Yet again, Kadarius Toney should be the starting QB for the Florida Gators in 2018

Embed from Getty Images

Last offseason, I wrote that Kadarius Toney should get a serious look as the starting QB in 2017. Fans scoffed.

I then wrote that if Toney wasn’t going to get a fair look, then Luke Del Rio should get another shot. I was threatened on Twitter (in jest, not a Jim McElwain “threat”).

The reason I stumped for both of those players so hard was because of an article I read about why the Florida Marlins signed a decent third baseman (Martin Prado) to a 3-year, $40 million contract at the exact same time the best general manager in the game – Theo Epstein – was signed to a 5-year, $50 million deal.

Epstein has built World Series winners in Boston and Chicago. Prado has averaged around 3 wins above replacement (WAR) per 162 games. How is it that Epstein is worth less money?

The author posited that the reason Epstein is worth less is because many of his skills can be taught while you can’t teach someone to hit a 95 mph fastball. In other words, Prado is more valuable because he has rare inherent skills.

In this case, Epstein is paid based on what the market dictates. But I firmly believe that this is an error in logic and analysis of the people setting the market. We can see Prado’s value immediately because we can see him hit a home run against an elite pitcher and his statistics are flashed on the scoreboard, updated with every at bat.

Conversely, it is much more difficult to quantify Epstein’s contributions. Yes, he has built World Series winners for teams with extended track records of being unable to do so (86 years for Boston, 108 years for Chicago). But is that all attributable to him? Is he a product of the people that he puts around him? And could you hire one of those people instead at a reduced price?

But just because it is difficult to measure doesn’t mean that we should avoid trying to assess Epstein’s contributions. I’m sure the Cubs and Red Sox value his contributions. And Prado’s teams have yet to win their division.

Application to QB Analysis

So what does this have to do with the Gators next quarterback?

Well, the first thing most people look at when analyzing a QB is size and arm strength. As a Buffalo Bills fan, I watched in horror in 1999 as Wade Phillips decided to start Rob Johnson over Doug Flutie in the playoff game against Tennessee that became known for the Music City Miracle.

Flutie wasn’t a great QB, but he had led Buffalo to a 10-5 record during the regular season. He was benched under the auspices of giving him rest in the season finale. But he was really benched because he didn’t have elite arm strength and he was small (5’10” 180 lbs) while Johnson had a better arm and more prototypical size (6’4” 205 lbs).

Johnson was a spectacular flop in Buffalo, including completing 45 percent of his passes in that playoff game against Tennessee. And ever since then I’ve wondered, while arm strength and size are easy to measure and rare, do they actually correlate with winning QB play?

It is easy to see a cannon for an arm and prototypical size and assume that you can coach the more “nuanced” intricacies of the position. But just because that’s the assumption, is it actually true?

This is relevant to Florida fans because QB Feleipe Franks has prototypical size at 6’5” and 220 lbs. As he illustrated at the end of the Tennessee game this year, he can launch the ball as far as is needed. He’s pretty mobile for a man his size as well. He has everything you would look for in a QB.

Except that he has a track record of being inaccurate.

Maybe you think Franks proved he doesn’t have what it takes in 2017. You may be right. But now Dan Mullen is bringing in 4-star commit Emory Jones, a QB who appears to fit his system really well and stands 6’3” 195 lbs. The problem is that the stats I can find on Jones indicate he’s just slightly more accurate than Franks.

And so that begs two questions. First, does accuracy correlate to excellent quarterback play? Second, are there ways to predict how accurate a QB will be?

Does accuracy correlate to QB excellence?

To examine this, I sorted all FBS QBs by completion percentage and then examined the completion percentage, QB rating and TD to INT ratio of QBs who ranked 1-10, 41-50, 71-80 and 101-110 for both 2016 and 2017. If accuracy truly tracks with completion percentage, we should see a significant uptick in QB rating and TD to INT ratio with more accurate QBs.

And that is exactly what we see. QBs who are more accurate have higher QB ratings and a much higher TD to INT ratio.

Franks season in 2017 is a great example. He completed 54.4 percent of his passes and had a 1:1 TD:INT ratio. Luke Del Rio has a career 56.6 completion percentage and a 1:1 TD:INT ratio as well. In 2012 Jeff Driskel completed 63.7 percent of his passes and had a 2.5 TD:INT ratio. In 2014, Driskel’s completion percentage dropped to 53.8 percent and his TD:INT ratio fell to 0.9.

