I've finally managed to automate the process, so now we can have a continuously updated table of Air Yard stats. For now, only the current season is available, although if you do a search on the site you'll find posts with Air Yard stats for previous seasons.
As a review, Air Yards is a concept introduced back in 2007, the first year of this site. I always wondered why Donovan McNabb would get credit for 80 passing yards for a screen toss to Brian Westbrook caught 2 yards behind the line of scrimmage. Everyone is familiar with Yards After Catch (YAC), a particularly popular buzzword during the ascendancy of the west coast offense. I thought that looking at a QB's portion of passing yards that are not generated by YAC would be an interesting way to measure performance. Air Yards is simply the arithmetic complement of YAC. It's the yardage a pass travels through the air forward of the line of scrimmage to the point of reception.
I thought of Air Yards as somewhat analogous to the DIPS (Defense Independent Pitching Statistic) concept in baseball. DIPS strips away factors beyond a pitcher's direct control and can be a better predictor of future performance, and therefore a truer indication of pitcher ability. With Air Yards we may be throwing some of the baby out with the bathwater, but the result is a higher signal to noise ratio.
The concept has gained some limited acceptance. I see it mentioned or discussed at other sites occasionally, but I've never seen anyone produce the numbers. One place where it's been accepted is at ESPN. The idea behind Air Yards is one of the core concept's within its QBR stat.
There are two derivatives of Air Yards that interest me. First is Air Yards per attempt (AirYPA). This is an indication of how deep a QB and his offense are able to get completions. It's a function of a lot of factors, but generally the better passing teams are going to get deeper completions and have fewer screens and check-downs.
The second is %YAC, which says what proportion a QB's passing yards have come from receiver YAC. A high %YAC is an indication the QB is benefiting from lots of yards gained from his receivers' abilities to gain yards in the open field.
Don't get me wrong, Total YAC is not a bad thing. The more completions a QB makes, the more YAC he'll see, which is good. But a disproportionate amount of YAC can be an indication that a QB's success in terms of total yards might be misleading.
Here is the Air Yards page, which will be permanently available through the main menu under Stats | Offense. As always the table is sortable.
It looks like the Total Pass Yds and YAC columns are switched?
It'd be interesting to see it broken up in different ways like YAC by running backs vs. wide receivers, or for different Air Yard groupings to try to isolate screens and checkdowns.
Neat to see RGIII and Luck on different ends of the YAC% spectrum.
Ok. Thanks. I'll fix it shortly.
Columns should be fixed now. Thanks again.
Do receivers get a recorded yards after contact, and if so would it be possible to divide catches by yards before contact and yards after contact? This would give quarterbacks credit for finding a wide open receiver with room to run but limit the receiver breaking tackles and then gaining extra distance and might be a better indicator of quarterback performance.
"throwing some of the baby out with the bathwater"
It seems to me that this might be throwing a lot of the baby out with the bathwater. YAC could be just as much determined by the quarterback's ability to find open receivers and make accurate throws to lead those receivers as it is by his receivers' running ability.
Have you looked at how good of a predictor of future performance AirYPA is? My intuition tells me that YPA would probably be a better predictor.
Like Wez said, it'd be nice to be able to isolate screens, which have very little to do with the QB, while a QB's ability to throw timing routes accurately seems too important to ignore.
Anon, you can read Brian's research into who is responsible for YAC at the below link, or I can just tell you the conclusion: receivers are mostly responsible for YAC.
http://www.advancednflstats.com/2007/08/yac-receiver-correlation.html
For anecdotal evidence, I present you this quote from an Air Yards article written in 2008:
"The 2008 season's list of leaders in %YAC include Cassel, O'Sullivan, Campbell, Favre (again), Losman, and Wallace. But Matt Cassel is good, right? Maybe not. Keep in mind how good the team around him was. He was handed the keys to a Ferrari. If a QB racks up his passing yards with YAC, he's either throwing lots of short check-downs and screens, or he has spectacular receivers--or both. Neither is necessarily an indication of a particularly skilled passer."
Brian, if you one day run out of things to do it'd be great if you could code the advanced stat pages to allow for multi-year results. For instance, I'd love to see who had the most EPA from 2008 to 2010 without having to look at each year's page and manually add it up.
