clock menu more-arrow no yes

Filed under:

Advanced Analytics: Early 2021 Season Projections

New, 21 comments

Will FSU make a bowl game this year?

COLLEGE FOOTBALL: SEP 03 Virginia Tech at Florida State

The spring season is in the rear view mirror for the Florida State Seminoles as they gear up for a summer offseason program. During the summer, teams can see incoming transfers and highschool recruits join the fray, or lose pieces to the portal or, god forbid, injuries. At this point in the offseason projections are still being tuned in order to best predict the 2021 season. Recently Draftkings Sportsbook released win totals for all 130 FBS teams. With those totals we can compare them against other projection models to give a rough first estimate of the 2021 season.

The two projection models we will be comparing today are ESPN’s Football Power Index (FPI) model, and my own personal ELO rating system. To give a quick summary of my model, an ELO rating system basically assigns teams a skill rating based on previous game outcomes and margin of victory (as well as a home field adjustment). Once you calculate those ratings you can determine the probability that one team will beat another. One drawback from my model is it does not currently include any sort of opponent adjustment or factor in returning production. Luckily for us, ESPN’s FPI factors in both of these features. For the purposes of this article, think of the difference in my model vs FPI as a sort of offseason rating.

For my model I acquired data from, courtesy of @CFB_data, using the @cfbfastR package

Due to some disastrous games in the 2020 season, my model does not think very highly of the Noles, giving them just a 15.5% chance to get to 6 wins and bowl eligibility. ESPN’s FPI however believes the Noles reach that mark 65.4% of the time, while Draftkings set the bar at 5.5 wins. An “Over” bet on 5.5 has implied odds of 47.62%, which means a little over half of the time FSU does not reach 5.5 wins. Big time additions in the transfer portal and a complete offseason should explain the difference between models. This should be interpreted as a team that isn’t yet elite, but is vastly improving.

Clemson is the consensus favorite across all projections, which shouldn’t surprise anyone. Draftkings appears to be higher on UNC and Miami than ELO or FPI is, and lower on NC State and Wake Forest. Rounding out the bottom we have Syracuse, UMass and FCS Jacksonville State. All three of those games have to be wins, plain and simple. Games with a wide range of outcomes include Wake Forest, Boston College, NC State and Louisville. In state rivals, Clemson and Notre Dame are low percentage games, but you can never say never.

Part of FPI’s methodology includes returning production, which looks at snaps and other statistics from players returning for the 2021 season. Here is that figure charted with offensive efficiency per play in the form of Expected Points Added (EPA). As you can see FSU returns a lot of production from an offense that finished a shade below average nationally. On the other hand you have teams like Florida who were extremely efficient last year, but return less than half of that production.

Moving over to the defensive side of the ball you can see where the strength of the Clemson team will be this year. They returned nearly all production, and were hyper efficient in terms of Expected Points Subtracted (Just EPA inverted). FSU returned over half of its production last year, which may not be a good thing considered how inefficient they were at stopping offenses. In order for FSU to meet or possibly exceed their win projections, it has to be with an improvement on this side of the ball.

The 2021 season can be a turning point for FSU if they meet or exceed expectations, which at this point are set at making a bowl game. FSU returns a lot of production on the offensive side of the ball, and adds dynamic talent in McKenzie Milton and Andrew Parchment. An elite offense can take pressure off of a defense that needs to show significant improvements. They may not be marquee matches, but the “toss up” games against Louisville, NC State, Boston College and Wake Forest are important to the overall health of the FSU football program.

If you liked this post and would like to see more graphs and stats, you can follow my CFB analytics twitter account at @CFBNumbers