PDA

View Full Version : Luck, Griffin, Tannehill, Weeden - A Metrics Breakdown



NUGap
03-07-2012, 09:24 PM
What I wanted to do was breakdown some stats and give a numerical backing to some of the things you see when you're watching the tape of these top 4 quarterbacks. The numbers are definitely meant to compliment what you see, not replace it. Essentially what I've done here is take 4 games each from Luck, Griffin, Weeden, and Tannehill, and chart every throw they made within those games. I noted what formation they threw the ball from, the outcome (did the receiver catch it, did he drop it, was it a fluke interception..etc), and the distance of the throw.

It's important to note that in order to limit this to the quarterback's ability, I noted only where the receiver caught the ball. I'm going to give some charts and then a little commentary.



Overall Comp%







Griffin
Luck
Tannehill
Weeden


Total
Comp%
Comp%
Comp%
Comp%


Screen
85.71
88.89
93.55
90.48


1-5 Yards
75.00
85.00
64.29
75.56


6-10 Yards
60.00
70.83
57.14
64.71


11-20 Yards
68.00
46.15
55.88
55.56


20 + Yards
60.00
41.67
31.82
52.38


Total
69.17
68.47
62.00
69.52




These are obviously the completion percentages for each quarterback and the various depths I assigned. Again, these depths are only where the QB threw the ball and the WR caught it.



%TotalThrows







Griffin
Luck
Tannehill
Weeden



%Throw
%Throw
%Throw
%Throw


Screen
15.79
8.11
20.67
22.46


1-5 Yards
24.06
36.04
18.67
24.06


6-10 Yards
22.56
21.62
23.33
18.18


11-20 Yards
18.80
23.42
22.67
24.06


20 + Yards
18.80
10.81
14.67
11.23




These are how often each QB threw to each zone, thus Weeden threw screens 22.46% of the time and Luck threw 1-5 yards 36% of the time

Here are some of my thoughts and a little summary of this data:

Andrew Luck:


Only 8.1% of Luck's throws were screens, but an unprecedented 36% of Luck's passes were 1-5 yards from the Line of Scrimmage. The QBs, on average, threw 25.1% of their passes in the 1-5yd range
While he threw a ridiculous amount of passes in 1-5yds, he also completed them at an 85% rate, better than any percentage at any position or depth, except for screens
Out of the 4 passers, Luck went deep the least, only opting to throw past 20 yards 10% of the time, for a comparison RGIII threw deep 19% of the time

Griffin:


The thing that sticks out at me for RGIII is the fact that he throws the various distances pretty evenly. the range of how often he throws the different depths is between 16% (screens) and 24% (1-5yds)
We all know the lore about his deep ball, but here's some numbers to back up the ridiculousness. He throws the deep ball (20+) 18.8% of the time and completes it 60% of the time. Luck only completes the deep ball 42% and Weeden 52%
He also has the highest completion percentage for the 11-20 yard pass at 68% (Luck 46%, Tannehill and Weeden 56%)

Weeden:


With Justin Blackmon at receiver, the mention of Oklahoma State's offense conjures up images of big plays and deep throws. 22.5% of Weeden's throws were screens (side note: when I did Matt Barkley, an amazing 25% of his throws were screens)
Weeden doesn't have the top completion percentage at any of the depths, but is very solid at all of them, with a good 52% completion rate at 20+ yards, he can hit the deeper passes pretty well
Strangely, from my sample, the percentage of his total throws at 1-5 yards and 11-20 yards were exactly the same, choosing to throw at those depths 24.06% of the time

Tannehill:


As you can tell by my stats, I'm a big fan of completion percentage as a key predictor for college QBs, if that holds true - color me unimpressed with Tannehill
He throws deep 15% of the time, the second most of the QBs, but completes them at a paltry 32%, the worst of the QBs
21% of the his throws are screens, and out of all the depths for him - he throws the 6-10 yard ball the most at 23%
While the other QBs have a overall completion % for the season at around 70-72%, Tannehill is lucky to hit 63%

If you're up for it, I went a little bit deeper, creating two of my own metrics.

