Risk vs Reward: How Running QB's Can Impact Winning

It started with Michael Vick in the 1999 season and it carried on through Bryan Randall, Vick's brother, Marcus, Tyrod Taylor, Logan Thomas....the list goes on and on. With all of those successful starting quarterbacks in the program's brief history as a national brand, it's hard to blame anyone for expecting a quarterback with some wheels.

But when the coaching regime transitioned from Beamer's more pro-style approach to Fuente's "Big XII" offense, the reliance on a running quarterback increased to a level that has seemed, frankly, unsustainable.

In 2016, transfer QB Jerod Evans was the catalyst of possibly the best offense in school history. Evans was just 29 rushing yards away from being a 1,000-yard rusher and 3,000-yard passer, a rare feat for any level of college football. That offseason, Evans made a surprise decision to enter the NFL draft. The most reported reason for his exit, albeit via hearsay, was that Evans didn't think his body could take another 226 carries. Regardless of the reason for his departure, Virginia Tech hasn't reached that level of offensive prowess or winning ever since. And since then, the Hokies have started five different QB's.

I'll spare any reader a long diatribe on play-calling when it comes to the quarterback run. If you're reading this, you most likely have watched a fair amount of Virginia Tech football and understand the high likelihood of a quarterback run on 3rd down. But the quarterback has become not just a wild card in the rushing offense. Over the past four seasons, the quarterback is accounting for more and more of the Hokies' rushing attack. And in those four seasons, the Hokies have become a stagnant offense that has struggled to live up to expectations consistently. According to leading analytical measures, Virginia Tech has risen from objective ineptitude and settled into offensive mediocrity.

Virginia Tech is getting more and more of their rushing production from the QB position as time goes on in the Fuente era. Granted, there are a few surface reasons for this that need to be addressed. First, the immediate transition from Evans to Josh Jackson and then Ryan Willis was a huge change in athletic abilities. Jackson was an if-needed runner, and Willis was a decent runner on a team that was "pass first". So an increase is to be expected. Second, Hendon Hooker was as shifty as a runner as you're going to find. His ability to makes guys miss at the second level gave the Hokies a much-needed boost in the running game. He was the best runner on the team for most of his time in Blacksburg. Which brings me to my last point which is that the Hokies really didn't have any running backs that were better runners than the quarterbacks. Until Khalil Herbert's nationally acclaimed season in 2020, the Hokies were thin at running back almost every year. An early departure from Travon McMillian followed by unfortunate injuries early in the careers of Jalen Holston and Deshawn McClease facilitated a running back room that lacked experience and explosiveness.

All of these reasons are valid and make sense to a micro viewpoint. Overall, however, they point to a real curious question regarding the identity of Virginia Tech's offense: Why is the quarterback playing running back?

This will be year 6 of the Fuente era. Starting QB Braxton Burmeister is arguably the best runner in the backfield. Last season, he and fellow QB Hendon Hooker combined for almost 34% of the total rushing yards for the Hokies. That's even with RB Khalil Herbert reaching almost 1,200 rushing yards on 150 carries, the entrance of Rutgers transfer RB Raheem Blackshear, and a finally healthy senior RB Jalen Holston. Safe to say, it looks like Burmeister will be running a fair amount this season. Behind him? Freshmen QB's Knox Kadum and Tahj Bullock. It doesn't take a genius to realize that Burmeister's health is critical for the Hokies success. Still, no one expects the offense to change much from last season.

But hey, I'm in the stands and those guys are on the field for a reason. Full stop. If that's what the program thinks is the best way to win football games, then that's that.

For the rest of this article, we're going to look at why the Hokies might be leaning into a running quarterback. We'll look in-depth into the Hokies' offensive metrics and stats. We'll also take a look at a national landscape over the past 5 seasons. I think you'll be surprised, as I was, with what we find.

For those of you not interested in the explanation of the data I used, how I cleaned it, and the methods used to come up with these metrics, feel free to skip to the next subheading.

To understand why the Hokie quarterback is generating an increasing amount of production in the run game, I think it's important to look at the national landscape first. To do this, I collected the rushing data for the top 100 rushing quarterbacks in each season starting with 2016. Why top 100? Simple. College stats include sacks as negative QB rushing yards. It's really dumb, I know. To steer clear of clouding the data with pocket passers that never run and have awful, and sometimes negative, rushing totals, I made the cut line at 100. Also, 100 is just an easy number to work with. Next, I grabbed the total rushing numbers, win percentages and Beta_Rank information for those teams. A bit about Beta_Rank:

Beta_Rank is an advanced analytical approach to ranking college football teams in a myriad of different categories. For this article, we will be focusing on "Drive Efficiency", "Explosive Drives", and an over-arching "Offensive Score" (which is the sum of Play Efficiency, Drive Efficiency, Explosiveness, and Negative Drives). These are self-explanatory on a surface level, but for a more detailed explanation from the creator himself, Rob Bowron, check out this link: What is Beta_Rank?

