Last season's SP+ projections were accurate enough: Five of the top six teams in the preseason projections made the College Football Playoff -- No. 3 Oregon went unbeaten in the regular season, while No. 2 Ohio State won the national title -- and No. 4 Alabama and No. 8 Ole Miss nearly did. Boise State was projected as the best Group of 5 team and backed that up, and playoff surprise SMU started out ranked higher in SP+ than in the polls. The spreadsheets didn't see Indiana coming any better than the humans, and SP+ was just as blindsided by Florida State's historic collapse, but all in all, it was a solid performance.
There were still quite some caveats when looking at the rankings in real time, though. Michigan was projected seventh overall but had to replace head coach Jim Harbaugh and some key assistants after 2023's national title run. Iowa's offense was projected 114th, but Kirk Ferentz had finally discarded his son as offensive coordinator, and it was possible the Hawkeyes might actually try to score points. Troy was projected a respectable 64th, nearly tops in the Sun Belt, despite losing a special head coach in Jon Sumrall.
Michigan finished the season beautifully but still ended up 8-5 and 26th overall. Iowa's offense indeed jumped to 69th. Troy collapsed to 4-8 and 96th.
SP+ projections take incoming and returning talent into account, but not incoming or returning coaches. We know this can have an effect, but I haven't found a way to incorporate coaching changes in a way that actually makes the projections better. Maybe that will remain the case, but we're going to give it another go in the next round of projections in May. And we're going to try it with both head coach and coordinator change data. The correlations are too strong to not make another attempt.
When you compare a team's (or offense's or defense's) recent performance to its long-term performance baseline, a coach or coordinator change can prompt quite a bit of regression toward the mean. When an upstart enjoys a stellar run but loses its head coach, odds are good that gravity will drag it down quickly. When a historically strong program plays poorly and fires its coach, a rebound of some sort will likely follow. Granted, these statements are all true even without coach or coordinator changes. But the correlations get much stronger when a change takes place.
Let's walk through the "recency vs. long-term performance" concept and what it might mean for certain teams in 2025. (For UNLV, it could mean very bad things.)
Head coaching changes: It turns out that there's a pretty strong, negative correlation between (a) the difference between a team's previous-year performance and its 20-year average, and (b) a proceeding change in SP+ ratings. For teams with new head coaches, that correlation is -0.559 over the past 15 years.
Translation: If you overachieved last year compared to your 20-year average, and you lost your head coach, you are pretty likely to regress the following year. And if you underachieved and changed coaches, you're likely to improve.
That's a pretty strong predictor. And it could say something revealing about the teams, offenses and defenses that were positive or negative outliers last season.
Jump to a section:
Head coach changes | New offensive coordinators | New defensive coordinators