PRR- Perfect Receiver Rating, I just took out drops here, essentially trying to eliminate the impact of talent level, at least on a basic level. These had to be flat out drops, not the receiver could have caught it, but he had to have it in his hands and then drop it.

Adjusted Completion Percentage - I like this stat a lot, the idea is to see what would happen if you took RGIII and subbed him in for Andrew Luck in the game. This started when I was doing Matt Barkley (before he announced he was staying), because he had a completion percentage being driven up by the ridiculous amount of screens he was throwing. I wanted to see what his overall completion percentage would be if he had to throw less high percentage passes (screens/1-5yds) and more low percentage passes (deep balls). Thus on this one, I averaged how often they threw each depth to make a standard percentage for each distance, here's how it came out:



Screen
16.76%


1-5 Yards
25.71%


5-10 Yards
21.42%


10-20 Yards
22.24%


20 + Yards
13.88%





So when you average out how often they all threw the screen, it comes out to 16.76% and so on. Then using their completion percentages, I made them all throw passes at this rate to see what their new overall completion percentage was (their completion percentages at each depth would be the same)

I then combined these metrics to see what would happen if they all threw the same depth of passes to the same receivers.

Here's the PRR (taking out drops) completion %s:



Perfect Receiver
(PRR)






Griffin
Luck
Tannehill
Weeden



Comp%
Comp%
Comp%
Comp%


Screen
90.00
88.89
100.00
92.68


1-5 Yards
77.42
91.89
69.23
85.00


5-10 Yards
62.07
70.83
66.67
73.33


10-20 Yards
80.95
50.00
61.29
59.52


20 + Yards
62.50
50.00
33.33
57.89


Total
73.60
73.08
67.88
75.58





I will say this is probably skewed towards Weeden a bit because I used the Stanford and Texas games and his receivers had 12 drops in those two games. But still interesting.

When you use those completion percentages with the Adjusted Completion %, here's what you get:



PRR-AdjComp%
Griffin
Luck
Tannehill
Weeden



Comp%
Comp%
Comp%
Comp%


Total
74.95
71.75
67.09
74.36




RGIII has the highest overall completion % at a whopping 75%, Weeden came in at 74.3%, Luck at 72% and Tannehill at 67%

Griffin is definitely helped by the fact that the AdjComp% gives him more high percentage throws like screens and 1-5 yd throws and take away some of the inaccuracy of the deep ball (even if he throws it well, it's still fairly low percentage

Luck isn't helped a whole lot, just because it takes away the weight of his crazy accuracy in the 1-5 yard range

Tannehill gets some bump too, just because it ensures that he throws less of his god awful 32% deep ball.

That's what I have for you. I know it was long, but I do a bunch of this stuff before the draft and I need somewhere to post it, so I'm not just hoarding info. I hope this helps you when you're looking at QBs and maybe helped you view something in a new light. If you want to take a look at the whole spreadsheet, or want more info (I have overall drop%, how effective they are in the shotgun/under center/ pistol), let me know and we can work something out. Thanks for reading it all guys, I really appreciate it.

ROADRUNNER
03-08-2012, 09:51 AM
Thank's for doing it well done...............

Fin Thirteen
03-08-2012, 10:19 AM
Really nice work. Certainly helps the Weeden fans press their case that he's upper first round talent impacted by his age.

You can't have a perfect stat analysis (eg factoring where the throw allowed YAC, rather than the receiver got it off his own work), but one thing that would make a big difference to this analysis is whether they were all close games against similar standard opposition. You'd have to ignore weak opposition and junk-time scores.

Overall, very impressive. Certainly does RG3's case no harm. Excellent consistency at every distance. I'm not sure it damages Tannehill TOO much either, because we all know he's raw and inexperienced. With him, it's about what you believe his numbers COULD BE, rather than what they are. That's certainly a risk at the #8 spot, but at least he's over the magical 60% completion, no matter which stat you choose.

ckparrothead
03-08-2012, 10:36 AM
Hey guys.

