 and Big 12 also play 9 conference games (versus 8 for the SEC, ACC, and Big Ten).

In the future, it may make sense to require teams to play 10 P5 games per season in order to have eligibility for the playoffs or higher quality bowls.