Building an Offensive Line Metric

Offensive line play is absolutely critical to a team's success. If you can't block, you can't do anything else. No matter how pretty of a play you draw up, it won't matter if the offense doesn't have any time to run it. I'm not advancing a radical theory here, just about everyone who watches football knows that the line is important.

The problem is that we don't have a whole lot of tools in our toolbox to evaluate offensive lines. Lacking any sort of objective measurement, it becomes too easy to just pay lip service to line play and then move on to the glamour positions. Few things in sports irritate me more than the term "skill positions." Quarterbacks and receivers aren't more skilled than the anonymous linemen, they are more glamorous, hence my use of "glamour position." No one writes a story about the head cheerleader going out with the left tackle.

I've been tinkering with an offensive line statistic for some time now based upon two basic skills: yards per carry and sacks allowed. Now, of course the skill of a running back or the ability of a quarterback to evade the rush will greatly impact these numbers, but the reverse has always been true: great running backs need linemen to open up holes for them to run through. All football statistics are inherently interconnected, and it's nearly impossible to isolate one person's contribution. A catch requires an on target throw, and the decision for the quarterback to throw to that target. But an offensive lineman's contribution is even more interconnected than anyone else's, so much so that I rate the line as one unit, not on an individual level.

Let's look at run blocking first. The average SEC team averages 4.42 yards per carry. We don't look at total rushing yards due to the wide variance of attempts by team. Arkansas rushed the ball 367 times, Alabama 570. Alabama's a much better rushing team than Arkansas, but the sheer number of attempts makes the gap look bigger than it is, though the gap is still pretty massive.

The NCAA counts sacks as rushes, so we also need to strip out that data to look solely at rushing. We subtract out a team's sacks allowed from their rush attempts and add the yards lost from sacks back to the team's rushing total. This way, we're only looking at rushing in yards per attempt, giving us a new modified average of 5.085. The standard deviation from the average is 0.604. From there, we can look at how many standard deviations each rushing offense is from the average, multiply the number by 100 just for ease of presentation, and come up with an RSCORE:

A score of zero is league average. The real surprise here is not only that Alabama isn't first in the league in run blocking, but A&M is by a rather wide margin. We can chalk this up to the Johnny Manziel effect, as Manziel rushed for 1410 yards on a 7.01 average. It's impossible to strip out how many of those yards were on designed runs and how many were on improvisational scrambles, so I'm not going to take the data out, but I have to admit that I don't entirely trust it.

It's one thing to give the offensive line credit for a great running back, but giving them credit for a quarterback scrambling after a play break downs seems like we are looking at the wrong thing. I'm tempted to look at things again only with all quarterback rushing taken out entirely, but that could be a future version of the RSCORE. Have it in the comments, but take the rating of A&M with a giant grain of salt.

The other big surprise is how terrible South Carolina was. Their running game struggled more than I was previously aware. LSU is hovering right at league average. We could blame this on Manzeil again, pushing the modified average through the roof, but just over five yards per carry for the modified average is the annual norm. One man is not breaking the whole system.

Turning to pass blocking, we can't look at just the total number of sacks for the same reason we don't look at the total rushing yards. Some teams pass a lot more than others. Auburn only attempted 257 passes all season, while Texas A&M attempted 492.

What we do here is take those sacks removed from the rushing data, and add them to the total number of pass attempts. This is the same concept we use with the ATVSQBPI: sacks are failed pass attempts, and should be treated as such. We then look at the number of sacks allowed per 100 pass attempts. The average SEC team allows 7.483 sacks per 100 attempts with a standard deviation of 3.522.

The reason we are using standard deviation is because the scale for evaluating run blocking and pass blocking is so wildly different. Not only is the pass blocking average higher, it has a much wider variance than rush blocking, in which teams only vary from the average by about a yard. Unlike rush yards per attempt, a high number on sacks allowed per 100 attempts is a bad thing. Here, a team wants a low number, not a high one, so I multiplied that team's standard deviations by negative 100 for ease of presentation and to flip the scale. That way, the PSCORE is on the same scale as the RSCORE, and a high rating is a good thing:

Did you realize Tennessee only allowed 8 sacks all year? Even more impressive, Tennessee threw the ball more than every other SEC team save A&M. The Tennessee line quietly had spectacular year protecting the quarterback. The Vols only allowed 1.677 sacks per 100 attempts.

On the flip side, Auburn and Florida had some terrible pass blocking. Auburn especially was near the conference lead in sacks allowed while attempted the fewest number of passes. The Tigers were the only team to have less than 300 modified attempts, yet only South Carolina and Florida allowed more sacks. Florida was second in the SEC in fewest pass attempts.

Most teams in the SEC were with one standard deviation of league average, but LSU was on the wrong side of the average, surprising no one.

Let's put things all together and come up with an OLINE score by simply adding the RSCORE and the PSCORE.

Texas A&M leads the conference in OLINE score, but take that with a gigantic grain of salt because of Manziel's rushing ability sending their RSCORE through the roof. Alabama was maybe the best line in the SEC, but A&M did have the better pass block score as well.

Arkansas had one of the more fascinating performances, as they couldn't run block for squat, but they were one of the top schools at protecting the quarterback. With one foot in a bucket of boiling water and one in a bucket of freezing water, they are, on average, comfortable.

Auburn finishing in last is no great surprise, but it is a shock to see South Carolina and Florida hanging near the bottom of the table. Florida could at least run block, but get dragged down by some incompetent pass blocking. South Carolina, apparently, couldn't do either, but at least they were consistent.

LSU's patchwork line finished eighth in the conference, which was slightly higher than I expected. Actually, I thought LSU would have a decent RSCORE, but the PSCORE was a pleasant surprise. Sure, it wasn't good, but I was expecting it to be pretty awful. LSU's line did not look good in pass protect last season.

Which gets us to the obvious limitations of the metric. It's nice to have some sort of objective measure on how the lines performed, but sometimes, there is no substitute for the good old eyeball test. This can give us a framework to evaluate the lines at season's end, but the OLINE metric is not a replacement for observation and evaluation.

This is a long way of saying that this just confirms what we already knew: LSU needs to get better at pass blocking, Alabama's line was awesome last year, and Auburn's was terrible. But it does tell us something we didn't know. Not every good team needs a good offensive line to win. Florida and South Carolina showed that there's more than one path to 10 wins.