James. That's pretty interesting. Definitely not what I was expecting. I remember back in JP Losman days in Buffalo, Losman was a stud with the deep ball--I believe TFS had him rated as one of the best if not the best deep ball passer. But his short pass accuracy (based on the "eye" test) was just awful. Constantly seemed to hang his receivers out to dry, and couldn't hit them in stride to save his life, causing the receivers to always have to stop or adjust for a pass. Losman seemed to be in the middle of the pack on YAC/A. So not sure what that means. Perhaps the TFS analysis was flawed, and Losman wasn't very good getting the ball downfield as it suggested--though his AY/A isn't awful. Losman was always a puzzle to me. Classic case of a guy who looked great in shorts, but seemed to always come up short in the games. Sounds a bit like that other RoboQB from Buffalo's past--Rob Johnson. Too bad we don't have data going back further to look at the class of '83. Those guys had the reputation for slinging it downfield. Different era, obviously, but would be interesting to compare to the modern passer.
Alex Smith is leading the league in AirYPA? How in the heck did that happen?
I can think of five biases right off the bat:
[1] gives no credit to the QB for accurate ball placement
[2] gives no credit to the QB for proper usage of gifted YAC receivers
[3] penalizes QB's for substandard pass protection
[4] penalizes QB's for the offensive strategy which may or may not be related to their capabilities
[5] rewards QB's who have gifted downfield receivers
In order to apply this concept in a useful way to the QB, the receivers/protection/strategy must be quantified and controlled for, otherwise this is simply another (useful) description of a team's passing offense as a whole. Of course the same kinds of limitations apply to other passing stats as well.
So add this one to the list of 'individual' stats that are useful to win arguments and confirm previously held beliefs but dubious in their ability to describe whatever the 'truth' is.
I think a better alternative would be to throw out all passes to RB's and FB's, and only count yards to WR's and TE's. That way, quarterbacks would still get credit for timing routes and hitting their receivers in stride, but not for dumpoffs and screen passes, which involve little passing skill.
flower guy - Don't traditional passing stats commit #2, #3, #4, and #5 just the same or worse? And don't traditional passing stats give the QB credit for the elusiveness of his receivers? And it does give credit for accurate placement in as far as the completion is made.
Take a deep breath and just admit this is one alternative way to look at QB performance. It's imperfect but useful and enlightening.
So I'll just add your comment to the list of 'criticisms' that are useful to make the commenter feel smart and dubious in its ability to shed any light on the issue.
I can't help but think that 'Yards Before Catch' (YBC) would be a clearer name for this stat than 'AirYards'.
Don't traditional passing stats commit #2, #3, #4, and #5 just the same or worse?
I mentioned that other stats have similar limitations.
Take a deep breath and just admit this is one alternative way to look at QB performance.
This statement is always true and I admit it.
So I'll just add your comment to the list of 'criticisms' that are useful to make the commenter feel smart and dubious in its ability to shed any light on the issue.
If you feel that pointing out large flaws in something is of dubious value then so be it. After I wrote this post I began thinking of ways to reliably quantify the missing factors, so I'm actually interested in resolving these things. BTW, why the personal attack when I made none? Is this the norm?
And he completed his first 30 yard pass last week...
Wrote this email to a friend who sent this out, worth posting as well -
Commenters noting same issues I have with it, which is some offenses are designed to get lots of YAC (i.e. any West Coast scheme), and teams which throw downfield a ton will be skewed upward.
Note: haven't looked at leaders yet.
...and I'm right. Weeden, Rodgers, Rivers, Fitzpatrick all punished. Mark Sanchez can only hit guys deep on rare occasion, is rewarded. (As are Vick, Cutler, Newton.)
Conceptually nice, practically meaningless. Watch any film of a West Coast offense and you can see it's designed to get a ton of YAC. The idea is to create a higher % rate of completions while forcing the defense to cover a whole bunch of options.
Ironically, leaders in passing % as of now: RGIII, Ponder, Rodgers. Bottom: Sanchez, Luck. Where are they here? The reverse.
It's a nice concept, but I think it needs more work.
The best would be something that calculates all yards until first serious defensive tackling opportunity. Or, really, the old YPA metric I think is more telling than this.
Also, let me say that this is a great stat for analyzing passing offense. Air yards and YAC gives you a good starting point to ask these eight questions:
Is AY/YAC significantly impacted by the limitations/strengths of the QB/receivers/protection/strategy?
It's also a great QB stat when comparing Matt Cassel to Tom Brady using only their time with the Patriots (small sample for Cassel notwithstanding).
You, sir, are my hero!
Air yards still doesn't imply the "safety" of a throw. For instance, a wide receiver bubble screen might have 20 air yards, but it is 0 yards downfield. What about a stat just showing yards from line of scrimmage. So, a basic screen pass might be worth -4 yards, a wr screen 0 yards, etc.