You, sir...are a gem.

ckparrothead
03-08-2012, 10:41 AM
One thing to keep in mind about the completion percentages for Ryan Tannehill.

His receivers dropped 64 passes this year. That is a really, really HIGH amount. If his receivers had only dropped balls at the rate that NFL receivers drop balls, he would have had about 29 more completions. I'll do the numbers later, but even assuming a 100% even distribution of those drops, all those percentages for him which lag behind the others do go up quite a bit I would think.

NUGap
03-08-2012, 11:09 AM
Thanks for the comments, I appreciate it. There are definitely a good amount of flaws - the YAC and whether or not it was the QB who allowed it or just a good play by the receiver. Or if the receiver caught the ball on the 5.5 yard line, is it a 1-5 or a 6-10? In these cases, if they had forward momentum or caught the ball in stride - I gave them the longer depth. Touchdowns are especially screwy, I wouldn't use any of the TD data. Mostly because if they throw from their own 40, catch it at the 50 and then run 50 yards, you can't list it as a 60 yard pass - it's only a 10. So I had to end up making notes when something like that occurred.

You'll notice that all the season completion %s in my charts are about 2.5-3% under their whole season total. This is because I tried to pick out 1 excellent game, 2 average games, and 1 poor game all against solid competition. My next step (assuming I have time), is to factor in the quality of the defense, passing yards allowed, etc.

And FWIW, my 4 game sample confirms what you said on a smaller level - I had 6.02% of Griffin's passes as drops, 6.31% of Luck's as drops, 8.67% of Tannehill's as drops, and 8.02% of Weeden's as drops. In a perfect world, I'd do every one of their games, but they're damn hard to find.

As far as Tannehill, when I took away the drops, it helped him to the tune of 6% points. I can't put my finger on it, but there's something I'm just not seeing when I watch him. Even disregarding the stats, I definitely don't see the value at 8. I think I'd take Weeden, even with the age, all day and every day, later in the draft.

Skree
03-08-2012, 11:37 AM
Thanks for the comments, I appreciate it. There are definitely a good amount of flaws - the YAC and whether or not it was the QB who allowed it or just a good play by the receiver. Or if the receiver caught the ball on the 5.5 yard line, is it a 1-5 or a 6-10, in these cases, if they had forward momentum or caught the ball in stride, I gave them the longer depth. Touchdowns are especially screwy, I wouldn't use any of the TD data just because - if they throw from the 40, catch it at the 50 and then run 50 yards. you can't list it as a 60 yard pass - it's only a 10. So I had to end up making notes about all of those.

You'll notice that all the season completion %s in my charts are about 2.5-3% under their whole season total. This is because I tried to pick out 1 excellent game, 2 average games, and 1 poor game all against solid competition. My next step (assuming I have time), is to factor in the quality of the defense, passing yards allowed, etc.

And FWIW, my 4 game sample confirms what you said on a smaller level - I had 6.02% of Griffin's passes as drops, 6.31% of Luck's as drops, 8.67% of Tannehill's as drops, and 8.02% of Weeden's as drops. In a perfect world, I'd do all games, but they're damn hard to find.

As far as Tannehill, when I took away the drops, he was helped to the tune of 6% points. I can't put my finger on it, but there's something I'm just not seeing when I watch him. Even disregarding the stats, I definitely don't see the value at 8. I think I'd take Weeden, even with the age, all day and every day, later in the draft.Nice work thank you for sharing the info !

ckparrothead
03-08-2012, 11:43 AM
A couple of things to keep in mind about Tannehill are:

1. That adjusted PRR percentage of 67% is really good in its own right
2. Screens aren't the only way to pump up your completion percentage
3. He and Luck were the only players running something resembling a pro offense
4. Forget how many starts he has, he's only been practicing at the QB position 2 years I believe

Gonzo
03-08-2012, 11:59 AM
Wow, great job! Welcome to the board!

NUGap
03-08-2012, 12:06 PM
To your point about your pro-style offense...

