Welcome to our daily college football newsletter. If you like what you read and want it delivered to your email inbox every morning, subscribe to our Substack.
**
If you’re new to The Barking Crow, we owe you a little Joe Stunardi backstory. If you’re familiar with modern college basketball, you may have some guesses, and those guesses are probably correct. Either way, indulge us for a minute. We’re establishing credibility here.
In 2017, when All Things NIT was born, something quickly became clear: If we were to be a proper NIT blog, we would need an in-house NIT bracketologist. Our intent was to be silly, and it was official All Things NIT policy that all bloggers on the site must be named Stu. With ESPN’s Joe Lunardi the preeminent NCAA Tournament bracketologist, a friend suggested a Studonym: Joe Stu-nardi. This innocent suggestion has led to a greater personal identification with the name “Joe” and a handful of college basketball beat writers learning, mid-interview, that Stunardi is not my real-world surname. Each has responded with some variation upon, “I really should have realized that.”
Ignoring work done simulating high school cross country meets (that’s a biographical step too far), NIT bracketology was my introduction into statistical modeling. I was tasked with bracketology, and a probability model seemed the best way to do it. When All Things NIT became The Barking Crow in 2019, we added College Football Playoff modeling to our offerings. Here we are.
Why do I share this backstory? Because this year, for the first time, bracketology (Lunardi’s invention) will be part of the college football mainstream. You’re going to see a lot of bracketologies during broadcasts and on social media. Most making those brackets will be new to the practice. We at The Barking Crow are not. We’ve been doing it for years, in football as well as basketball. Issues new to new bracketologists will not be new to us. And that, at long last, brings us to the important warning at the heart of this whole post: There are two kinds of bracketologies out there. It helps to know the difference.
All bracketology is either Predictive or Reflective, or an unfortunate mix of the two. Predictive bracketology predicts what the eventual bracket will look like in a given sport. Reflective bracketology reflects what it looks like right now.
Like Methodism vs. Presbyterianism, the Predictive vs. Reflective debate centers on cosmotheological differences of fundamental import to those in the trenches of the conflict. Like Methodism vs. Presbyterianism, almost everyone on the sidelines does not give a shit about this disagreement. Should you? I don’t know. I can’t speak to Calvin, but I do think it’ll make your football season less confusing if you can put some context around the brackets bombarding you every time you glance towards ESPN.
Reflective bracketology is like a scoreboard. It shows you where things stand right now. Predictive bracketology is a view into the future. It shows you the likeliest place things will land. In basketball, the industry standard is Reflective bracketology. This is what Joe Lunardi has always done, and since we all copied Joe Lunardi, most of us follow his lead. At The Barking Crow, we differ. We don’t engage in Reflective bracketology. It’s not that we think there’s anything wrong with it come February and March. It’s that in November, December, and January, Reflective bracketology is impossible. The season hasn’t progressed far enough for a bracket to make sense.
An example:
Automatic bids in basketball go to the 31 conference tournament champions. Conference tournaments don’t begin until the waning days of winter. Lunardi—a hero of ours—squares this circle in November by including the conference tournament favorites in these slots. In January, he squares it by including the leader in the conference standings. In effect, then, his November bracketology is Predictive and his January bracketology is Reflective, except that January’s reflects the standings, not the tournament outcome. December? It exists in chaotic limbo, mostly Predictive but partly unrealistically Reflective.
If this sounds confusing, it’s because it is. The college basketball bracketology industry standard is to be terribly confusing for the first three months of the season. If you know what you’re looking at, it all makes sense. If you don’t want to have to learn about bracketology industry standards, it does not make sense. Like most proposed federal economic policy, the process is coherent, but it’s stupid. It’s not Lunardi’s fault. His approach does make sense. But he was tasked by the market with taking a useful tool in February and March and expanding it backwards into January, December, and eventually all the way to April, when his predictions now launch for the following year’s tournament. This expansion necessitated the patchwork process of processes. The result? A mass befuddlement of innocents.
We, then, maintain Predictive bracketology the whole season long. With the exception of a few days in March when a live scoreboard is helpful, we think this is what fans really want anyway. They want to know where things are headed. Predictive bracketology runs into its own issues, but it’s a more consistent approach, and it’s more aligned with whatever real-world situation is about to unfold.
**
Enjoying what you’re reading? Subscribe to our Substack to receive it every day in your email inbox.
**
In college football, Predictive bracketology makes even more sense than it does in basketball. The College Football Playoff rules center on conference championships. Conference championships aren’t played until the final two days before the bracket is unveiled. The CFP field will be determined by who the conference champions are, not by who leads the conference standings right now. Not until conference championship matchups are determined is Reflective CFP bracketology helpful to the casual fan.
I think most college football bracketologists are doing Predictive bracketology, at least for a few more weeks. The temptation to slide into Reflective status, however, will probably grab them in early October. They’ll give a 4-seed to some 5–1 Big 12 team who’s 3–0 in conference play and destined to finish the year unranked. After that, it’ll be a hodgepodge until the Reflective and Predictive brackets finally converge upon the actual 12-team field.
Others have opted to begin the hodgepodge already. Here’s Andy Staples’s projection from Monday:
Andy Staples is one of the best college football analysts in the country. He knows the sport very, very well. He also understands the new playoff format. He’s one of many starting college football bracketology this year. He’s using the Lunardi method: Reflective when it’s anywhere close to an option, Predictive when there are gray areas to be filled. Where does this lead him? Georgia Tech is the ACC leader, so Georgia Tech’s in the playoff field. Nobody else in the picture has played a game, so Staples’s own predictions fill in the other eleven spots, plus his six spaces on the graphic for peripheral contenders.
