Through 27 games, I was able to track the passing of the “Big Four” quarterbacks: Teddy Bridgewater, Johnny Manziel, Blake Bortles and Derek Carr—the presumed top four quarterbacks that have a reasonable chance of going in the first round.
Games tracked for Bridgewater: Ohio, Eastern Kentucky, Kentucky, Rutgers, South Florida, Houston, Miami (Bowl)
Games tracked for Manziel: Alabama, Arkansas, Missouri, Louisiana State, Auburn, Vanderbilt, Ole Miss
Games tracked for Bortles: Akron, Penn State, South Carolina, Louisville, Connecticut and Baylor (Bowl)
Games tracked for Carr: Rutgers, Boise State, Cal Poly, San Diego State, Utah State, San Jose State, Southern California (Bowl)
For this game tracking, I chose not to look at completion percentage, “true touchdowns,” or “true interceptions” but simply an evaluation of the ball placement. If the quarterback reasonably puts the ball in place for the receiver to get it, regardless of the actual catch, it’s counted as an accurate throw.
This will mean there are no penalties for drops, but there are penalties for spectacular catches where the receiver bails out the quarterback for poor ball placement. It also penalizes all dropped interceptions and touchdowns, while eliminating throws that could not reasonably be blamed on the quarterback in those situations.
Any drive resulting in a touchdown did not get any additional touchdown credit, even if a receiver dropped the ball in the end zone earlier in the drive. All distances are “true” insofar as they are recorded where the ball was thrown, not the final yardage gained on the play, which also means all throws in the end zone have additional yards tacked on based on where in the end zone the ball was thrown (a throw from the 15-yard line eight yards into the end zone would be recorded as a 23-yard pass).
Only passes that were “aimed” were included, which means no throwaways or receiver/quarterback miscommunications. A few penalties that made it impossible to determine ball placement were excluded as well (usually egregious holding).
Each box of four includes “completions,” aimed passes, “true touchdowns,” and “true interceptions” based on ball placement as indicated:
There are a number of ways to track the data, all of which provide different contexts. I’ll attempt to walk through a few of them and see if there are any useful conclusions to draw.
The first way to evaluate the data is to look at a simple spread of the throws by a two-dimensional look at a field:
That can be useful, but is a little hard to read, so breaking it up by individual quarterback can help as well:
It might also be useful to granularize the depth of the throws, so a one-dimensional chart only looking at distance of throws (in 5-yard bins) will help, too:
While all of those charts can be useful, it’s difficult to wrap one’s head around the different numbers, especially given the different nature of the offenses. One can even out the attempts (say, 250) and match it to the distribution of passes from 2012-2013 in the NFL, to project to a generic pro-style offense:
This is a little easier to look at and figure out. In a traditional passing distribution, Bridgewater is quite a ways ahead of the other quarterbacks, and Carr’s gaudy passing totals (he passed twice as often as others in the same amount of games) are adjusted to numbers that make more sense relative to each other. Here, Bortles’ high accuracy is ameliorated by the low touchdown total and average interception total, and Carr’s obscenely low interception rate can be put into context.
Both Carr and Manziel had their accuracy rating reduced from their original totals because the high number of screen passes and passes below 4 yards of depth were washed out, while Bortles’ passing distribution helps him.
You can use this to project passer rating and the even better metrics yards per attempt and adjusted yards per attempt onto the quarterbacks.