I broke down each play by what formation they were in, here are the completion %s:




U-Center

Shotgun


Tannehill

71.67
55.56


Luck
65.12
75.93




I also have the breakdown of Tannehill's percentages at each depth in each formation:




Shotgun

U-Center


Screen
100.00
88.89


1-5
60.00
75.00


6-10
50.00
69.23


11-20
54.17
60.00


20 +
9.09
54.55


Total
55.56
71.67




I hadn't even taken the time to look at these. The under center numbers do show a lot of promise for Tannehill, considering the prevalence of shotgun/ spread formations. It could also mean that Tannehill was forced to throw out of the shotgun more when they were down in the 4th and threw incompletions, trying to make something happen. Maybe I'll go through and take out "trailing situations".

TedSlimmJr
03-08-2012, 12:23 PM
It's an interesting assortment of statistical statistics you've compiled here and it's probably a lot of fun to analyze, etc... but I typically don't like to bombard myself with the stats when it comes to quarterbacks...but that's just me.

I think anyone who's observed a fair amount of college football this past season, particularly in referrence to these QB's here... should be able to determine that Ryan Tannehill is required to make more NFL caliber throws on a down-to-down basis than most other QB's in this draft. You can see it with the amount of deep/intermediate, odd number routes on the route tree that he's required to throw towards the perimeter and outside the numbers. Your statistics also bear that out.

However, you have to watch the tape in order to understand the QUALITY of the throws individually. Every 12 yard, 18 yard, or 22 yard throw these QB's make isn't identical in quality. Some are easier to make than others simply based on the read and route combinations. For example, although Griffin has tremendous downfield accuracy with the deep ball, they're all coming off of IZ play action.

Tannehill was running a timing based WCO under Sherman, and one of Tannehill's biggest drawbacks right now in his development is that he doesn't throw with the level of anticipation yet that he'll need in the NFL. When his anticipation gets really bad, that's when he struggles the most... typically in the 2nd half of games after opposing defensive coaches have made their halftime adjustments.

I don't think completion percentage is relevant whatsoever in terms of predicting what type of quarterback any prospect is going to be in the NFL. You're hard pressed to find any QB's coming from the Big-12 in any given year that are viable NFL prospects that don't complete 65-70% of their passes. Those type of completion percentages are essentially built into the offenses by design. Sam Bradford and Colt McCoy both completed around 69-70% of their passes in college. I didn't think either was going to make an elite caliber NFL quarterback. Pat White was a 65-66% passer at WV. Brian Brohm, Colt Brennan, and John Beck were all 65-70% passers in college. Tim Tebow was a 66-67% passer for his career at Florida in Urban Meyer's offense. He struggles with accuracy more than any starting QB in the NFL.

Matt Ryan only completed 59% of his passes his Senior year. Matthew Stafford was a 55% passer until A.J. Green came along... which barely allowed him to complete over 60% of his passes his junior season. Brett Favre's best season at Southern Miss he only completed 55% of his passes. Eli Manning was a 58% career passer at Ole Miss... he only completed more than 60% of his passes twice. There's a ton of examples... the point is, completion percentage is way down on the totem pole in terms of determining what type of QB your getting.... if it's on the totem pole at all.

Intangibles like pocket presence, toughness, durability, leadership, work ethic, are all a little more crucial in my opinion. I think you always have to be careful with the statistics when you're dealing with system quarterbacks.

NUGap
03-08-2012, 01:00 PM
No, I absolutely agree with what you're saying. In no way am I saying that Brandon Weeden is better than Andrew Luck because he had a better AdjComp% or even a better regular season completion percentage. Like I said in the original post, the idea of the stats is to supplement what you see on tape, not to supplant the other factors.

For instance, let's say the media fell in love with Dan Persa as a quarterback, he had two consecutive 73.5% seasons and now they're touting him as a better QB than RGIII. Every time you hear a media pundit talk about Dan Persa, it plants a seed - wow he may be good. Now every time I see Dan Persa, when he throws the standard 5 yard dink and dunk pass in the Northwestern spread offense, I make a mental check. Yep, he did good. However, when he throws a wounded duck past 20 yards, I think - well I've seen him do other good things and everyone says he can do it. He must be good. If I had broken down Dan Persa, the idea would be to go back and say...gee, Dan Persa is only completing .0001% of his 20+ yard passes - maybe I should check out his tape more closely. Obviously, this is an exaggerated example - but the point is to say, it's a way to check what you see.

