The effect of mobile quarterbacks on running back success

In 2009, Chris Johnson had one of the best statistical seasons the NFL has ever seen. He became just the 6th player in the history of professional football to rush for over 2,000 yards in a season — he’s since been joined on that list by Adrian Peterson who became the 7th to accomplish the feat in 2012 — and broke Marshall Faulk’s total yards from scrimmage record, combining for a whopping 2,509 yards combined rushing and receiving.

Johnson was sublime during that season and his offensive line — Michael Roos, Eugene Amano, Kevin Mawae, Jake Scott, and David Stewart — deserves a lot of credit too, but CJ never becomes CJ2K without Vince Young’s legs. Johnson’s production jumped from 99.3 yards per game with the immobile Kerry Collins behind center to a staggering 141 yards per game with the elusive Young handing him the ball. The strange thing about this production was that Johnson’s yards per carry were better with Collins (6.27 yards per carry) than they were with Young (5.36 yards per carry) during this season.

There are a couple mitigating factors with the yards per carry statistic here that I want to point out. Because Collins only started 6 games, Johnson’s huge game against the Texans in which he rushed for 197 yards on just 16 carries — including touchdown runs of 57 and 91 yards — really carries an inordinate amount of weight in his overall yards per carry. That 91 yard carry alone added almost a full yard per carry to his total with Collins — though to be fair, he also had some big runs with Young under center — so it’s a little bit of a sample size problem. On the other side, the Titans shameless fixation on getting Johnson over the 2,000 yard mark during a meaningless Week 17 game resulted in him getting 36 carries against a Seahawks team that knew exactly what the Titans were trying to do, effectively watering down his yards per carry average at the end with Young.

In 2010, the Titans again relied on a combination of Young and Collins to get through the season at quarterback, and again, the pattern repeated itself. During games that Collins (or the similarly glacial Rusty Smith) started, Johnson averaged 72.4 yards per game and 4.11 yards per carry. When Young started, he averaged 98.1 yards per game and 4.49 yards per carry.

The thought that a running quarterback could help a running back be more effective isn’t exactly novel. In fact, it probably qualifies as football common sense, particularly in the age of faster defenses and the zone read. However, I wanted to test this theory and put some real numbers to it to see how well it holds up and attempt to quantify the advantage a mobile QB offers.

The metric I ultimately chose to use for determining rushing success was Football Outsiders’ Rushing DVOA. You can read the long form explanation of what DVOA really is here, but the the TL;DR version is that they breakdown the success of a given play based on the specific game situation and normalize that based on difficulty level of opponents and league averages in that same situation. It’s better than more basic stats like total rushing yards because it gives far more credit to a 18 yard touchdown run on 3rd and 3 against a great run defense than it does for an 18 yard run from a throwaway draw play on 3rd and 25 at your own 12 yard line against a bad run defense.

The statistic I decided to use to determine quarterback mobility was total quarterback rushing yards by team for the season.

But didn’t you just say that counting stats suck?

Well yes — and they generally do — but in this case it’s not a bad measure of quarterback mobility. The top 5 quarterbacks in this statistic last year were Cam Newton (754), Russell Wilson (586), Tyrod Taylor (427), DeShone Kizer (419), and Dak Prescott (357). Down at the bottom you find Phillip Rivers (-2), Drew Brees (12), Eli Manning (26), Tom Brady (28), and Jay Cutler (32). The results through the middle and over multiple years were pretty stable and really match with what the eye test says about the running ability of these players so I think it’s the best available stat to use in this application.

I tracked this data back 5 years (2013-2017) to get a decent sample size and also try to capture most of what I consider the “zone read era”. After all, the wide scale incorporation of the zone read has added a real weapon to ground games with running quarterbacks. Defenses have to account for the threat of the QB keep on these plays which means at least one fewer player for the offense to block and often a group of second level defenders who are a beat slower on their reads while they figure out who has the ball after the mesh point.

The less obvious benefit to having a mobile QB comes on plays like the outside zone run — which Titans fans will likely be seeing lots of in the coming season — which often feature boot action off the back side of the run. That boot action, when performed by a QB who offers a viable run threat, holds the backside defenders and prevents them from being able squeeze down cut back lanes and chase down zone runs from behind. This allows the back to be more patient while also stretching the lanes between defenders.

It is easy to make arguments for why running quarterbacks should help a running back be more effective, but do the stats really back that up?

Plugging in the 5 years of data for all 32 teams gave me 160 data points to work with. I plotted these points on an X-Y plane with the X-axis representing the team’s Rushing DVOA for a given season (positive is good, negative is bad) and the Y-axis representing the team’s Total QB Rushing Yards for that same season. When you plot all of the data points and fit a regression line to them it looks like this.

Visually, it’s pretty evident that more QB rushing yards generally correlates to stronger team performances in team Rushing DVOA. The regression analysis run on this data resulted in a correlation coefficient of +0.40 which indicates a moderately strong correlation between having a prolific rushing quarterback and having an effective rushing offense.

