clock menu more-arrow no yes mobile

Filed under:

Visualizing Quarterback styles under Paul Johnson

A statistical breakdown of each QB's performance under CPJ, with a helluva lot of graphics.

Kevin C. Cox

I was inspired by Grant Heffley's two excellent posts this week on the different styles of quarterbacks that we have had under Coach Johnson. But I am a visuals kind of guy so I set out to try and visualize the differences between the four horsemen; Nesbitt, Washington, Lee, and Thomas. I hope you like charts because I kind of went overboad with them, my apologies if not.

The first thing I want to look at is the distribution of yards gained on all carries by each quarterback. Was one QB a make-or-break QB that either lost 5 yards or gained 30? Maybe one did a better job of not losing yards than the others. The following plot shows just that. The X-axis is yards gained and the Y-axis in each plot is the percentage of carries for each QB that gained a given yard amount. The dashed line in each plot represents a gain of 0.


  • As expected, to me at least, Nesbitt did the best job of not losing yards on his runs (least amount of area to the left of the dashed line = less carries for loss). Actually, you almost could argue Tevin Washington did as good of a job of not losing yards when he ran the ball.
  • Vad Lee was probably the most hit or miss of any QB. The area under his density curve for runs between 5 and 15 yards is greater than any of the others.
  • Thomas has a ton of spikes downfield, but I think thats just because he hasn't had that many carries. His plot should even out as he starts to run more often this year.
Was there any change in the distribution of yards gained on runs by year for any of the 4 quarterbacks? To make it easier to view the difference by year I have plotted the distribution with a boxplot. The middle line in the boxplot represents the median of the data, with the box representing the 25th and 75th percentiles of the data.The lines extend out to the max and min non-outlier yards gained, with each dot representing an outlier. The smaller the box the tighter the distribution (less variance in yards gained per run) and the closer the middle line (the median) is to one side of the box the more skewed the data is to that side.The dashed red line is the median for all runs, 3 yards. A good example of skew are the runs by Tevin Washington in 2012; His box is very tight and the middle line is to the left of the overall median, which means most of his runs were between 0 and 3 yards.

  • I think the 2012 plot shows where everyone's high expectations for Vad Lee came from. As a runner in 2012 he didn't have any huge losses (no dots to the left of his line), he gained more yards more often than any one else has (the right line of his box - the 75th percentile of his carries- is farther right than any other QB besides 2009 Tevin, who had practically no carries), and he had a couple huge runs that year as well.
  • 2013 was a different story. Vad had more small-gain carries that year (25th percentile was less than 2012's) and had fewer huge runs (max run was about 22 yards).
  • Look how small Nesbitt's Inter-Quartile Range was in 2008 and 2009 (the width of his box). Dude was consistent.

That was all about the distribution of carries; how often did you gain a certain number of yards on each carry. Well I want to look at how good each quarterback was when he ran the ball. Two methods to do that are Yards per Carry and Success Rate. Yards per Carry is exactly that; how many yards do you gain on average when you run the ball. Success Rate is more focused on did you gain enough yards for the play to be successful. This is defined as gaining at least 50% of the necessary yards for a first down on first down, 70% on 2nd down, and 100% (so getting the first down) on 3rd or 4th down. This is a good way to measure "move the chains" running backs. The following plot shows the success rate of each quarterback in each year on the x-axis, the yards per carry on the y-axis, and is broken up by down (except 4th since there weren't that many carries to visualize). The size of each quarterback's name is the number of carries they had on each down. I had to exclude Justin Thomas last year and Tevin Washington in 2009 because they really didn't have enough carries in each down to get a good sense of their performance.


  • Tevin Washington's 3rd Down performance in 2012 really stands out. He had over a 60% success rate on 3rd downs, so when he ran the ball on third down he got a first 60% of the time, that is pretty great. I am starting to appreciate Tevin a lot more with this analysis.
  • Compare that to Nesbitt on third downs, who still had a great success rate (2009's was just at 60%) but also averaged closer to 5 and 6 yards per carry on 3rd downs. He made a lot of big plays on third down for us, and he wasn't just gaining a lot of yards on 3rd and longs. His success rate's were all really good.
  • Obviously Vad's 2012 performance is the outlier in the room. And what happened in 2013? He went from averaging about 5 and a half yards per carry on 1st downs in 2012 to less than 4 in 2013, and went from over 8 yards a carry on third downs to less than 3.5 in 2013. That is a huge drop in performance.

Next I want to look at where each QB's carries occurred on the pitch field. Was someone predominately used in goal line situations in certain years? Did we rely on a QB to move us down field and then let the running backs punch it in? The next plot shows the spot on the field on the x-axis and the percent of the QB's carries that occurred at each spot on the field. It has been split up by year again to allow you to compare QB changes a little better.


  • The difference between 2008 and 2009 is interesting for Nesbitt. In 2008 our offense wasn't as good, and neither was our defense compared to 2009. So we had the ball more often in the middle of the field and just after a kickoff. But in 2009 we were rolling so we had the ball more often on our opponent's side of the field in scoring position. At least I think that explains the discrepancy.
  • In 2010 Nesbitt had a spike in carries in the redzone. Could this be from the loss of Dwyer that year? I don't know.
  • Tevin's distribution of the location of his carries looks nothing like Nesbitt's from years prior. Perhaps this chart measures changes in the offense more than changes in the QB.
  • It appears as if Justin Thomas rarely had many "move the chains" drives where he carried us down the field. He had the least percentage of carries past midfield of any QB.

Ok, I think that was about all the running analysis I can handle right now. There is one more stylistic difference I want to explore. And that is, who does each QB throw the ball to? Does one rely on his receivers downfield, or perhaps one is constantly dumping off to his running backs? The x-axis on the next plot is the position of the intended target (so incomplete and complete doesn't matter, just who you threw it to) and the y-axis is the number of attempts.


  • The one huge problem with this graph is that you could just plot it as "With or Without Bebe" and you would get the same trend as breaking it out by year and QB. Have we changed our style because a.) We don't have the receiving talent like we did in 2008/2009, or b.) our offense has changed to focus on the playmaking ability of our running backs, or c.) our QBs are garbage and can't complete a pass down field or d.) all of the above? I have no idea.
  • Tevin really relied on our running backs in the passing game in 2012. I like the visual comparison between him and Vad. Vad seemed to be either more able or more willing to throw to his receivers downfield, while Tevin was looking for the running backs.

Wow this post got long. Sorry about the dump of graphics, but I really wanted to visualize the differences between the quarterbacks over the years. If there are any other comparisons you would like to see then just let me know in the comments. And obviously please leave any of your observations or criticisms in the comments.