/cdn.vox-cdn.com/uploads/chorus_image/image/71019528/usa_today_17918379.0.jpg)
We’re 70 days from kicking off the season, and we’re 0 days from kicking off Advanced Stats Week here at FTRS. Today, we’re going to take a 30,000 foot look at what to expect from our Advanced Stats coverage for the 2022 season. It’s a great time to be a nerd in the world of college football, as we’re privy to more and more wonderful data to help make more sense of this sport we love. Akshay will help us dive deeper into the present and future of that world over the next few days.
Game Charting
Here at FTRS, we will continue to provide weekly Advanced Stats previews of the upcoming game, Advanced Stats reviews of the most recent game, and mid-season check-ins to help evaluate our progress from last year and throughout the upcoming season. This year, we should have some improved graphics and other visuals to help you make sense of what you’re seeing week by week.
The foundation of my Advanced Stats work on this site is charting every single play of every single Georgia Tech football game. Let me take you behind the curtain a little bit. On each play, you can see below what goes into the charting process on my spreadsheet:
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/22685350/Screen_Shot_2021_06_27_at_8.34.30_PM.png)
- yards gained
- play call type
- pass (in)completion
- air yards on passes thrown
- whether or not the defense created a havoc play or generated pressure
- the quarterback in the game
- the offensive player who finished the play with the ball
- whether an offensive lineman blew their block
- which defensive player was responsible for havoc or pressure
- which defender was in coverage on pass plays
Then, we get an output telling us whether the play was successful, the EPA for the play, the expected completion rate for pass plays, and the line yards on run plays.
Adding all of that up for the whole game, we’re able to get a very thorough understanding of how the game went and why it went like it did. So, what kind of information can you expect to glean from our Advanced Stats reviews and previews this year?
There are a few key metrics that have very good descriptive and prescriptive value: success rate, yards per play, and EPA/play.
If you need a refresher, success rate is the percentage of plays on which the offense gains 50% or more of needed yards on 1st down plays, 70% or more of needed yards on 2nd down, and 100% or more of needed yards on 3rd or 4th down. It is a fantastic benchmark for understanding how efficient an offense is at staying ahead of the chains and keeping the ball (or how effective a defense is at keeping an offense from doing the same).
Yards per play is self-explanatory. I will add two caveats here: I include scrambles and sacks under passing yards per play, so the numbers we report won’t match exactly what you see in the traditional box score. Additionally, for all of our stats, we don’t include plays that happen in garbage time (defined as one team having a lead of 44 or more in the 1st quarter, 38 or more in the 2nd, 28 or more in the 3rd, or 22 or more in the 4th) .
EPA/play means expected points added per play. Am I in better or worse shape in terms of my likelihood of scoring on this drive after the last play? This metric does a great job of combining the efficiency and explosiveness of an offense, and it gives context that can be missed by looking at yards per play. What do I mean? 4 yards is 4 yards. Except 4 yards on 3rd and 3 means a whole lot more than 4 yards on 3rd and 17 or 2nd and 20. EPA/play does justice to that by taking an average amount of expected points given our down and position on the field and then comparing that to the situation after the play has happened.
You can look at these three metrics to get a great top-line feel for how a game played out. Beyond these key metrics, we’ll be digging in to quite a few other areas to analyze GT’s performance this year.
Rushing stats
- line yards: This metric, pioneered by Football Outsiders, attempts to parse out the credit on run plays between the offensive line and the running back. The offensive line gets full credit for the first four yards of every run, 50% credit for yards between 5 and 10, and no credit for yards after that. It’s imperfect, but it helps us to see whether the offensive line is consistently opening holes, or if running backs are having to rely on the very occasional big play to build their stats.
- opportunity rate: Related to the line yards metric, the opportunity rate tells us what percentage of the time a called rushing play gains 4 or more yards (assuming 4 yards are available to be gained).
- highlight yards: On the other hand, the highlight yards are the yards attributed to the running back in the above scheme.
- power success rate: On third or fourth down with 2 yards or less to go, how often does the offense run for a first down or touchdown?
- stuff rate allowed: What percentage of called run plays are stopped for no gain or a loss?
- run rate on 2nd and long: This is an evaluative tool for offensive play calling. On 2nd and long (8 or more yards to go), called pass plays have a much higher average EPA/play.
Passing Stats:
- YPA: yards per called passing play, including sacks and scrambles
- pass % on 1st down: This is another descriptive tool for offensive play calling.
- Air yards/attempt and Air yards/completion: How far from the original line of scrimmage did the ball travel to the intended receiver on all pass attempts and on pass completions?
- CPOE: Using the above information and a data set adapted from Josh Hermsmeyer, we can calculate the average expected completion percentage for the quarterback based on how far down the field he is throwing the ball. Then, we can take the results of each pass play and calculate Completion Percentage Over Expectation, which is one of the premier metrics for assessing quarterback effectiveness.
- YAcatch: How many yards is the receiver gaining after the catch?
- FLOPS: Failures on the line or penalties. How many obvious failures did each offensive lineman have over the course of the game?
- Quarterback situational stats: EPA/play when facing pressure and when not, as well as CPOE on throws more than 10 yards and throws more than 20 yards
Defense:
- havoc rate: Havoc rate is a favorite around here. It is a rate stat telling us what percentage of the time the defense causes a tackle for loss, forced fumble, interception, and/or pass breakup.
- pressure rate: For both individual defenders and for the team, how often is pressure generated on opposing pass plays? This is more stable and more predictive than just looking at sack totals.
- CPOE allowed: For individual defenders in coverage and for the team, how does passing completion percentage compare to expectation based on average depth of target?
- run stuff rate: For individual defenders and the team, how often are opposing run plays stopped for no gain or a loss?
- YAcatch allowed: How many yards after the catch are individual defenders allowing when in primary coverage?
I’m excited to chart and to share finding throughout the season!
The Binion Index
Last season, we debuted a new rating system for all of FBS that performed quite well in its game predictions. We correctly picked the winning side of the point spread in about 55% of last season’s games, and I’m excited about the evaluation and tweaking of the model that I’ve been able to put in so far this offseason.
You can glean some offseason value from the model by checking out the win total predictions we put up here, and we will have our final pre-season ratings posted and available sometime in August.
Nerd Resources
Finally, I also want to alert you to a couple of other resources in case you’re interested in doing some more digging on your own.
There’s a great new resource available for college football fans that also debuted last year, in large part thanks to one of our own here at FTRS, Akshay Easwaran. You can glean tons of wisdom from a project that he’s played an integral part in at: https://gameonpaper.com/cfb/.
You can also access a tremendous amount of information at collegefootballdata.com, and if you’re so inclined, you can do some fantastically interesting work on that data in the R or Python softwares packages using the CFBFastR package.
It’s going to be a great year here at FTRS as we dive deep into the 2022 edition of the Yellow Jackets. We can’t wait to share it with you.
What do you want to see from our coverage this fall?
Loading comments...