Correlation coefficients are basically the measure of how related two stats are and they fall on a spectrum from -1 to +1 with -1 indicating that the two stats are perfectly inversely correlated and a +1 indicating that the two stats are perfectly positively correlated. A coefficient of 0 would tell us that the two numbers have nothing to do with one another. When dealing with measurements in a sport that has as many variables as football, it’s hard to find coefficients higher than +0.60, so a +0.40 tells us that these two things are related and the influence is pretty strong.

In case you’re wondering that data point all by itself in the top right belongs to the incredible 2014 Seahawks rushing attack when Russell Wilson’s 849 rushing yards helped spark a rushing offense that finished with a DVOA of +29.0%. It was easily the most dominant rushing offense of the last 5 years.

Rushing DVOA is a complex stat that takes in to account a ton of factors so I also wanted to see if the influence held up when compared to a more simplistic measure of rushing success. I decided to do the same analysis using Total QB Rushing Yards as the Y-axis again, but replacing DVOA with Team Rushing Yards Per Carry on the X-axis.

Again, visually we can see the positive correlation between the two stats and when you fit a regression to the data, you get a correlation coefficient of +0.49. That tells us that QB rushing yards is even more closely correlated with a team’s yards per carry average than Rushing DVOA was. That makes sense considering that DVOA — being a far more complex stat — has a lot more noise to sift through.

To get a better feel for how much quarterback mobility can help, I decided to break the quarterbacks up in to three groups by level of mobility and look at the average Rushing DVOA scores for their teams. The three groups that I created for the quarterbacks were Runners, Mobile, and Pocket.

I placed players in each group based on my perception of their playing styles — so yes, there is going to be some subjectivity here — but for the most part I thought it was pretty easy to categorize each quarterback without a great deal of internal debate.

The “Runner” category included players like Cam Newton, Russell Wilson, Marcus Mariota, and Tyrod Taylor. This is not meant to say that these guys are “running QBs” in the way that the national media uses it — we’re all tired of hearing the old “can Marcus Mariota win from the pocket” debates — but rather that these players run enough that the defense has to specifically account for the threat of their rushing ability. The idea in my mind was that “contain ________” would be written on the dry erase board in the opponent’s defensive meeting room during the week.

The “Mobile” category included guys like Ryan Tannehill, Andy Dalton, and Aaron Rodgers. These are guys that have some athleticism and can be asked to run some zone read from time to time, but don’t rise to the level of defenses needing to scheme to stop the QB run.

And finally, the “Pocket” category was guys that are complete non-threats on the ground. Players like Tom Brady, Drew Brees, Carson Palmer, and Eli Manning made up this final group.

After placing all the quarterbacks in to the various categories, I then averaged the Rushing DVOA for each category going back over the last 5 seasons.

Pocket: -8.5% Rushing DVOA

Mobile: -5.8% Rushing DVOA

Runner: +1.4% Rushing DVOA

Again, the numbers scream that having a quarterback who can run makes a huge difference in the overall success of a team on the ground. There are obviously outliers in the data that can easily be illustrated. Last year the Saints finished 1st in Rushing DVOA despite having a relatively immobile QB in Drew Brees. Their excellent offensive line and talented backs were good enough to push them to the top of the NFL despite not getting much help from their signal caller when it comes to threatening on the ground.

On the opposite end of the spectrum, the Seahawks finished 23rd in Rushing DVOA despite the presence of Russell Wilson. Their awful offensive line play and banged up running backs were too much for Wilson’s effect to overcome.

All three of the results above confirm the feeling I had in 2009 and 2010 when Vince Young was helping Chris Johnson put up ridiculous stats: quarterbacks who double as talented rushers provide a significant advantage to their backfield mates.

What does that mean for the NFL? The league is still definitively a passing league. In fact, most statistical analyses indicate that rushing success barely matters when it comes to winning games. If you have to choose between having a strong passing attack and a strong running attack, you should choose the passing game every time. However, an efficient running game can help move the chains and keep the offense in rhythm which still has some value.

The best solution for NFL teams is obviously to “get you a QB who can do both”, but those are in short supply. Guys like Carson Wentz, Russell Wilson, Cam Newton, and yes, Marcus Mariota, are a nightmare for opposing defensive coordinators to deal with. However, players like Tyrod Taylor who help produce a very strong running game with their legs, but struggle in the passing attack, are not going to help teams win games more than guys like Brees who won’t do a ton for your run game (besides force safeties out of the box), but consistently destroy defenses through the air.

The Titans duo of Derrick Henry and Dion Lewis could put up big numbers in Tennessee in 2018. Both are talented runners working behind a strong offensive line with a highly mobile quarterback. Throw in an improved offensive scheme that is expected to take more advantage of Mariota’s athleticism and you have a recipe for a big season statistically.

With fantasy draft season approaching I think this could be something to keep in mind from that perspective as well. Running backs that suddenly inherited a running quarterback like Tyrod Taylor in Cleveland or Alex Smith in Washington could see a significant boost in productivity thanks to the skill set of their new colleagues.