I used a split for FBS to see at what "level" of FBS football could some of these trends differ. For this, I grouped the Power 5 + American Ath. Conference + Notre Dame. The rest of FBS football (Mountain West, Sun Belt, CUSA, MAC, and Independents less Notre Dame) was in a second group. I chose this line of demarcation because the quality of play at the top of the AAC is, quite frankly, better than the bottom of most Power 5 conferences. Also, elite offenses over the past 5 years from Central Florida, Houston, Cincinnati, etc. needed to be included among the best in the country. The inclusion of Notre Dame speaks for itself.

The term "QB production rate" is used throughout the article are simply means the percentage of total rushing yards for a team in a given season that are by the quarterback. I didn't include analysis on rushing attempts due to sacks, QB kneels, scrambling, etc. Again, this is not about play calling, but more about production on the stat sheet. I also wanted to capture teams, like Virginia Tech, that have struggled at the running back position in the past few years.

I'll have to admit, the analyses that I go through are relatively elementary in the advanced analytics world. I dove into other practices, most notably neural networks, but didn't see any results that changed the big picture. Linear analysis is the easiest to present and, ironically, often the most accurate.

This is the first and most obvious question to ask. Do teams with quarterbacks that can break a game open on the ground win more or less than teams with a more traditional rushing attack? To answer this, let's look at a simple chart plotting the Percent of total rushing yards for a team by the QB and the corresponding team's win percentage.

To understand complex relationships, we can use a basic statistical measure called a p-value. A model's p-value is the probability that a new data point added to your model would go against what the current model is predicting. A higher p-value (which is measured on a scale of 0 to 1) means the model is less significant, while a smaller p-value means the relationship is more significant. In general, a p-value of <0.05 is considered significant.

You can see above that this model's p-value is 0.4666. This means that we cannot say that the percent of rushing yards from the QB impacts winning. The slope of this model is slightly negative, which could point to having a higher QB production rate may negatively impact winning; however, this is only minimal. This seems like a nail in the coffin, but let's peel back one more layer. Let's look at what happens when we split this model into our two groups, Power 5 + AAC + ND and all the others.

By splitting up the FBS on our previously discussed line of demarcation, we can see that the "lower-level" teams are negatively impacted when their QB production rate is higher at a substantially significant rate. On the other hand, "higher-level" see almost no impact at all on their winning. If anything, we can see a slight positive slope.

This is the exact opposite of what I expected to see. I would've thought that a dynamic quarterback at the "lower-level" of college football would have a greater impact. It could be that "lower-level" programs with consistent winning seasons such as App State, Boise State, Brigham Young, Marshall, San Diego St, etc. have strong cultures of developing top running backs. Those teams can then focus on developing passing quarterbacks to compliment their run game and create a balanced offense that is hard for other "lower-level" teams to stop with less athletic defenses. But this is just a theory.

Considering Virginia Tech, the question of 'why have we seen an increase in QB production rate' now has more questions than answers. It's clear to us now that getting a higher rate of your rushing yards from the QB spot doesn't help with winning. So why is it done?

Often, we have to go deeper into our statistical studies than we initially thought. Seeing a problem and saying "if this then this" is not going to get it done on most occasions. Our next step is to find out what statistical measurement does impact winning at the "higher-level" of college football. From there, we can backtrack to see how QB production rate plays into those statistics.

First up, let's look at how the overarching "Offensive Score" from Beta_Rank can predict win percentage.

In what should come as no surprise, teams that are better offensively win more games. That's not news to anyone. Our next step is to see how QB production rate fits into Offensive Beta_Rank Score.

It doesn't take a statistician to realize that there's no relationship between QB production rate and Offensive Beta_Rank Score.

So, that leaves us at an odd place in the case of Virginia Tech. On one hand, we see that Hokie QB's are accounting for a larger percentage of team rushing yards year over year for every year since 2017, even with one of the school's best-ever running back seasons with Khalil Herbert. On the other hand, we see no positive correlation between QB production rate and offensive output and/or win percentage.

Let's look at some more offensive metrics that make up the Offensive Beta_Rank Score from the Hokies over the past few years.

Here is where we can see the positive impact that a running quarterback can have on an offense. There is positive growth in both Play Efficiency and Explosive Drives that coincides with an increase in QB production rate. In 2020, the Hokies also saw a decrease in the Negative Drives metrics and finished the seasoned with the 24th best score in that metric. (The 2017 score for Negative Drives was the 18th best nationally.)

Where the Hokies have seen a negative trend is in Drive Efficiency. Drive Efficiency is the point value that is not captured by Explosive Drives, Negative Drives, and Play Efficiency. This is where the Hokies struggled last season and where the offense needs to improve to enter the upper echelon nationally. The metric of Drive Efficiency can be looked at as how an offense can score on possessions that are comprised mostly of "average" plays. These are plays that go from around 3.5-7.5 yards on average. For reference, Virginia Tech has ranked #52, #93, #27, #36, and #119 nationally, respectively, since Fuente's arrival in 2016. That's right, the Hokies were the 119th ranked offense in Drive Efficiency in 2020.