Bridgewater: 112.8 passer rating, 8.8 yards per attempt, 8.7 adjusted yards per attempt
Bortles: 90.3 passer rating, 8.0 yards per attempt, 7.6 adjusted yards per attempt
Carr: 106.6 passer rating, 7.6 yards per attempt, 8.3 adjusted yards per attempt
Manziel: 93.8 passer rating, 7.5 yards per attempt, 7.1 adjusted yards per attempt
You can also project it onto the Air Coryell offenses the NFL saw in the last two years:
Not much changes, given that the heavier emphasis on deeper passing hurts all of the quarterbacks, and the small difference between “traditional” pro-style and the Coryell style of offense means there aren’t dramatic changes. In that “system,” the numbers project as follows:
Bridgewater: 113.2 passer rating, 9.0 yards per attempt, 8.9 adjusted yards per attempt
Bortles: 97.7 passer rating, 8.2 yards per attempt, 7.8 adjusted yards per attempt
Carr: 106.9 passer rating, 7.8 yards per attempt, 8.5 adjusted yards per attempt
Manziel: 95.0 passer rating, 7.7 yards per attempt, 7.4 adjusted yards per attempt
Naturally, one of the biggest issues with all of this data is that there’s too much information to filter and it’s fairly difficult to process. There are easier ways to represent some of the data. The first of which is to reduce them to the three simple rate stats described above, but there are other ways to look at it, too. One can look at differential completion percentage by two-dimensional distance:
Or, one can look at one dimension distance:
The distance chart is not very useful, however, as it’s subject to highly random spikes given small sample sets at the far distances (Manziel between 30-34, Bridgewater and Bortles between 35-39, etc.), but it can still provide a small amount of context.
The most important thing is not to read too much into the data. Naturally, any rate statistics for the quarterbacks can be subject to the strength of opposing defenses, the skill of his receivers to get open and demand targets, the nature of the offense they were playing in (both Manziel and Carr play in spread variants that emphasize yards-after-the-catch and a short passing game, which is why their completion rates are low across the board despite highly effective offenses), etc.
But it does provide some context for the claims we make about quarterbacks. It doesn’t mean many of these claims are wrong but that they’re subject to a lot more explanation when they disagree with relevant data. For example, Bortles has a reputation for being a generally inaccurate thrower compared to the other Big Four quarterbacks—but has a better accuracy rating than two of the other ones.
It is not necessarily that the reputation is inaccurate; the UCF quarterback could have been working with larger windows or produced accurate passes in unsustainable ways (e.g. placed behind the receiver or after unnecessary extra adjustment). In this specific case, I do think that Bortles has underrated accuracy, but that there is something to be said about those other explanations for his reputation. Merely hitting one’s receiver is not enough.
The data can also be used to ask questions that reveal something about the quarterbacks’ games. Carr already had a low interception rate before game tracking (1.2%, 5th-best among all college QBs with 300 passes, below Marcus Mariota, Teddy Bridgewater, Cody Fajardo and Bryce Petty), but that could have been easily explained by the high percentage of screen passes in his game.
It doesn’t seem that simple, however, after adjusting all of his attempts for distance, where he massively outperforms his contemporaries in the same system. For Carr, it’s difficult to explain this away due to variance; he threw twice as many passes at intermediate distances than the other quarterbacks despite his system simply because he passed more. A different explanation could posit that Carr plays a little too safe, although that feels odd considering his willingness to audible and throw deep.
It’s true in a sense, however. Carr doesn’t necessarily avoid risky throws when the play calls for it, but he is particularly sure to place the ball in exclusive place for the receiver, where a defender can’t get to it. Sometimes, he’s a bit too attentive to this and it’s difficult for the receiver to get there as well and that’s a problem at times.
Another explanation is that because the defense is a bit too keyed in to the screen game, they aren’t necessarily in a position to punish him for lazy ball placement. The second explanation is a little true, but largely unfounded. For the most part, his low interception rate has to do with his tendency to keep the ball away from contested catches, which is why his deep misses are almost always overthrows.
Regardless, the data can be immensely useful so long as it isn’t used to replace judgment. The fact that distance adjusted passer rating is possible doesn’t mean that it is an effective way to rank the quarterbacks, but the data can allow us to test our assumptions, or force us to explain in detail why they still may be true. It can also lead to new questions which reveal something new about the prospects as well.
View Comments
What would happen to the data if you put the USC game in for Carr? I know there is a talent mismatch there but it is also the only defense that could be considered even just above average that he played against. Your analysis and fit projections are great for Carr. I worry that his only game against good comp was poor. I value your analysis on this issue because you breakdown the film & numbers.
Is that game a huge "I get rattled under pressure" ponder like Red flag or would you take this guy if you were RS in 1st round somewhere.
Arif did use the USC game for Carr. Look above. USC is Southern California.
