Climbing The Ladder: An Early Look at Each CFB Team's Odds to Improve
Every team wants to succeed, which team's have the best chance to do so?
Every team regardless of their process has the same overarching goal for the offseason: work towards improving on the successes of last season. Success in football can take many forms, but for now we will focus on the most common form, more wins.
In order to determine which teams have the best chances to improve their win totals from the previous season, we will need some data to work with. The data we will be working with is as follows:
Last Year’s Win Percentage and Wins Over Expectation: This gives us the actual results of the games, as well as whether these wins had a little bit of luck baked into the results.
Last Year’s F+ Ratings: F+ Ratings are the combination of ESPN’s Bill Connelly’s SP+ Ratings, and Brian Fremeau's FEI ratings. These advanced metrics will give us an idea of how efficient a team was on offense, defense, and as a whole team.
A Teams 4 Year Recruiting Average: Teams that recruit the best players are typically the best teams, but can it help you improve your win total from one year to the next?
Returning Production: How much of your production from the previous year do you return for the following year. It is still early in the offseason, so there is still time for rosters to move around, but for now here is ESPN’s Bill Connelly’s latest installment of his returning production numbers:
Our data has been collected, and now it is time to build a model that will produce a probability a team improves their win percentage from the previous year. Our model type of choice will be a simple, but trusted logistic regression.
Feature Importance
When the testing model, it performed relatively well, accurately predicting whether a team will improve their win percentage year over year ~80% of the time. Considering wins and losses have an element of randomness to them, this will work. Popping the hood of the model, we can see that Returning Production is by far the most important variable to whether a team will improve their win total. Some make the mistake of attributing returning production to overall team strength, when in reality it is more a metric looking at a teams progression/regression from the previous season.
Previous years win percentage comes in second to returning production. This sounds about right, as it is much easier to improve on a 1 win campaign vs. a 11-1 season. Wins above expectation is also up there in feature importance, as teams with low wins above expectation were actually stronger than their record indicates (*cough* NEBRASKA *cough*).
At the tail end of importance are advanced metrics and recruiting. While these numbers give us a great idea of program strength and team quality, they aren’t as useful when look solely at whether a team will improve its win percentage.
Now that we have an understanding of the variables that are important for our model, we can look at what it thinks of each team entering the 2022 season.
2022 Odds to Improve Win Percentage
This is the end result. As mentioned before, when a team has a high win percentage the year before, their odds of bettering that are very low. This is illustrated by the downward slope on the graph above. Teams above the trend line are sitting in pretty good shape given their previous win percentage, where teams below the trend line are not in as great of shape.
It can be hard to look at all FBS teams at once, so we will start dividing them up into groups in order to see which teams are in for good, or possibly great seasons. First up, we can look at teams who were just below the bowl eligibility line to see which ones have a good chance of crossing that threshold.
Bowling In Your Sights
Both Louisville and Florida State arguably could have made bowls last season, and return a vast majority of their production. The model has both teams cruising to a bowl game this season. Of course, both of these teams have expectations that exceed bowl games, but for now just making a bowl feels like a solid floor. For reasons I will address at the end of the article, I would have to disagree with Bowling Greens place on this list. I believe their nation leading 92% returning production is dragging them higher than they probably should be on this list.
Outside of the top 3, you have some name brand teams like LSU, Texas and USC looking to get back on track. LSU is looking to usher in the Brian Kelly era, while Texas has loaded up on offense for Ohio State transfer Quinn Ewers. USC has been in the headlines for most of offseason with splash additions of HC Lincoln Riley, QB Caleb Williams, and countless more players from the transfer portal. All three of these teams should be playing postseason football, but it will be interesting to see how high each of them can climb in 2022.
Next Man Up
These are the teams that made bowls in 2021, and are looking to take another step forward in 2022. Tennessee returns 76% of their returning production including QB Hendon Hooker, who could climb the draft boards with a stellar season. Considering the issues off the field around the Tennessee program, Josh Heupel taking this team to 7-6 in 2021 was highly impressive. Year 2 of the Heupel high tempo “Everyone go deep every play” offense with Hendon Hooker… sign me up!
Similar things can be said about South Carolina and HC Shane Beamer. While they haven’t had to deal with NCAA violations like Tennessee, their place in the college football world is surrounded by football titans. 7-6 with a bowl victory against rival North Carolina is a heck of a first year for any South Carolina coach. They bring in QB Spencer Rattler, who is looking to prove his place as an elite QB in college football.
Miami, Penn State and Boise State have been successful in recent memory, but have taken a step back and are antsy to get back up the hill. Like USC, Miami has been in the headlines quite a bit this summer, as new HC Mario Cristobal looks to turn the recruiting tides back to what they once were in Coral Gables.
Big Season Potential
Last, but definitely not least, the teams looking to have a season to remember in 2022. Leading the way is NC State, who appear to be a popular pick for New York 6 or even ACC championship predictions. NC State always seems to have that “big game” potential, but it feels more tangible now than ever. They return 82% of their total production, which ranks 8th in the nation. QB Devin Leary, the 49th ranked QB based on offensive PFF grade, has the potential to soar up the draft boards with a heisman esc season. If that happens, we may finally see NC State appear in late season playoff graphics.
Utah is another trendy pick to potentially crash the College Football Playoff party. A healthy Cameron Rising makes this Utah offense dangerous, they just have to avoid an early season slip up in order to stay in playoff contention come November.
Limitations Turning Into Future Projects
I mentioned before that I wasn’t sold on Bowling Greens chances in 2022. The reason why is also a limitation to this model: Not taking into account a teams out of conference schedule. In 2022 Bowling Green plays Marshall, Mississippi State, and UCLA in their out of conference loop. While the model did do a good job overall, I would have liked to see what an out of conference variable would do to the predictions. I think it would be interesting to dive into the out of conference schedule and analyze how teams schedule out of conference and what that does to their win totals.
Conclusion
Winning more games than you did the year before is the goal of every single team every offseason. Using returning production is a great way to gauge which teams have a decent shot at increasing their win percentage that given year. Of course, things like out of conference schedule could also be a major factor, and will have to be examined more during this offseason. As we work our way through this offseason, we will put this together with an out of conference analysis in order to get the clearest picture we can get on the 2022 season.
If you want to dive in to the data like I do, check out @CFB_Data and @cfbfastR on Twitter, where you can learn how to get started in the world of College Football data analysis!
If you want to see more charts and one off analysis, follow my twitter page, @CFBNumbers
Cover Photo: William Howard/Icon Sportswire/Getty Images
At a quick glance there seems to be an over-representation of P5 teams above the line and under-representation below the line. Its notable there are only 2 Big Ten teams below the line which taking schedules into account doesn't seem entirely feasible, unless I am missing something?