Running Back Statistics: A New Way To Judge

Using statistics or metrics to evaluate football players is always an inexact science at best and a completely futile exercise at worst. We as fans are working off of limited knowledge of schemes and individual player responsibilities, making it difficult to keep statistics in context when comparing across leagues or time periods. We also have to deal with the difficulty inherent in separating an individual’s performance from that of the team. Even Jerry Rice would struggle with Blaine Gabbert as his QB. Nonetheless, many people have made a valiant effort at tackling these problems. I believe we can do better though, and in this post I will explain a new way to evaluate running backs.

Yards Per Attempt

First, let’s look at the best metrics currently available, and why I believe they fall short. The first and probably most familiar is yards per attempt (YPA). Shown above is the top 10 running backs ranked by YPA. The statistic is straightforward enough, take the total number of yards a running back gains and divide by the attempts it took to get them. This gives you the average yards a running back gained on any individual attempt. This is a great way to compare running backs, but it’s a little biased. As any statistician will tell you, averages are susceptible to outliers. One or two crazy events can skew an average badly. This applies to YPA, but there’s a kicker: for YPA, there can only be outliers in one direction. The lowest possible amount of yards one can reasonably expect to gain on one play is zero (maybe -1 or -2, but that doesn’t change things much). In contrast, the most possible yards a back can gain is 99. In short, large outliers are only possible in the positive direction, considering most YPAs are around four. This means that running backs who crack off one or two big runs per game but are otherwise terrible can still have a great YPA. This is where our next metric comes in.

Success Rate

Football Outsiders keeps track of a wonderful metric known as success rate, and the top ten backs based on that metric are shown above. A full breakdown of success rate can be found at Football Outsiders website here but I will give a quick and dirty explanation. Basically, success rate measures how often a running back makes a "successful" play, which is defined primarily by down and distance. For example, a two yard run on first and ten would not be a success, but that same two yard run counts as a success if it comes on third and one. Basically, it tells you how often a running back does the job that is asked of them, but rewards no extra credit for big gains. We like extra credit though, so this won’t do on its own either.

Rushing Index

Enter my new metric: rushing index. This concept will be very familiar to baseball fans, as it closely resembles OPS+. The idea is that I added together success rate and YPA, normalizing to league average, which I set to a value of 100. This means a player with a rushing index of 120 was 20% above average, and a running back with a rushing index of 80 was 20% below average (sort of, it’s more complicated than that, but roll with it). But wait, you might be saying, how in the world did you add YPA and success rate, they are on two completely different scales! This is true, which is why I had to do the averaging before adding them together. The basic concept is this: I calculate the average YPA across qualifying players (minimum 100 rushes), and calculate what percentage above or below average each player was. I then do the same for success rate, and add the two together, dividing by two so that the average comes out at 100. So without further delay, here are the rushing indexes of your qualifying running backs for 2012.

The Results

So the first thing that jumps out is that C.J. Spiller get ranked above Adrian Peterson. That's kind of crazy, but I think it can be explained by the difference in carries. Rushing index (RI) does not place any value on volume as long as the minimum number of rushes is met. RI is not meant to be a catch-all, single number to describe running backs, just a more fair way of comparing than other options.

Aside from that craziness, there are some other things worth note. Only three players (AD, Alfred Morris, and Ahmad Bradshaw) were in the top ten for each success rate and YPA. I like this, it means players who were getting penalized using one metric or the other are getting full credit here. Perhaps the most surprising guy is Jamaal Charles, who ranks sixth in RI despite an abysmal ranking of 29 in success rate. Part of what causes this is the lack of variance in success rate. There simply isn't a huge difference between the between the best and worst guy, so being the worst in success rate won't drag a player down as much as being the worst in YPA. Expanding on this, RI implies that there really are only one or two elite backs, followed by a bunch of guys who are a little above average and then a bunch of bums.

Speaking of those bums, Trent Richardson is not looked upon favorably here. I mean, ouch, that's bad. Especially for the third pick overall. There's hope I guess, but that's what it is at this point. Forget breaking big runs, Trent wasn't able to pick up the bare minimum to be considered successful. It is my personal opinion, however, that much of Trent's struggles can be attributed to our less than stellar coaching last year.

RI is nice, but it is still packed with flaws, like any metric in football. Some of the obvious ones have already been mentioned. It doesn't account for differences in scheme, it doesn't account for quantity, and it doesn't account for quality of teammates. It also isn't adjusted for strength of opponents, and perhaps most importantly, it only ranks backs in terms of running, and does not account for contributions made in the passing game, either through receptions or blocking. Still, I think this was a fun exercise and these numbers should be useful in the future. Hopefully, given enough time, I can create this data for other years and try to pick out trends. But all of that will have to wait for another post.