Thanks for pointing that out and not crushing me. Still wonder where Arifs study of Carr and that game in particular have him feeling about Carr and his value if you have to take him somewhere in the 1st compared to QBs he has below him in his ratings in the 2nd - 4th rds. I like Carr in a good pocket but worry about him playing on a Vikes team with bad D that forces you to have to score 30 to win against teams in full pass rush mode. That was Ponders big problem. He felt pressure that wasn't there and reacted poorly often.
Also if we are talking about big games. Look at Carr's 2012 bowl game against SMU. A terrible blowout 43-10. Carr is scared in the pocket under pressure. He takes his eyes off of down field and runs into the back of his lineman. His right tackle did not help him much either.
Two bowl game flops for Carr, says he has a tough time stepping up in big games. Now take Bortles and his bowl game win this year. He had the IT factor, for at least one game.
Carr's best tape for me is in 2011 when he played in a more pro-style offense. He is making good down field throws. The problem is, has he progressed or recessed since then? It is hard to tell with the spread offense he has been in the last two years.
Bowl games do a terrible job at projecting. Rodgers, Manning and Brees all did terribly in their bowl games and Brady did awfully until the final quarter and overtime.
I think if you break Carr's tape down throw by throw, you can account for the type of offense he was in, and I would say he plays the game intelligently within the constraints of his offense. In that sense, he's progressed.
I like Carr's accuracy, arm strength, and leadership. I would be more than happy if the Vikings were able to trade back a few spots to land him.
When I watch tape on Carr, he can sling it as well as anyone. when it comes to:
Quickness of release
Ball placement
Velocity on throws
However, he does get away with throwing leaning backward or falling away from pressure.
I would love to see him plant that right foot and drive his body into his throws all the time. Stand in the pocket just a bit longer and not be afraid to step into throw and take the hit.
I think he is a tough kid that can handle pressure. He has great speed and can run, but he is a pocket first QB who can get the job done in the air.
It is unfair to compare Carr to Ponder. Ponder is not a quick read, get rid of the ball pocket QB.
Carr knows what he wants to do with the ball when he is at the line!
I do not see the dear in the headlights look with Carr.
Take that last comment with a grain of salt, because I like what Ponder can do out side of the pocket if he has a nice slot and TE to work with down field.
As WTFVikes fan points out, USC is there.
But I wanted to get into his actual performance against USC--of his 56 aimed passes in that game, only two were attributed as completions when they were failures from a ball placement perspective. *Twelve* passes were attributed as incompletions when they were success from a ball placement perspective. Most of those were drops. It was the least friendly receiver corps I've ever tracked (Carr was missing his top receiver in that game).
Despite his completion rate of 31/55 (getting rid of one throwaway and adding one or two penalties) at around 56%, his true accuracy was 43/55 (78%), which is higher than his career accuracy.
Over a third of passes were made under duress.
For the most part, I would say USC was a good performance with a bad statline. Personally, I would take him.
I actually just watched that game again and what bothered me most wasn't the amount of missed throws Carr had, but the moments he had them. Would I be correct in saying several of those incompletions were on plays that should have been big gainers or TD's? Carr's line play was pretty terrible and the play calling didn't help him at times against USC, but it really bugged me that when he had some opportunities he missed on them. I haven't really watched any other games of his so that has really left a negative impression in my mind. It sounded like those blatent misses were uncharacteristic for him and these stats would seem to support that.
Couple of points I have come up with in watching Carr in virtually every game he plsyed in 2013.
First off, its kind of a misconception that he played against all weak D's. He actaull faced higher ranked pass D's then Manziel and Bridgewater.
In the USC game, the Fresno State OC (Schramm) really is to blame......I know unlike UL, A&M, and UCF, Fresno State had no rushing game, which allowed teams to key on the pass that much more, but in the USC game Schramm only ran the Ball 6 times.....None in the first Quarter....This allowed a USC D to pin thier ears back and focus on the QB.......Even a great QB is going to struggle when that happens.....See Payton Manning in the Super Bowl for example....
