8 factors that predict potential underdog College Football Playoff contenders

New York Times

15 hours ago

The AP's preseason poll dropped on Monday afternoon, and tucked inside were a list of the most obvious contenders to make the 2025-26 College Football Playoff — names like Texas, Penn State, Clemson, Georgia and last year's title-game combatants, Ohio State and Notre Dame. The Athletic's own Matt Baker recently disproved the notion that ranked preseason teams have an edge because they're ranked; rather, these squads are ranked highly because they're also the most likely playoff frontrunners, with more talent than anybody else in a sport where that tends to matter a lot (and in an era when it matters more than ever).
Advertisement
Either way, those 25 teams aren't necessarily going to sneak up on anyone. But we're also interested in teams from outside the preseason ranking who might still crash the playoff party. So I combed back through data since 2000 (the earliest season of turnover-margin data in Sports-Reference/CFB's database) for indicators that might predict an unranked AP preseason team's ability to still finish within the top 12 of the final pre-bowl poll — a general proxy for being in the playoff mix going into the committee's final weekend of decision-making.
Many of these elements relate to the concept of regression toward the mean: the idea that teams tend to revert in the direction of longer-term norms over time, especially when we account for more volatile or luck-driven stats. It's useful to be able to create a hierarchy of those types of factors to look at — or regress away — when trying to identify playoff-worthy dark horses.
With that in mind, let's run down the key questions to ask around any potential surprise contender — and the potential beneficiaries of those variables in 2025 — in order of predictive importance.
This factor is not necessarily a surprising consideration — good teams tend to stay good going forward, especially in college football — but it's the most important metric for determining whether a team from outside the preseason top 25 will break through in spite of the pollsters' concerns. Of the 66 teams in our dataset that made the leap from outside the preseason poll, more than half (35) were coming off a season with a +5.0 Simple Rating System (SRS) score, which puts a team as roughly a top-50 team in the previous year's pecking order, and more had a previous rating of +10.0 or better (12) than were negative the previous season (nine). It's not impossible for a surprise team to really surprise, rising from a poor previous rating to crack the top 12, but it's rare.
More often, those teams come from the top ranks of the prior year, which means BYU and Louisville get the most credit in this category, followed by USC, Iowa, Minnesota and Virginia Tech.
Many of these teams suffered key personnel losses to some degree or another, so caution is advised, particularly before we get to the other indicators. Situations like Florida State last year — whose SRS declined by a shocking 21.7 points year-over-year — can happen. But as a baseline, the clear majority of teams stick within a touchdown of their previous SRS, for better or worse.
Advertisement
The second-most important predictor also involves SRS — but it concerns whether a school was abnormally good (or bad) by its own standards a year ago. In the spirit of regression to the mean, if a team has established a particular long-term level of play, then deviates from that level in one season, it is likely to return or at least move back to the long-term norm the following year.
That means we're looking for squads whose 2024 ratings were far worse than usual. Newer FBS teams like James Madison (and Delaware, Missouri State, etc.) were excluded from consideration because they lacked a five-year sample of previous seasons.
Tulsa and Kent State top the list, but are usually just mediocre and were among the worst teams in FBS last season. They are good candidates to improve, but not to make the CFP. More realistically, Air Force, Oklahoma State and Florida State all have the potential to improve quite a bit through a combination of regression, the portal and recruiting (in the case of OSU and FSU), and the other built-in ways that prevent competitive programs from staying down for too long.
Much to the chagrin of mid-major fans across the nation, the college football power structure favors elite teams in elite conferences whenever it can. So it makes sense that a potential dark-horse candidate from a more prestigious conference is going to get the benefit of the doubt in the rankings before a team trying to make their bid from a smaller conference.
In terms of our 2025 candidates in the categories we've looked at so far, that big-conference bias is good news for Iowa — who has the best combination of 2024 SRS and the potential to regress toward an even higher rating based on its previous norms — plus Louisville, USC, BYU, Minnesota, Missouri, Wisconsin and the rest of the power-conference teams we've discussed. Meanwhile, the American and, to a lesser extent, Sun Belt are the epicenter of teams that have solid indicators, but their conference might get held against them, barring an automatic bid.
Advertisement
Though it sometimes gets mentioned in betting circles, one of my favorite semi-hidden stats remains 'yards per point' (YPP). At its core, YPP measures how efficiently a team converts yardage into points on offense — and how inefficiently it forces opponents to do the same on defense. Teams with a lower YPP than their opponents are essentially turning field position into points at a better rate, which translates into a higher scoring margin and, usually, more wins. Like turnover margin (more on that stat later), it can be volatile from year to year, bouncing around with red-zone execution, third-down conversion rates, special teams performance and 'bend-but-don't-break' defense. But when a team consistently posts a strong YPP differential, like the New England Patriots did throughout their dynasty, it's often a sign of something more sustainable.
