Dan Mullen’s QBs are really effective runners.
At least, that’s the reputation he has built. But that’s not necessarily what the statistics indicate.
In 2007 and 2008, Tim Tebow averaged 4.1 yards per rush. No other Mullen starting QB would average more than 3.7 yards per attempt until Dak Prescott averaged 6.2 yards per rush in 2013. In fact, prior to Prescott, Mississippi State QBs averaged just 2.9 yards per rush. That increased significantly with Prescott and Nick Fitzgerald, with those QBs averaging 5.5 yards per rush.
This is instructive because during those Prescott/Fitzgerald years, Mississippi State averaged 7.5 yards per pass attempt while only slightly lower at 7.2 yards per attempt prior to Prescott’s arrival. Yet the Bulldogs only averaged 27.4 points per game during that time versus 32.3 with Prescott or Fitzgerald under center.
I’d submit that the key difference between those QBs was not necessarily their ability to throw the ball (although Prescott was definitely the best at that). I think it can be argued that the difference is that Mullen finally found QBs who were efficient running the ball.
Why QB running is so important
Note that I am specifically talking about efficient running of the QB. With the exception of Tyler Russell (43 carries for minus-5 yards in 2012), Mullen still ran his QBs prior to Prescott’s arrival. Tyson Lee ran the ball 74 times in 2009 (2.4 yards per rush) and Chris Relf ran the ball 317 times in 2010 and 2011 (3.4 yards per rush average). The issue is that they weren’t very efficient.
When I collated the rushing yardage from every FBS QB during the 2017 season (including sacks), they averaged 3.07 yards per rush on 42 attempts. So while Relf ran the ball a ton in his two seasons, he was only slightly better than the average in the country.
Compare that to Prescott in 2013. Not only did he have 134 rushing attempts, but he averaged 6.2 yards per attempt. That means that every time Mullen decided to have Prescott run the ball, he was more than three yards better than a typical QB run.
Recently I pointed out how yards per play is a key indicator tied to scoring and this points out why efficient running of the ball is so important.
Let’s say that we have a QB who averages the exact yards per attempt of all other 2017 FBS QBs (7.41 yards per attempt). The team runs the ball on first down and gets stuffed, setting up second-and-10. The QB then runs a read-option play that nets 2 yards on second down. On third down, the QB completes a ball for his exact yards per attempt average. The result is a punt. Had he gotten 4 yards on that read-option play, the drive is still going.
Obviously this is an example that requires lots of very specific things to happen. But that’s what statistics really are: just a collection of averages that describe to us what happened.
And in real life this is how you explain Nick Fitzgerald and Mississippi State.
Fitzgerald put up a throwing stat line last season very similar to another QB that Gators fans should be pretty familiar with.
Statistical passing comparison of Feleipe Franks with Nick Fitzgerald in 2017. (Will Miles/Read and Reaction)
In every practical sense, those are identical performances. The only difference is that Fitzgerald played more and so attempted more passes. These were both really ineffective performances through the air.
The difference is that Fitzgerald also ran for 984 yards on 162 carries (6.1 yards per rush). That’s the difference between the two offenses. Mississippi State averaged 32.9 points per game in games where Fitzgerald threw more than 10 passes. Florida averaged 22.6 in games where Franks did the same.
How do we measure this?
Clearly Fitzgerald’s running ability is important. But so is the lack of consistency in the passing game. It’s not a coincidence that the Mississippi State offense averaged 35.7 points per game under Prescott the two previous seasons and then “only” 31.3 under Fitzgerald.
So how do we account for each player’s abilities? Well, what I’ve attempted to do is quantify how much better or worse than average a QB is based on his averages both running and passing the ball. I acknowledge that this doesn’t take into account turnovers or completion percentage. But what it does do is measure exactly how efficient a QB is on any given play.
If he runs for 4.07 yards per rush, he’s exactly one yard better than his peers on that type of play. If he throws for 6.41 yards per pass attempt, he’s exactly one yard worse than his peers on that type of play. If he were to have those averages for 100 passes and 100 runs, he would be exactly average as the positive from the rushing would cancel out the negative from the passing.
I’m calling this metric “Yards above Replacement (YAR)”, as it measures exactly how much better a QB is on a per-play basis than his peers. Because this is a rate statistic, it is impacted by sample sizes. A QB with three rushes – one of them for 90 yards – will obviously rate very high using this statistic. That’s why I’m also quantifying “Total Yards above Replacement (TYAR)” as well, as this just adds up the total value from each type of play.
Yards above replacement (YAR) versus points per game for Power-5 QBs in 2017. (Will Miles/Read and Reaction)
The above chart shows that the metric correlates well with scoring on a point per game basis. This was plotted for the QB from each Power-5 team who played the most in any game that he threw more than 10 passes.
Obviously, there’s quite a bit of scatter in the data. But out of 31 QBs who had negative YAR, only three led offenses that averaged more than 30 points per game. Interestingly, I do think those three players show where this metric can be improved. Ryan Finley (NC State), Kelly Bryant (Clemson) and Luke Falk (Washington State) all have a negative YAR, but they all have completion percentages over 65 percent.
I’ll certainly be working to include that into future versions. But this version can tell us some really important things about the differences between QBs last season and what to expect next season.
Yards above replacement for Feleipe Franks and Nick Fitzgerald in 2017. (Will Miles/Read and Reaction)
Comparing the 2017 seasons of Fitzgerald and Franks using YAR shows Franks as having negative contributions through the air (6.4 yards per attempt) and on the ground (0.3 yards per rush). Conversely, Fitzgerald has negative contributions through the air (6.2 yards per attempt) but positive on the ground (6.1 yards per rush). Thus, Fitzgerald has a positive YAR and TYAR while Franks is significantly negative.