Like many fans, I struggled to put a name to the Hokies offensive struggles last season. I think we can write the term "Drive Efficiency" on our collective drawing board of explanations. In a year where the Hokies had one of the most productive running backs in the nation, one of the best offensive lines in the nation, a returning starting QB, and 3 returning contributors at WR/TE...the Hokies still scored less than 28 points in over a third of their games. Surface level, it doesn't make sense. If we take a look at the Drive Efficiency Scores for 2016-2020 vs the raw total QB rushing yards of the corresponding teams, we can see that Virginia Tech's 2020 struggles were somewhat of an outlier.

I initially expected to see a modest negative relationship between offenses that relied heavily on running with their quarterbacks and winning percentages. If you look at the past teams to make the College Football Playoff, a majority of them have steered clear of running with the quarterback. Even with great athletic QBs like Trevor Lawrence, Justin Fields, and even Ian Book, teams that are at the top of college football get a larger proportion of their rushing production from running backs than other teams. So when the results came back negligible (no significant relationship between QB production rate and winning percentage), I had to wonder: why would teams even risk injuring their quarterback? And that's where I think the difference lies.

Everyone's hurting a bit towards the end of a football season, but to have your starting QB playing with significant pain is not going to be a recipe for success. This is where the Hokies have struggled over the past few seasons. Hendon Hooker struggled to be at 100% in his time as a starting QB, and the drop-off in his production last season showed that. Going back to Evans, his career as a Hokie may have been cut short due to the physical impact of running that much. Heck, Josh Jackson's leg broke in the ODU game in 2018 (a play that could happen at any time, but you get my point). The teams that win at a high level do so because their quarterback can be just that, a quarterback not a running back.

QB development at Virginia Tech seems to be focused heavily on perfecting the RPO as opposed to developing as a pocket passer. Granted, the offense is designed to be RPO-heavy. However, Jackson, Willis, and Hooker all failed to develop as passers at the level needed for the program to increase its win totals. As you can see below, the offensive metrics for Effective Pass and Effective Rush have moved inversely as quarterbacks have spent more time in Fuente's system. I added annotations to show how incumbent QB's have become better at the run game than the passing game over time (QB name followed by the number of years in the program).

It's pretty safe to say that when your quarterback isn't developing as a passer, you have to rely more and more on the run game. In the Hokies' offense, a large part of the running game is executing the RPO against a defense that has to defend the intermediate passing game first, then attack the run second. Virginia Tech quarterbacks have struggled to make defenses focus on the passing game, which has put an increasing burden on the run game to carry the offense. Tech's inability to consistently move the ball down the field with the intermediate passing game, as shown by the decreasing scores in Drive Efficiency, has handcuffed the offense.

Fuente feels as if Burmeister may have a better grip on the passing game than some previous Hokie QB's, however, which is a welcome sign for Hokie fans who are desperate for the "BIG XII" offense they were promised six years ago. Recently at ACC Media Days, Fuente commented on Burmeister's development as a passer:

It could also be argued that the Hokies have a deeper pass-catching group than that 2016 team. Many who cover the conference say the Hokies may even have a better talent pool at the position group than Clemson and UNC, two teams with Heisman caliber QBs. That says a lot about what this offense could look like in 2021. Still, the ball needs to get to those guys at a much higher clip than what we've seen in the past few seasons in Blacksburg. Will the Hokies attempt more than last year's 23 passes per game next season? Yes. I'd put money on that. Will it be to the tune of 32 attempts per game like in 2016? Doubtful.

This offense is still going to be run-heavy. Last season, the Hokies ran the ball 219 times more than they threw the ball. To steer things back to our main topic, Hokie QB's only threw the ball 67 more times than they ran it last season. You don't flip that script in one offseason without bringing in a new QB. If I were to put a number on it, I think you could see that Hokies throw the ball 28-30 per game.

Why? Pretty simple, Braxton Burmeister is one of the best runners on the team. He's too good of an athlete not to run him 12-15 times a game. That would be less than the 17 attempts per game that Hokies QB's racked up last season, but still a high number for a thin QB room. Fuente said during ACC Media Days that Burmeister will have a huge role in Tech's running game, but understands that some of that burden has to be put on the running backs. "...we need to find a way to share that load out of the running back room," he said.

How Fuente and OC Brad Cornelson handle the running load for Burmeister this season is one of the most intriguing schematic storylines of their tenure in Blacksburg. The offensive talent is not a problem at Virginia Tech this season. Depth is. And with a defense that is poised to be one of the most improved in the ACC in 2021, a high-level offense could see the Hokies back in Charlotte for the ACC Championship Game. That only happens if Burmeister is healthy. I know it, the coaches know it, the whole ACC knows it. The question yet to be answered is this: How much risk will the Hokies be willing to take with their speedy signal-caller?