Think of it like combine numbers. We know RGIII is fast, but when he runs a 4.34, we still get all excited. When we see that Dwayne Allen ran a 4.89, we take a little bit of a pause. I'm not going to move Allen to the 7th round now, but I'm going to go back and look at his tape a little harder. Think of this whole little exercise as extended combine numbers for the QBs, verify verify verify.

I've done math in the past on if season completion percentage leads to success, just for fun - it doesn't. Your points are valid, I can bring up Chad Henne who had a career comp% of 59.2 or Jamarcus Russell who had a senior year at 67.8%. But there is no negative or positive correlation, r = 0, so to speak. More data never hurts though. If I (or a team) can have one more factor, no matter how small or large, to weigh when evaluating quarterbacks - I am that much more prepared than the next guy.

SF Dolphin Fan
03-08-2012, 11:34 PM
Great work. It really points to Robert Griffin III rare ability in the deep passing game. Also probably the fact that Andrew Luck didn't really have the receivers to make the deep game work at Stanford. I think Coby Fleener was his biggest deep threat.

ckparrothead
03-09-2012, 02:23 AM
For what it's worth, my velocity measures tells me what we already know: Brandon Weeden has a damn good arm. Best in class. Velocity is second to none. Robert Griffin's velocity doesn't even touch it. Nor Ryan Tannehill's. Actually Tannehill and Griffin come out in a tie. Both are ahead of pre-surgery Peyton.

One thing people may not realize is that Matt Flynn's velocity on similar throws of similar distances is actually consistently higher than Griffin's and Tannehill's. I'd attribute this to the consistency you see from a more mature pro arm. Tannehill and especially Griffin can top out pretty close to where I've seen Flynn top out, but their velocity is inconsistent so Flynn's average is higher.

TenFinnfan
03-12-2012, 07:44 AM
Great work. Do you have a previous report on Flynn that you posted.May have missed it. Be great to see.
Tenfinnfan

NUGap
03-12-2012, 06:19 PM
I thought about doing one on Flynn to compare to these guys, but since he only started two games, any comparison to these QBs would be pretty off. Even though I 'only' did 4 games of the college QBs, if I had picked any two for them - I probably could have manipulated the stats to say anything I wanted. Four is a better sample size. So as much as I want to do Flynn to compare, I'm not planning on it.

Finfanforever
03-14-2012, 11:11 AM
For what it's worth, my velocity measures tells me what we already know: Brandon Weeden has a damn good arm. Best in class. Velocity is second to none. Robert Griffin's velocity doesn't even touch it. Nor Ryan Tannehill's. Actually Tannehill and Griffin come out in a tie. Both are ahead of pre-surgery Peyton.

One thing people may not realize is that Matt Flynn's velocity on similar throws of similar distances is actually consistently higher than Griffin's and Tannehill's. I'd attribute this to the consistency you see from a more mature pro arm. Tannehill and especially Griffin can top out pretty close to where I've seen Flynn top out, but their velocity is inconsistent so Flynn's average is higher.

CK,
You are the GM for the Dolphins...let's assume Manning is out of the picture, do you sign Flynn or draft Tannehill or Weedon. If you go for the rooks...is #8 too high? Where do you see Weedon (what round & pick). Excuse me for putting you on the spot but I really value your opinion. Thanks

Finfanforever
03-14-2012, 11:12 AM
Awesome post! Great in depth work

J. David Wannyheimer
03-14-2012, 10:46 PM
I really enjoyed reading this thread. Thanks.

Wildbill3
04-27-2012, 12:10 PM
Deserves a Bump.

sjsharkfanredux
04-27-2012, 02:52 PM
Great work NUGap on the break down of game statistics. Your posts were intriguing reads.