Analytics have become commonplace in sports. Casual baseball fans know about Moneyball and the value of on-base percentage over batting average and have at least heard of WAR (Wins Above Replacement), even if they don’t really know what it’s good for. Casual basketball fans were probably exposed to John Hollinger’s PER (Player Efficiency Rating) system at some point, know about Houston general manager Daryl Morey’s background, and may have even picked up on stylistic shifts that were impacted by analytics — more corner 3s, a shunning of long 2-pointers, et cetera.
The analytics revolution has not quite reached mainstream status with the country’s most popular sport, however. Football is a more random, complicated sport, with a pointy ball and 22 players who all carry out unique tasks at one time or another. There is more luck and specialization involved in football than in most team sports, and it makes it more difficult to draw obvious conclusions about players, teams or front offices.
Still, at the NFL level, there has been progress. A lot of teams have analytics departments, and sites like Football Outsiders, Advanced NFL Stats and others have been gaining a foothold. Any breakthroughs for football analytics, however, have taken place at the professional level. In college football, where the head coach is the general manager and some graduate assistant is the analytics department, things are a little trickier. Some teams and coaches have a much better feel for when to go for it on fourth down, but that’s only one aspect of stats in football.
But the college football statistics community does exist and has been putting out some interesting work for a while. We are on an everlasting quest for more hands on deck, but we get a little further, a little more detailed and a little more engrained with each passing season.
At this stage in the game, what you need to know about college football analytics can be more properly explained by certain truisms instead of specific measures. Here are five points that you need to know about college football and its stats.
Visit Football Study Hall to get in-depth team previews, advanced stats and features for the 2014 season .
1. There are Five Factors to Winning a Football Game
So much of football boils down to where you start, how you move the ball, how you finish, and whether or not the pointy ball bounces in your direction. Or to put it another way: The five stats that matter most in football are efficiency, explosiveness, field position, finishing drives and turnovers.
• If you win the field position battle, you win the game 72 percent of the time.
• If you win the turnover battle, you win the game 73 percent of the time.
• If you finish drives better than your opponent, you win 75 percent of the time.
• If you are more efficient than your opponent, you win 83 percent of the time.
• If you are more explosive than your opponent, you win 86 percent of the time.
The college football box score hasn’t changed much since the 1920s, and if we were to rebuild it from scratch, we would be well served to build it around these five factors. These concepts are in no way advanced stats, but they are and could be the basis for such.
2. The Difference Between Standard Downs and Passing Downs is the Difference Between Winning and Losing
You can define standard downs as first down, second-and-7 or fewer, or third- or fourth-and-4 or fewer. Passing downs are the other plays: second-and-8 or more, third- or fourth-and-5 or more.
We’ve all heard coaches preach the importance of staying on schedule. It is a cliché, but sometimes clichés exist for a reason. A team’s Success Rate on standard downs was, on average, 48 percent in 2013; on passing downs, it was 32 percent. Once you fall behind schedule, it is rather difficult to catch up.
Interested in learning more about advanced stats? Check out Bill Connelly's book for an in-depth look at all things college football, the issues facing the sport in future seasons and a detailed breakdown of advanced statistics and what they mean.
3. In Advanced Stats, Adjusting for Opponents is Everything
One of the biggest problems with college football stats is that you cannot simply look at them and come to immediate conclusions. Fresno State averaged more yards per game than Texas A&M in 2013, and Northern Illinois averaged more than Ohio State. Marshall and Rice won 10 games while Washington won only nine. We know to pause and ask, “Yeah, but who have they played?”
Statistically speaking, there are countless ways to adjust for the quality of the opponent at hand (some better than others), but no matter how you do it, you have to do it. In essence, it is what makes “advanced stats” advanced, and while we account for this in every sport, it is never more vital than in college sports. The talent gap from team to team is just too wide.
4. Garbage-Time Stats are Mostly Garbage
One of the least productive moments of the BCS era (1998-2013) came when decision-makers decided margin of victory should play no role in the BCS formulas. They didn’t want to encourage teams to run up the score against overwhelmed opponents. (This ignores that human pollsters are still very much swayed by big margins.) If Team A beat Team B by one point, it was the same as beating that team by 38. It intentionally removed the most telling piece of data for systems that use only points scored and allowed.
There’s a better way, anyway. Play-by-play and full-drive college football data can be found publicly now, either at the NCAA’s official site, on school sites or at CFBstats.com. And when you use data beyond simple points scored or total yards gained, you can filter out what happens in garbage time, when the game is out of reach. You can look only at what transpired when a game was considered competitive, which retains the important piece that we gleaned from point differential (level of dominance) while removing the part nobody likes (running up the score).
