Over the past number of years there have been numerous attempts to evaluate the passing game in collegiate and professional football. The NFL uses its out-dated Passer Rating system, ESPN uses its Total Quarterback Rating (QBR), while others around the internet use tools like Wins Above Replacement. Each system has its benefits and shortcomings, but as a whole, they all do a decent job of telling us who is the best quarterback in a respective league. While much has been made over time about evaluating quarterbacks, running backs have been strangely left out of the advanced evaluation process.
In my searches on the internet, I was only able to find one site who evaluated running backs with a rating system, but unfortunately, they only covered the NFL and followed the same flawed techniques of the Passer Rating system. Given the importance of the rushing game in college football, I set out to evaluate running backs in a new and more advanced way.
In order to do a rating like this, we need something in which to measure like points or wins added. As I stated last week
, since the college game has so many different teams with so many different playing styles, it's difficult to quantify an individual player's contribution to team wins. Because of this, I used points scored as the dependent variable in the rating. Next, I had to figure out just what made a running back valuable to his respective team. Was it rushing yards or rushing yards per attempt? Was it fumbles, fumble percentage, or some other unknown variable? After running multiple regressions, I came to the conclusion that the biggest contribution to the rushing game came from rushing yards per carry and successful rushes
is defined as a rush without a fumble lost). Strangely, during the experiment, different ways to evaluate fumbles like fumble percentage, fumble rate proved insignificant. We all know that fumbles lost is a significant stat because it signifies the loss of a team's most important asset; a possession. To correct this, I simply took a team's rushing attempts and subtracted fumbles lost. This way we get a more advanced way of tracking fumbles.
Just like last week, touchdowns are absent from the ratings for multiple reasons.
First, a rusher could benefit from a team who passes the ball down-field and falls short of the goal line, setting up numerous short yardage rushes for touchdowns. Second, short yardage fullbacks score frequently in short yardage situations. Finally, touchdowns don't directly show the ability to move the ball down-field, they show the final result. Yards per carry and successful rushes show how well a team moves the ball.
Without further ado, the below chart shows last season's team leaders in points added through rushing. The values are significant at the 1% level and are as follows: Yards per Carry (41.2), Successful Attempts (0.12).
When the factors are combined, we find that Oregon, Army, and Georgia Tech were the nation's leading rushers in 2012. Also, we see that the regression rating gives us results that the eye test would very much agree with. The teams who led the nation were not only very efficient with their rushing attempts, but they also did so at a high volume without losing possession. This is exactly what I wanted to accomplish with these ratings. (Last season, Kentucky was ranked 83rd and Neal Brown's Texas Tech offense was ranked 63rd in estimated points added through rushing.
) However, some teams like Georgia, who had some of the nation's elite rushers in Todd Gurley and Keith Marshall, were penalized because they were part of an offense with a prolific passer. This system isn't perfect by any stretch of the imagination, but it gives a very good look at which team has the best rushing unit. Now we turn our focus to evaluating the individual rusher.
First and foremost, be very careful when examining individual rushers on the same team as the final result will not be a true reflection on points added to a team. Let's examine Georgia's two premier rushers, Todd Gurley and Keith Marshall, for instance. Gurley averaged 6.2 YPC and Marshall averaged 6.5. Marshall had a very respectable 117 carries while Gurley had 222 on the season. Once everything is calculated, Gurley receives a points added rating of 284.5 while Marshall receives 283.1. While Gurley still comes out ahead, there can be a problem because of volume. If Marshall rushed for 6.6 YPC, he'd be ahead of Gurley despite rushing many fewer times. Individuals need to be placed into categories like players with 100+ carries, players with 150+, players with 200+, etc.
Second, when doing this individually, the individual totals don't add up to team totals. For example, when Gurley and Marshall are added together, Georgia would be expected to score around 567 points on the ground. When all of Georgia's running backs were averaged, they were estimated to score 267 last year, a clear difference. These ratings were meant for teams, not players, and though they can theoretically be used for players, be careful.
No system is or will ever be perfect
. There are simply too many factors and too many players to divide credit for in football. For instance there's no possible way to account for blocking with running backs even though that's a very important part of what a running back does. Shortcomings are a part of every system, but now we have a better way to evaluate the rushing game