2023–24 College Football Championship Predictions and Bowl Recap
Playoff Expansion Can’t Come Soon Enough
Throughout the season, each team’s performance metrics were tracked by objective formulas (No inclusion of pre-season ranks, opinion polls, scouting grades, or any other human-fabricated numbers) and ranked only by what has happened on the field. For more about tracking performance objectively and the terminology referenced here, please refer to the season kickoff. Or check out how the season progressed with these quick, simple links.
Well, that didn’t go as planned. For the first time in 5 years of tracking performance data using PASS, sTem and tWin% (See links above), the predictive results of bowl games were wildly off the mark. Time to blow it all up? Not quite. There’s always insight to be gleaned from the game data collected — More on those outcomes below. There’s no question that bowl games bring in the cash; there’s also still a lot of reflection for the NCAA to do on what postseason play looks like for any of this to feel meaningful.
The transfer portal has been around a few years now and the NFL Draft even longer. It still seems that more and more players opt out of playing in bowl games with each passing year. Depending on the source, it looks like there were about 853 total players who opted out of bowl games this year, which boils down to nearly 10 scholarship players per team. With 85 scholarship players on each team, roughly 12% of them were absent, which leads to a pretty large margin of error. That margin compounds when looking at a whether a player is a starter, their position, and how big of an impact they may have had.
It’s widely assumed and accepted that a change at starting quarterback makes the largest impact upon a roster. But what if multiple players at the same skill position are missing? North Carolina was one team that had the unfortunate experience of both, starting a backup QB with little experience and missing 6 WRs due to the portal, the draft, and injury. A dynamic passing attack behind experienced quarterback Drake Maye was the Tar Heels’ defining feature; UNC went into the Duke’s Mayo Bowl looking very much disarmed.
Coaching changes can also have huge impact. Losing a head coach can obviously change the morale and temperament of a team. Changes in offensive & defensive coordinator can also really affect a team’s identity, or in some cases make it evaporate entirely. Look at Penn State’s defensive performance against Mississippi in the Peach Bowl following the departure of Defensive Coordinator Manny Diaz. The Nittany Lions allowed 540 yards after allowing a season average of just 223 yards, including twice(!!!) holding their opponents under 100 total yards. No offense to Mississippi’s prowess, but that change gifted the Rebels a convincing win when they were likely to be flattened on a level playing field.
And that wasn’t even close to the most extreme change in the makeup of a coaching staff. James Madison’s 11-win team was so raided for talent that it had to hire 5 temporary coaches just to suit up for the Armed Forces Bowl. Air Force’s run-heavy attack was good this year, but not nearly as good as a threadbare Dukes team would have some believe. These are the kinds of discrepancies that season data unfortunately doesn’t have a tangible answer for.
Similarly, a lot of definition-of-average 6–6 teams from the Power 5 conferences looked better than their Group of Five foes simply by having deeper rosters. Season data shows that teams like Tulane, SMU and Bowling Green all had strong odds to dominate the likes of Virginia Tech, Boston College and Minnesota. Take away high-flying QBs like Michael Pratt or Preston Stone along with other star position players, and those G5 programs didn’t have the depth to replace them.
As we’ve seen throughout most of this season, Minnesota and Boston College were ranked #132 and 133 in PASS season performance (Dead last). While those teams managed to eke out the 6 wins necessary for bowl eligibility, they did so with less prowess than any other team playing their same opponents. This is a great illustration of how the Power Five’s true power isn’t always in the quality of starting players, but having generally deeper rosters to handle injury and upheaval. It’s an unfortunate conclusion to the year for Group of Five teams who conquered much greater heights in the regular season.
I think most can agree that the upcoming change to the playoff will be most welcome. The overwhelming majority of bowl games better resemble Spring scrimmages than the kind of postseason play seen in other collegiate sports. A coaching staff tests out a lot of underclassmen to see what they’ll be working with, and maybe throws out some trick plays while resting starters to avoid injury. The fan attendance numbers are often less than the average regular season game. It’s still football, it’s still fun, it still brings in massive media dollars, but hardly reflects regular season levels of competition.
If it were up to me and my passion for tracking objective performance numbers, I wouldn’t open the transfer portal until after the National Championship has been played. That would also apply for coaches, who would be held to the same standard as players and unable to leave their posts until after postseason play concluded. Just imagine if an NCAA basketball coach announced their departure for another job right before March Madness began! And while opt-outs to protect player futures remain a healthy and justifiable thing, incentivizing programs to put forward the best possible version of a team for bowl games should be strongly encouraged.
Bowl Outcomes
It’s not all bad! There’s no denying that going 17–24 in the 41 FBS bowl games was a failure (The FCS’ Celebration bowl excluded). That number really should have been 18–23, but I overlooked my own rule of coin-flip situations (Texas Tech vs. Cal), where the team with the higher sTem score for tempo should be chosen.
For all of the reasons described above, the field of data being evaluated shifted drastically from the expectations of years past. For the sake of normalizing this data set to check for ongoing viability, let’s remove the bowl games where the absence of a successful starting QB affected the outcome — 13 games. Let’s also remove the bowl games where coaches had already departed their posts — 7 more games. In the remaining 21 games the PASS predictions went 15–6 at a 71.42% rate, which is in line with the in-season predictive rate of 72.12%.
Best prediction
Michigan over Alabama
Score: 27–20 (OT) // Predicted score: 29–19
Off by just one total point with +2 to Michigan and -1 to Alabama. It’s amazing how well the simulator performs when both teams starters all play and the coaching staff remains intact. What’s even more impressive is that the yardage predictions were both within 10 yards, with Alabama projected at 283 (Gaining 288) and Michigan projected at 357 (351 gained). Despite Michigan’s absolute dominance this year, the majority of ESPN users still chose the hype of history past and took Alabama.
Worst Prediction
Ohio State over Missouri
The Cotton Bowl might not be the worst pick in terms of going against the grain, as a majority of bowl-pickers also chose Ohio State. But the simulator, along with most people thought this would be a higher flying score, at 28–19. That made for a game that was 30 under the projected score, the worst O/U of all projections. QB Kyle McCord’s departure was just one of 17 opt-outs for the Buckeyes and Mizzou took full advantage with many more starters on the field.
A special flop recognition also goes to Georgia’s utter dismantling of Florida State. The Seminoles were one of the 4 best regular season teams, even including the snaps they took after the injury to starting QB Jordan Travis. While 19 players opted out on both sides of this matchup, 12 of FSU’s looked to be starters based on snaps taken, while only 5 of Georgia’s were by the same measure. That was extremely noticeable on defense, where Florida State had previously been one of the nation’s best at stifling opponents.
National Championship Prediction
This is the same set of information provided as in the Bowl Game Guide, featuring the simulated final score, spread based on tempo, win likelihood, the team’s ranks for performance/tempo/strength of schedule, and a yardage prediction.
#1 Michigan — 34 (-8), 88%, #1 PASS / #1 sTem / #66 oSoS, 403yds
#5 Washington — 21 (+8), 12%, #5 PASS / #12 sTem / #48 oSoS, 341yds
In the coming weeks I’ll return with a recap of the season featuring final updated rankings, conference ratings, and final thoughts for the 2023–24 season. After the informational storm of bowl season it’s time for a bit of January rest. Thank you for reading, I hope your new year is already off to a wonderful start.
Author’s note: If you are gambling this season, remember to never gamble more than you can afford to lose. If you or someone you know faces issues with problem gambling, please seek resources that can help, such as https://800gambler.org/