Carr was playing in a one dimensional O, unlike the other noted QB's he really did not have the luxary of a solid rushing game to help his pass game. D cords that were Facing Fresno new that Fresno State was not going to beat you with the rush.....So they were 100% focused on the pass game.....It amazing Carr was able to do what he did when everyone new what was coming....
IMHO Carr is the BEST overall QB in this years draft, to support my point look at the following. (Keep in mind Carr faced this team when their was 100% Healthy, Bortles and Bridgewater got Rutgers afer they were down multiple starters to injury.)
We actually had the unique opportunity to evaluate these quarterbacks on a somewhat level playing field. Bortles, Bridgewater and Carr all played Rutgers at home this season.
Here is how the three QB’s did:
Bridgewater – 21-31 (67.7%) 310 yds, 2 TD’s, 1 INT
7 carries for 11 yards
Bortles – 21-30 (70.0%) 335 yds, 1 TD, 0 INT
13 carries for 32 yards and 1 TD
Carr – 52-73 (71.2%) 456 yds, 5 TD’s, 1 INT
4 carries for 24 yards
Great comment, and welcome to the VT website MR. T
I would also like to welcome all of the newbees popping up here!
How do these stats compare to a similar analysis of other qb's college careers? (Luck, Big Ben, Ponder, Brady, David Carr?). That would help show the significance of the data analysis.
I wish I could tell you, but I don't have the resources to match that effort. That's why I don't go into the business of projecting here; without historical data to compare it against, it's not useful as a projection tool, just an analysis tool.
Arif,
Great data study, but are you telling us Manziel is one of your top 4 QB? Are you telling us you have a 1st round grade on him?
Ron Jaworski and Phil Simms don't seem think he is first round NFL talent.
Analysts are either high or low on Johnny Manziel. What is your real take on him when you watch him play the game.
No, those are the consensus top four quarterbacks. Personally, I think Manziel is a high-second round talent, but that's not where he'll go.
This is really good. Thanks for doing it. Given that as many as three of these guys will be off the board when the Vikings pick, I think it would be useful to extend analyses to the next tier of prospects. This is a labor-intensive project, so I understand why you've limited this to these four. But in general, I'm interested in these four plus at least McCarron and Garoppolo.
I will expand it to include Brett Smith, Jimmy Garoppolo, A.J. McCarron, Aaron Murray, David Fales and Zach Mettenberger.
If you've seen Arif's other works, this is not a labor intensive one. It required legwork, but it's still shorter than LOTR.
I really do not know what to say about all these graph's and stuff. Can you simple it down with some stick figures drawn with crayons?
I want to know what the rock, river, skip scores are on these guys and which ones can just "wing" it. Just get Norv a piece of clay to mold.
Interestingly enough, johnny, they all scored a 12.5, so nothing there to help us make a choice. It should be noted that Manziel showed up late for the test, but it was recorded anyway. Do you know, have these guys taken their Wonderlich test yet?
Coach, I think that they take it during the combine.
wonderlich? was fran's old playboy girlfriend at the combine???
Ha!
Would you penalize young Johnny for that Coach? What if he had a good reason?
The list of good reasons for being late to your NFL rock skipping test is very, very short, cart. But I'd go with, "That won't happen again, will it son." ("No sir" is the correct response here) "Oh, and by the way, that wasn't a question."
arif, idk how you support yourself, but i hope the VT stock options adam gave you can be exercised soon so you can justify spending your time on this kind of thing, i think it's pretty cool. i have a headache now, but it's still cool
maybe this is misguided for something i missed, but i focused on that last chart, and the 5-9 and 10-14 boxes, because i would think that's where a QB's most value might be. wouldn't that be the area where you'd throw for firsts downs, and have a lot of third and longs where the team often has to rely on the QB? bridgewater stood out there, but he'll be gone by pick 8. when we see the second tier QB article, then i'd also look for a guy who does well there
all this being said, i still wouldn't be surprised to see someone high up fall to us
Gruden seems to like Johnny Football.
http://espn.go.com/videohub/video/clip?id=10527644&categoryid=2378529
Carl, I find it hard to determine a QB, or any other player, Gruden doesn't like publicly.
You know, I don't really like any of these guys.
I like Jimmy Garoppolo.