Jason O. Watson
So as some of you may remember, we introduced a pair of research and analytics interns back in August. What have they been up to? Well, Mike Nixon was hammering out a piece or two a week on Dana Holgorsen, Chip Kelly, etc., but for the most part they spent the first few months of their jobs charting football games. It takes forever when done manually (and we're working on some nice pieces of automation), but I'm happy to announce that we can now start playing with some of the work they did.
Over the course of 14 weeks and the bowl season, Mike and Chris charted a total of 109 games. Granted, that is only about 13 percent of the games from the 2012 season, but a) charting takes forever and b) this is a healthy, robust sample size. This is not a healthy, even, representative sample, nor was it intended to be. This was intended to be a series that both aided in writing about given college football teams during the season and drummed up interest for a vastly expanded charting project, including quite a few volunteers, for 2013.
In the end, here is the distribution of teams for which we charted games:
- 12 games: Alabama
- 10 games: LSU
- 9 games: Oregon
- 8 games: Stanford, West Virginia
- 7 games: Florida, Georgia, Texas A&M, USC, Washington State
- 6 games: Kansas State
- 5 games: Florida State, Mississippi State, Notre Dame
- 4 games: Arizona, Auburn, Oklahoma, Oregon State, South Carolina
- 3 games: Baylor, Clemson, Georgia Tech, Michigan State, Nebraska, Ohio State, Ole Miss, Tennessee, Texas, UCLA, Washington, Wisconsin
- 2 games: Arizona State, BYU, California, Fresno State, Louisville, Miami, Michigan, Nevada, Oklahoma State, San Diego State, Syracuse, TCU, Texas Tech, UL-Monroe, Utah
- 1 game: Arkansas, Arkansas State, Boise State, Bowling Green, Central Florida, Colorado, Duke, East Carolina, Eastern Washington, Kent State, Louisiana Tech, Marshall, Maryland, Missouri, Navy, North Texas, Northern Illinois, Northwestern, Ohio, Purdue, Rutgers, San Jose State, SMU, Toledo, Tulsa, UL-Lafayette, Utah State, Wake Forest, Western Kentucky
Again, this distribution came about mostly because of in-season storylines. We charted most Alabama and LSU games because they were considered serious title contenders (and their defenses are awesome). We charted a good portion of Oregon, West Virginia, Washington State and Kansas State games because of the Leach v. Holgorsen series (and shifts it took when it turned out Washington State wasn't necessarily worth charting every week). We charted game-of-the-week types, and we charted quite a few bowls.
So what does "charting" mean, exactly? In this case, it meant taking the play-by-play data from a given game, then going back and watching the game (twice, perhaps) to add extra detail to each play.
How far did a pass travel through the air; did a 30-yard completion come about through a long pass or through a screen catch-and-run?
How many players did the offense have in the backfield at the snap?
How many players were lined up wide? Shotgun, pistol, or behind center? How many offensive players were helping with pass protection?
Was it a Wildcat formation?
Did the defense blitz? Zone blitz?
If a quarterback ran with the ball, was it designed? Was it a scramble? Option?
If a defense got pressure on the quarterback, was it because of the coverage? A blown block?
If a pass misfired, why? Overthrown? Underthrown? Thrown out of bounds? Tipped at the line? Defensed?
Was there a broken tackle? Were there multiple broken tackles?
This is next-level data, and while, again, this isn't a perfectly even distribution of teams and games, we hope to be able to draw some general conclusions about the game of college football with this data and with this series of posts. We have a lot of games for teams coached by Air Raid disciples; what characteristics distinguish the Air Raid from your typical spread? How do great 3-4 defenses like Alabama's or Stanford's differ? Are there common themes to be found among big plays? What is the yardage value of a missed tackle? What tendencies can we derive from different formations (my first piece is going to be on the no-back set)? This is wonky, and hopefully really interesting, stuff. And if we can both teach each other about the game of football and drum up interest for a vastly more widespread charting project in the future, then the project will have succeeded. Enjoy.