The past nine QBs to win the Heisman Trophy have averaged a completion percentage of 67.1. The average QB rating and TD:INT ratios for those players is 177.7 and 5.7, respectively.

Previous work I’ve done indicates that elite QBs do get more out of each attempt. But it also indicates that there is very little difference in yards per attempt for QBs ranked 30-95 in QB rating. Instead, the key differentiator for those QBs is completion percentage, i.e. accuracy.

Can accuracy be predicted?

So if we believe that accuracy is the inherent rare skill that correlates with performance, that leads us to our second question. Is there a way to predict how accurate a QB will be?

Last offseason, I looked at a relatively small subset of QBs and their high school records. The conclusion was that accuracy in college tracks pretty reliably from high school.

There were two issues with that examination. The first was that the sample size was extremely limited (13 QBs) due to time constraints and the difficulty of finding complete high school statistics. The second was that the QBs I examined were players I was curious about (Sam Darnold, Tim Tebow) rather than selecting the samples more randomly.

I’ve now expanded the study to include 50 QBs, ranging from highly successful QBs (Deshaun Watson) to less successful QBs (Wes Lunt). The comparison that was made was the senior year of high school for each player compared to their career completion percentage in college. I did limit examination to QBs of Power-5 programs.

The first thing that I noticed was that the completion percentage of the 50 QBs in high school was 62.0 percent and the completion percentage of those same QBs in college was 62.1. That’s pretty remarkable.

Just as you would expect, there was some variation and were some outliers. Florida QB Luke Del Rio completed 70.1 percent of his passes in high school but only 56.6 percent in college. Conversely, Washington State QB Luke Falk completed 58.7 percent of his passes in high school but has completed 68.3 percent of his passes in college.

But if we look at a histogram of the data, what we see is that more than a quarter of the QBs see a 0-3 percent increase moving from high school to college. More than three quarters fall within plus or minus 6 percentage points.

The takeaway is clear. If you have a QB who is accurate in high school, chances are pretty high that they are going to be accurate in college. If you have a QB who was inaccurate in high school, chances are high that they will be inaccurate in college.

This doesn’t mean that it is a perfect correlation. There are outliers. But I’m willing to bet that many of the outliers have causes not related to a QBs ability to be accurate.

For example, Mississippi State’s Nick Fitzgerald had a high school completion percentage of 44.6 percent but that was only for 74 throws. That was likely because Fitzgerald’s offense was a triple-option scheme. It shouldn’t be a surprise that in Dan Mullen’s offense Fitzgerald has improved his accuracy to 55.4 percent.

Application to Florida’s QB situation

If accuracy is important and correlates from high school to college, then that means that it is a characteristic – just like arm strength, height, weight and speed – that coaches need to pay attention to when they recruit. Based on the data presented above, I would submit that it is the most important characteristic that coaches should look at.

It also means that we should be able to look at the completion percentages in high school of the players on Florida’s roster and draw some conclusions.

The first thing to notice is that new Florida commit Emory Jones’ profile is better than Feleipe Franks. He’s more accurate and that has generated more yards per attempt, albeit on 87 less throws. Jones also fits Mullen’s offense well and has had to throw on the run a lot in high school. But to expect him to come in and complete 70 percent of his passes just isn’t realistic.

The second thing to notice is that Franks was exactly who he should have been in 2017. He completed 54.4 percent of his passes, well within what we would expect from about 75 percent of QBs with his profile.

Toney and Trask have exemplary completion percentages with similar yards per attempt to both Jones and Franks. The difference is that Toney has the second largest sample size on the board while Trask has a limited sample size and played a lot as a backup.

Jake Allen did lead St. Thomas Aquinas to multiple state championships. But he did so putting up accuracy numbers on-par with Franks and Jones, doesn’t have the running ability of Jones or Toney and completed much shorter passes than any of the others.

I still think Jones will be the starter to open the season, and that may not be the wrong call. But I do worry about his ability to be accurate and whether that is a skill that can improve over time.

If you look at QBs with similar high school profiles (less than 62 percent completion percentage, more than 10 yards per attempt), only 6 of the 50 QBs examined fit that profile: Aaron Murray, Jared Goff, Brad Kaaya, Dak Prescott, Brett Hundley and Jake Locker.

