Does the weight of an offensive line matter in college football?

I hate college football because I love it so much.

When my beloved Utes lose it ruins my day. I have tried to chill out and distance myself from the competition and the game. At the end it really doesn’t matter. Sports are many things, but I need to view it as just an escape. When you flip on the tube or look at your twitter feed it doesn’t take long to realize that much heavier and pressing issues exist in the world, sports are a great way to get away. So, for a moment in this post, let’s escape into college football. Let’s pretend it is the most important thing we can talk about.

And the most important thing that I looked at recently was if the size of an offensive line impacts the performance of a team. Or does weighing more really matter? The long short of it is – that’s not the most important thing. The most important thing is protecting your most valuable scoring assets.

Groundbreaking? I know…really obvious stuff. You can quit reading now if you would like, but I am going to continue on and show the mathematical significance of the mentioned statements. I am not a quitter.

Question 1: Does the weight of an offensive line matter? Or if my team weighs more will we win more?

Answer: No, not really.

1. 1. There probably is a level, like DII, DIII, High School and lower, where the weight of an offensive line has a greater correlation to wins. In the FBS and the NFL these lineman are all basically the same size. The average weight of 254 college teams ranged from 263 to 320 lbs. A range of 56 pounds might seem like a lot, but if you look at the frequency of weight, its a different story. Most teams (90%) were in the range of 290 to 310 lbs, a much tighter range (see below). You are dealing with basically the same weight across teams.
   2. I correlated the wins of 128 CFB teams with their avg. O-Line weight and found that it had no linear relationship (a correlation is basically a quick and dirty, “Are we on to something?”). I was actually very surprised that weight + wins wasn’t highly correlated.
   3. The avg. weight of teams with wins above and below 8 was basically identical at 298 lbs.

4. Finally, if you look at the weight to wins below you can see that there isn’t a necessarily an ideal weight to get more wins.

5. The regression of QB Rating + wins has a highly significant relationship, whereas the weight + wins is not worth noting.

Question 2: If the offensive line doesn’t impact wins that much, how do they impact the game?

Answer: That question is misleading by nature so to create a sense of confusion and emotion so you will keep reading. Of course the offensive line wins games! Any football fan knows that you win games based on your front 7’s. The rise and dynasty of Alabama was built on the backs of great offensive lines. It is hard to gauge how good an offensive line based on their size. Its a crude data point to focus on. Just because you are big doesn’t mean you will win. Right, Fezzik? The better way to understand how an offensive line wins games is how other members of the offense performs. It’s better to focus on how well they are protecting the most important assets: the Quarterback and Running Back.

1. There is a positive moderate correlation of QB performance (QB Rating) + wins at a 0.56. This was the most significant correlation I found in the dataset I was working with (128 FBS teams, average weight, record, pt differential, conference, QB rating, team’s leading rusher, average margin of victory/loss).
2. The correlation of weight of O-Line + QB Rating is as significant as RBs performance (Yards gained) + wins (0.16).

Question 3: I still don’t believe you. Can you provide more evidence?

Answer: Football doesn’t have a great way to track the offensive line. It’s not like baseball where essentially every thing is tracked. The sack is probably the most accurate way to see the impact of the O-line. Of the 128 teams observed the amount of sacks given up, had an average of 12.7, with a the 0.7 standard deviation, and min number of 12 and a max of 15. That is a tight distribution. I thought that because of the tight distribution that any correlation would be insignificant. That isn’t the case. The correlation of sacks + QB rating or sacks + wins is moderately negative, at -0.22 and -0.26 respectively.  Weight doesn’t matter. If you were to correlate weight + sacks you get a 0.04. Snore.

Data used for this post was taken from CBS Sports, Sports-Reference.com, and cfbstats.com.

This slideshow requires JavaScript.