In our historical sample, teams that had lower YPP differentials between offense and defense in the previous season — negative numbers are better because they mean you traded fewer yards for the same points than your opponent did — tended to be more likely to make the leap from unranked to the top 12 after controlling for everything else. Our group of historical surprise teams had an average prior YPP differential of -1.3 and 62 percent of them were negative in the previous season.
That's a bit counter to the usual wisdom that YPP is heavily luck-based and likely to regress against teams who relied on it. In this case, it may be picking up on a team's strengths that don't always show up in other conventional stats, like situational execution, special teams efficiency and a general ability to squeeze more points out of opportunities. If that holds true, Army and Iowa stand out among the unranked masses for their efficiency on offense and ability to bend, not break, on defense.
In a less contrarian statistical finding, we won't be surprised to learn that teams that won fewer games than we'd expect based on their points scored and allowed — via the Pythagorean Formula — tend to be more likely to bounce back the following season. More importantly for this exercise, they are also prone to being overlooked by the preseason pollsters, who may tend to judge a team off of its standard win-loss record without digging into extenuating stats like its point differential.
That could give a boost to teams like UCF and especially Auburn, plus Virginia Tech slightly lower down the list, each of which was unlucky with close games compared to blowouts in 2024.
Like with Pythagorean luck, another well-known factor that regresses to the mean for football teams over time tends to be turnovers. Yes, there are some notable exceptions — we go back to Bill Belichick's Patriots for an example from the NFL — but generally speaking, a team with an outlier turnover margin in either direction has a tendency to move toward the middle the following season.
That means for breakthrough candidates, we're looking for teams coming off poor turnover differentials in 2024.
Most notable here is Florida State, whose uncharacteristically awful 2024 was driven at least some by that minus-16 turnover margin. Plus, lurking just at the periphery of this list are Auburn, UCF, Oklahoma State, Arkansas and West Virginia — all were at minus-eight or worse and have a number of additional factors pointing to better days in 2025.
Advertisement
One of the interesting philosophical arguments around preseason polls is what exactly they're attempting to rank. Most view them as an accounting of top-to-bottom roster talent on paper going into the season, which incidentally is why their basketball cousin is a good predictor of March Madness results even after controlling for season-long performance. However, they're not necessarily trying to peg where a team will end up by season's end.
In theory, the two rankings ought to be correlated, but differences come about because of scheduling, which is an area we can project in our search for surprise playoff contenders. In our historical dataset, teams that went from a harder schedule (in terms of opponent SRS ratings) during one season to an easier projected schedule (based on a weighted multi-year average of SRS for its opponents) in the following preseason tended to be more likely to catch the initial polls by surprise.
Factoring in schedules, teams like Virginia Tech, USC, FSU, Kansas and Washington (among others) are being underrated in their playoff potential.
This might be the most fascinating result in our entire experiment. In the historical data, there was a real effect where unranked teams whose coaches were in either their third or fourth season at a particular stop (consecutively, not overall) were more likely to finish among the top 12 in the pre-bowl rankings even after controlling for everything else.
Why might this be? My theory is that those seasons come in the sweet spot of a coach's tenure. By that point, they've had enough time to install their systems, bring in players who fit their style and establish the culture they want — but at the same time, their message is still fresh, morale is still high and opponents may not have fully adapted to their tendencies yet.
Among our common dark-horse candidates from the rest of the factors above, teams with coaches in the third or fourth year with their current program include USC's Lincoln Riley, Jeff Brohm with Louisville, Hugh Freeze with Auburn, Brent Pry with Virginia Tech and Matt Rhule with Nebraska. This is the least important factor. It's not impossible for a newer — or older, in the case of Kirk Ferentz in his 27th year at Iowa — coach to also surprise from unranked territory, but there does seem to be something about that sweet spot that makes a team more primed to break out.
Now, we combine all of the predictive factors from above into a single ranking, weighted by the importance of each sub-category:
Based on past trends, we would expect USC to be most likely to rise from outside the top 25 to serious playoff contention by the end of the regular season, followed by Louisville, Auburn, Virginia Tech and a bunch of Big Ten and Big 12 schools.
Not all of these teams jumped off the page in every category, but most were coming off a decent SRS season and/or had a subpar season by their standards. They're all in power conferences. The highest-ranked non-power candidate was Tulane at No. 27.
Advertisement
The odds are that at least a few of these teams — or teams like them a bit further down the ranking — will make the jump into the playoff conversation, as an average of 2.75 teams per year started unranked and finished 12th or higher in the pre-bowl AP poll each year in our sample. We've trended toward even more of that chaos in recent years; four teams have done that per season since 2021.
So if you're thinking about this week's polls, remember that while the AP's preseason top 25 is stacked with the sport's most obvious playoff threats, history says the field won't be limited to just those hyped-up names. Every season brings a few gate-crashers from outside the group of teams we thought we knew to watch, and our list is full of schools that check the right boxes to be those spoilers in 2025.
(Photo of Lincoln Riley: Kirby Lee / Imagn Images)
Spot the pattern. Connect the terms
Find the hidden link between sports terms
Play today's puzzle