If accuracy can be coached, then we should see a significant improvement from the first full year to the second full year of playing. But that’s not what we really see.

We see a relatively modest improvement of 2.1 in completion percentage and 0.2 in yards per attempt. The one outlier is Locker, who improved by 6.5 percentage points from 47.3 to 53.8. Interestingly, Locker had a completion percentage of 58.8 in high school. Based on this analysis, I would have predicted a significant increase his sophomore year based on his freshman season.

All of the other QBs first years in college were within 4.7 of their high school completion percentage and so the 2.1 average improvement in completion percentage tells us that this doesn’t appear to be a skill that varies based on coaching (with the caveat that this is a really small sample size). Instead, it appears to be an inherent skill that probably has to be recruited.

Takeaways

This analysis means that anyone stumping for Feleipe Franks is doing so based purely on their faith that Dan Mullen can make him into someone that he hasn’t been thus far. For sure, I do think Franks will improve, perhaps significantly.

But in a best case scenario let’s say he gets Locker’s bump in completion percentage and Prescott’s increase in yards per attempt with the same number of pass attempts and an improvement to 10 TDs and 5 INTs. That would put him at a QB rating of 135.7. That’s just slightly better than Austin Appleby’s QB rating of 128.0 in 2016.

This analysis also means that I believe that Kadarius Toney is the best candidate for the starting QB job once again. He has a completion percentage in high school that would project him to put up a completion percentage above 62 percent in a worst-case scenario. He also has the running ability to execute Mullen’s offense.

When Emory Jones committed, we discussed whether on the Gators Breakdown early signing day podcast whether Mullen might use Toney in the 2006 Tim Tebow role. That season, Tebow came in for starter Chris Leak and ran the wildcat formation whenever the Gators needed a yard. Certainly Tebow used brute force to get the first down while Toney would use quickness, but it was a logical leap to make that Toney could play that role with Jones as the starter.

But looking at these numbers, I wonder whether I should have thought about it in exactly the opposite way. Should Jones play the Tebow role while Toney is the starting QB?

This actually would make sense based on the needs of the roles. Jones is the larger man, more capable of picking up those short-yardage opportunities. Toney is smaller, and would likely be more susceptible to injuries in Mullen’s offense. Mullen would be able to protect him from runs up the middle while utilizing his ability to get outside to make explosive plays.

Maybe I’m wrong. I’ve certainly been an avowed Toney Truther ever since I watched his tape last summer. But part of the reason that I’ve been so convinced Toney could play QB is because of these numbers. All this reanalysis has done is increased my conviction that completion percentage is a way more important metric than arm strength or height.

If you take that same list of 50 QBs and look for a high school completion percentage greater than 65 percent and a yards per attempt greater than 10, you’re left with 9 names: Matthew Stafford, Will Grier, Mitchell Trubisky, Jake Browning, Charlie Brewer, Justin Herbert, Kenny Hill, Hutson Mason and Wes Lunt.

Those players averaged a completion percentage of 63.9 and a QB rating of 145.6 in their college careers. That QB rating would have ranked 35th in the country this season. The last time a Florida QB put up a QB rating that high with enough attempts to qualify for a ranking was Tim Tebow in 2009 and remember, Toney is on the high end of both of those qualifying metric (70.6%, 11.5 yds/att).

In no way am I suggesting that Toney will be Tebow. I actually don’t think he’ll get a very long look at QB. If you pay any attention to Florida’s beat writers, they focused on his inaccuracy on screen passes and how his passes didn’t look very pretty in practice during camp last season.

It reminded me of all the criticisms of Flutie in 1999. All of the experts – including those on Buffalo’s coaching staff and front office – focused on what he was not rather than what he was.

In Toney’s case, he’s small, has limited arm strength and doesn’t throw a particularly pretty ball. But instead let’s look at what he is. He’s accurate. He’s able to break outside the pocket and make plays. He’s explosive in space. And if you look at the film, he shows a pocket presence that was missing regardless of which QB was in the backfield for Florida this past year.

Toney is what he is, which means he is the best candidate to be the starting QB in 2018.

FEATURED IMAGE USED UNDER CREATIVE COMMONS LICENSE FROM arctic whirlwind

High school stats collated from max preps unless otherwise noted. college stats collated from sports reference cfb