A YAR of 0.32 may not sound like that high, but that ranks Fitzgerald 21st out of the 65 main QBs for the teams from Power-5 conferences. Conversely, Franks ranks 58th. If we look at TYAR, Fitzgerald ranks 22nd and Franks ranks 54th, while the top-5 were Baker Mayfield (Oklahoma), Lamar Jackson (Louisville), Khalil Tate (Arizona), Mason Rudolph (Oklahoma State) and Drew Lock (Missouri). Not coincidently, none of those teams averaged less than 37.5 points per game when those players threw more than 10 passes in a game.
Based on the equation shown in the chart above, every 0.1 YAR is worth 0.7 points per game. So in the case of Franks, Florida would have been expected to score 20.5 points per game in his games. The Gators actually scored 22.5, but that’s with the three pick-sixes against Michigan and Tennessee included. In Fitzgerald’s case, Mississippi State would have been expected to score 32.2 points per game in his games and the Bulldogs actually scored 32.9.
Projecting Florida’s offensive performance for 2018
I wrote last year for SEC Country that really bad offenses show huge jumps in performance for two reasons: a new QB transfers in or there is a change at offensive coordinator. Absent that, offenses typically show much more modest improvement.
Florida has the new offensive coordinator in head coach Dan Mullen, and Mullen certainly has a very different offense than his predecessor. The QB run will be featured in some capacity, but that doesn’t directly make the offense better. As shown above, if the QB averages less than 3.07 yards per rush attempt, he actually harms the offensive performance.
This is where Kyle Trask – who has looked like the best prepared to run the offense through the air – likely comes up somewhat short. Trask is not a statue behind the line of scrimmage, but he is limited. To expect him to have a positive effect in the running game is unlikely, meaning he’ll have to be significantly above average through the air to get the Gators above 30 points per game.
Feleipe Franks does have more mobility than Trask. His 79-yard run against Texas A&M showed he can break tackles and has deceptive speed. But his tape also includes him repeatedly running out of bounds to take a sack rather than throwing the ball away as well. Those plays count too.
But what we can do is see where Franks needs to improve to get Florida over that 30 point per game barrier.
Improvement needed by Feleipe Franks to improve the Gators offense significantly in 2018. (Will Miles/Read and Reaction)
If Mullen can get Franks to improve through the air at the same percentages as Dak Prescott improved from 2013 to 2014 with no improvement in the running game (scenario 1), Florida’s offense would improve 7.9 points per game. That would have ranked 65th last season.
If Mullen can’t improve Franks through the air at all, but can increase his carries to 100 with a 4.0 yard per carry average (scenario 2), Florida offense would produce 26.6 points per game (74th in 2017).
For Florida to get over 30 points per game, Franks is going to have to improve through the air approximately as much as scenario 1 while improving on the ground as much as scenario 2. I just don’t think that’s very likely, which means Florida’s offense under Franks will show modest improvement but will continue to struggle.
So what about Emory Jones? Well, I have written previously about what QBs with a similar recruiting profile typically produce in a season. Not a ton of them start as true freshmen, but those who do average 6.8 yards per pass attempt.
If we were to assume Jones would match that on the same number of throws as Franks in 2017, he’d need to make up the difference with his legs. But 6.8 yards per attempt is much higher than Franks achieved in 2017 (6.4) and so Jones would only need to average 4.5 yards per rush on 100 carries to achieve a predicted point total of 30.
This might actually be possible for Jones to achieve. He averaged 6.9 yards per rush on 129 carries his senior year in high school. Obviously the SEC isn’t Heard County High School, but the fact is that of the players being considered at QB, Jones is the only one who has any significant track record of being able to run the ball. If you don’t believe Franks or Trask will be able to throw the ball effectively in 2018, then Jones may be the best option.
Of course, this doesn’t exclude using multiple QBs, which is probably the best option. Kadarius Toney can throw enough that he should be able to keep defenses honest. Even when he was essentially zero threat to throw in Nussmeier’s offense, Toney averaged 8.6 yards per rush on 14 carries in 2017. The implementation of a Toney or Jones package would bring some of the value of a running QB while keeping the best throwing QB in there for the majority of the game.
Takeaway
This analysis shows why the fit of the QB within an offense is so important. If Mullen just runs Trask into the line of scrimmage repeatedly because that is his scheme, it can actually have a negative impact on the offense’s productivity. Conversely, he may be able to get away with less through the air if his QB can exploit his offense to provide explosive plays through the running game.
But this analysis also shows the climb that the Florida offense has to make in 2018. The Gators aren’t coming off of an average performance from Franks. It was in the bottom 10 percent of the Power-5. Modest improvement isn’t going to cut it based on the expectations of Florida fans to score points, and quite frankly, the expectations that Mullen has expressed as well.
The more I look at this, the more I think that the offense needs to open up either with Jones as the starter or with Trask as the starter with major playing time at QB given to either Emory Jones or Kadarius Toney where they are a threat to run and pass.
I also suspect that Mullen repeatedly telling the media that how his QBs progress between spring and fall with determine who starts is code to Jones that the job is his to take. Mullen isn’t stupid. He understands the value that Jones’ running ability will bring to the Gators. But if he isn’t going to be serviceable as a passer, then he can’t be the starter.
So we’re back to where we’ve been for what seems like a decade. There is significant uncertainty at QB that probably won’t be resolved until at least after the game against Tennessee in Knoxville. But there’s a pretty clear path that points towards Jones in Mullen’s offense.
And that’s probably the only path that gets the Gators to 30 points per game.