Again, I really like Andy Staples’s work, and I recommend it outside of this. But this is unnecessary. He’s justified it by saying things along the lines of, “The playoff format is confusing, and we need to demonstrate how it works so fans get it come November.” Maybe that’s true. But I fail to see how brackets like this don’t just confuse people more.
How does the playoff format work? Like this:
- The five highest-ranked conference champions get bids. The four highest-ranked champions are seeded 1 through 4, in order of ranking.
- The seven highest-ranked teams who didn’t win a conference championship get bids. These, plus the fifth conference champion, are seeded 5 through 12, in order of ranking.
How does the playoff format not work? The twelve playoff bids don’t necessarily go to the twelve highest-ranked teams. The four first-round byes don’t necessarily go to the four highest-ranked teams. Lastly: More than five conference champions can make the playoff field, if there are six or more of them ranked in the top twelve. This might be one of the least understood pieces of the process, especially thanks to graphics like Staples’s labeling that 12-seed slot as specifically for the Group of Five.
Is it confusing? A little bit. It’s more complicated than the four-team process.
Is it worth including Georgia Tech this week, to make a point? No. That point does not need making, and including Georgia Tech doesn’t make it anyway.
Here’s our model’s projected bracket from before Week Zero, typed out (we’ll get the graphics made in future weeks, and they will not be as visually appealing as On3’s):
Orange Bowl Side:
First Round: 9-seed Texas at 8-seed Alabama
First Round: 12-seed James Madison at 5-seed Notre Dame
Sugar Bowl: 1-seed Georgia vs. Alabama/Texas winner
Fiesta Bowl: 4-seed Kansas State vs. ND/JMU winner
Cotton Bowl Side:
First Round: 10-seed Penn State at 7-seed Oregon
First Round: 11-seed Missouri at 6-seed Ohio State
Rose Bowl: 2-seed Michigan vs. Oregon/PSU winner
Peach Bowl: 3-seed Florida State vs. OSU/Missouri winner
Here it is ahead of Week 1, following Saturday’s games:
Orange Bowl Side:
First Round: 9-seed Texas at 8-seed Oregon
First Round: 12-seed UTSA at 5-seed Notre Dame
Sugar Bowl: 1-seed Georgia vs. Oregon/Texas winner
Fiesta Bowl: 4-seed Kansas State vs. ND/UTSA winner
Cotton Bowl Side:
First Round: 10-seed Penn State at 7-seed Alabama
First Round: 11-seed Missouri at 6-seed Ohio State
Rose Bowl: 2-seed Michigan vs. Alabama/PSU winner
Peach Bowl: 3-seed Clemson vs. OSU/Missouri winner
How did we build these? We took all nine conference favorites, according to our model’s simulations, and we compared their average final CFP rankings and their playoff probabilities. Then we did the same with the average final rankings of the rest of the FBS teams.
The four highest-ranked conference champions get the first four seeds, so we gave those to the conference favorites with the highest average rankings. Before Week Zero, Florida State was among those four. After Week Zero, Clemson passed them. Clemson has a better average final CFP ranking than Kansas State, so Clemson gets the 3-seed, like Florida State did.
There’s a fifth playoff spot for a conference champion, and to fill it, we looked at the other five conference favorites. Since all currently have an average final CFP ranking of “unranked,” we turned to their respective playoff probabilities. James Madison and UTSA are very, very close in that category. Close enough that little strength of schedule ripples from this weekend’s game were enough to flip them.
From there, we took every other team in the FBS and lined them up by average final CFP ranking. The seven highest-ranked teams get at-large bids. We assigned those in order. As with JMU and UTSA, Alabama and Oregon are very close in our model’s simulations. Close enough that their average final rankings flipped based on the Week Zero games.
To illustrate this process, here’s the spreadsheet we use to organize it:
We start with the conference favorites. Once we have our five playoff teams from that group, we line up the at-large candidates by average ranking. From there, it’s just slotting teams into their specific bowl games according to the CFP’s procedures. This is what we’ll do every week. We won’t insert any opinions. We won’t try to keep up with the standings. We’ll keep our eyes focused on where things are currently aimed, and we’ll share that with you. Every one of our bracketologies will be, more or less, our model’s single best prediction of the eventual playoff bracket.
Georgia Tech can get to the playoff. The ACC is an open race. But the Yellow Jackets are not there yet. Playoff spots go to conference champions. Not to whomever reaches 1–0 first.
**
We had a trivia question for you yesterday: Which three teams have appeared in each of the last 23 preseason AP polls?
If you guessed USC or LSU, you were very close, but recent history let you down. USC missed the top 25 ahead of the 2019 season. LSU missed it ahead of 2022. Those teams each have made 22 of the last 23, but not 23.
If you guessed Alabama, you may have forgotten how bad things got before they landed Nick Saban. From 2002 through 2007, they weren’t in a single preseason AP Top 25, and in both 2004 and 2006, they never cracked the Top 25 at all.
The answers? Ohio State, Oklahoma, and Georgia, in order of average preseason ranking. Georgia wasn’t great until lately. Oklahoma hasn’t been great lately. But they’ve both always been good enough to start the year in the field of national awareness. The last time Oklahoma missed a preseason poll was 1999. For Georgia, it was 2001. For the Buckeyes, you have to go all the way back to 1988.
**
Tomorrow, we’ll tell you something important about North Dakota State ahead of their Thursday night tilt with Colorado. That, and all the other early Week 1 action.
On Friday, we’ll have a big look at the Labor Day Weekend games.
**
Thanks for being here. If you like what you just read and want it delivered to your email inbox every morning, subscribe to our Substack.