5. Pace Adjustments are Almost as Essential
If Florida State’s 2013 offense had run at Baylor’s pace, the Seminoles would have projected to average 633 yards and 63 points per game. If Georgia’s 2012 offense had played at Oregon’s pace, the Bulldogs might have averaged 575 yards and 47 points per game.
We get distracted by big, shiny point and yardage totals, and we sometimes fail to recognize the offenses or defenses that are truly the strongest (or weakest). If you play in the Big 12 or Pac-12, your defense is going to face a ton of high-paced, high-quality offenses and will by default give up more points and yards. That doesn’t mean the defenses in those conferences stink any more than it means that ACC or Big Ten defenses are better because they face fewer plays. If advanced stats aren’t your thing, you could still do yourself a huge service by looking at yards per play in the box score instead of total yards.
Measuring The Five Factors:
In this article and in the 2014 Athlon Sports’ College Football Preview, you’ll find a series of references to what we call the Five Factors. They are interrelated and are more descriptive than prescriptive — you can’t simply say, “We need to improve on turnovers” and make it so — but they are wonderfully useful in examining what went right or wrong for a team in the previous season.
So what’s the best way to look at these factors? Some are more simple and direct than others.
Field Position: Simply looking at a team’s (and its opponent’s) average starting field position is a clean way of determining how a team leveraged the field in its favor. A team can create an advantage (or disadvantage) through numerous means — good kicking or punting, good returns, turnovers, avoiding three-and-outs (and creating plenty for the opponent) — but average starting field position is the easiest way to measure the result.
Finishing Drives: In the 2014 Athlon Sports’ College Football Preview, we took a look at one simple measure to judge the ability to finish drives: points per trip inside the opponent’s 40. There is more separation between good and bad teams if you stretch the “scoring opportunity” definition to the 40, but you can get a good feel for drive-finishing ability by looking at the typical red-zone definition, too. WARNING: Avoid “red zone scoring percentage” averages. Over time, there is an enormous difference between scoring a touchdown and settling for a field goal, and “scoring percentage” treats them the same. Aim more toward touchdown percentages or the superior “points per trip.”
Visit Football Study Hall to get in-depth team previews, advanced stats and features for the 2014 season .
Turnovers: You can certainly look at turnover margin to roughly gauge the impact of turnovers. You can also look into the field position and points immediately created by those turnovers if you want to get a little bit more accurate.
To aim at both the effect and randomness of turnovers, you will find in the 2014 Athlon Sports’ College Football Preview a comparison of actual numbers and “projected” turnovers based on what a team’s turnover margin would have been with an average number of fumble recoveries (50 percent on average, obviously) and interceptions (a team normally averages one interception for every four pass break-ups). That will give you an idea for both who committed and forced the most and who was particularly lucky or unlucky regarding the bouncing of the pointy ball.
But what about efficiency and explosiveness? They are dominant when it comes to winning football games, but how do we most easily and effectively measure those terms?
Efficiency: Success Rate is an on-base percentage for football; it creates a definition of success for every play — 50 percent of necessary yardage on first down, 70 percent on second down, 100 percent on third or fourth down — and, over time, gives you a clean, easy look at how well a team stays on schedule and ahead of the chains. Explosiveness, your ability to create big plays and easy scores, is often seen as the most important factor in football, but you still have only three to four plays to gain 10 yards, and Success Rate tells you almost everything you need to know about how teams perform in that regard.
Explosiveness: In a pinch, Yards Per Play will suffice just fine when it comes to gauging explosiveness. From an advanced level, there are other options. PPP measures the equivalent point value of every play by assigning a point value to every yard line (based on the net points an offense is expected to generate from yard to yard). Isolated PPP looks at the point value of a team’s successful plays (as determined by the Success Rate equation above). With IsoPPP, you can boil offense down to two questions: How often were you successful? And when you were successful, how successful were you?
How do efficiency and explosiveness interact? Here are a couple of examples.
Miami’s offense was quite explosive in 2013. The Hurricanes ranked 11th in yards per play (6.8) and third in IsoPPP (1.38), but they were just 51st in Success Rate (44.5 percent). This paints the picture of an offense that could eat up wide swaths of yardage in a short amount of time but made too many mistakes to score consistently.
Alternately, Arizona State’s defense ranked a healthy 13th in allowing only a 36.5 percent Success Rate; the Sun Devils were able to create plenty of passing downs and three-and-outs, but they also ranked 67th in yards per play allowed (5.5) and 118th in IsoPPP allowed (1.33). The big plays they allowed were far too big.
Written by Bill Connelly (@SBN_BillC) of Football Study Hall for Athlon Sports. This article appeared in Athlon Sports' 2014 College Football Preview Editions. Visit our online store to order your copy to get more in-depth analysis on the 2014 season.