Writers of Pro Football Prospectus 2008

14 Oct 2015

ALEX: Week 5

by Scott Kacsmar

We've been writing a lot in the last few months about a new metric called Air Less EXpected, or ALEX for short. ALEX measures the average difference between how far a quarterback threw a pass (air yards) and how many yards he needed for a first down. If a quarterback throws a pass five yards behind the line of scrimmage on third-and-15, that would be minus-20 ALEX. The best application of ALEX is to look at third and fourth down, when it's really crucial to get 100 percent of the need yards to extend the drive.

The average ALEX on third down this season is plus-1.0, so the average third-down pass is being thrown one yard beyond the first-down marker (the "sticks"). Most quarterbacks have a positive ALEX, and the caliber of quarterbacks who are in the negative is usually very poor.

I wrote about this stat as "Air Minus Need Differential (AMND)" when looking at Zach Mettenberger in the offseason. That clunky name was also used in the Kansas City chapter of Football Outsiders Almanac 2015, which of course detailed the struggles Alex Smith has at being aggressive. That finally led to calling it ALEX, which we featured on ESPN Insider last week for the most aggressive and conservative quarterbacks since 2006.

Thanks to the growing interest from readers, we are starting a weekly ALEX feature here on FO. Starting next week, it will be a short feature each Monday afternoon. We'll look at the most significant ALEX plays from the week and update the rankings of qualified quarterbacks. These numbers are subject to change at season's end. The data on 2006-2014 is the same as what we use for stats like receiving plus-minus and YAC+, which excludes passes that are thrown away, batted at the line or when the quarterback was hit in motion. The 2015 data currently includes all passes. Eventually game charting will filter out those passes excluded in previous years, and correct any mistakes in the air yards data.

Week 5's Most Conservative Plays

The Lowest ALEX

Teams: Denver at Oakland
Situation: third-and-26 at own 39, leading 9-7 in fourth quarter with 11:32 left
Play: Peyton Manning to C.J. Anderson for 8 yards
Air yards: minus-4
ALEX: minus-30

Peyton Manning wasn't about to air out another pass to Charles Woodson, so the Broncos just went with a running back screen for an 8-yard gain. They were flagged for holding on the play, but Oakland obviously declined.

The Worst Play of the Week

Teams: Oakland vs. Denver
Situation: fourth-and-19 at own 49, trailing 16-7 in fourth quarter with 2:38 left
Play: Derek Carr to Marcel Reece for 4 yards
Air yards: minus-2
ALEX: minus-21

Did Oakland lose track of the downs? It doesn't get more desperate than fourth-and-19 when you're down two scores in the final minutes. Yet the call was a screen to fullback Marcel Reece for a 4-yard gain. This is the worst ALEX play on fourth down in 2015, but what about on record?

Going back to 2006, Oakland's screen ranks as the 10th-lowest ALEX pass on fourth down. The 2008 Lions of 0-16 fame still hold the title belt. Against Jacksonville, Drew Stanton threw a 2-yard pass to Michael Gaines on fourth-and-34 for minus-32 ALEX. The play gained 13 yards.

Since 2006, on fourth down with at least 10 yards to go, there have been 18 passes thrown at or behind the line of scrimmage. Only one pass converted for a first down, and it was a fake field goal by the 2012 Packers against the Bears. Here is the breakdown of converting passes thrown on fourth-and-10 or longer since 2006:

  • Thrown at or behind the LOS: 1-for-18 (5.6 percent)
  • Thrown beyond the LOS, but short of the sticks: 39-for-154 (25.3 percent)
  • Thrown at or beyond the sticks (positive ALEX): 118-for-341 (34.6 percent)

Come on, Oakland.

Week 5's Most Aggressive Plays

The Highest ALEX

Teams: Pittsburgh at San Diego
Situation: first-and-10 at own 31, trailing 20-17 in fourth quarter with 2:00 left
Play: Quarterback: Michael Vick to Darrius Heyward-Bey, incomplete
Air yards: 54
ALEX: plus-44

Michael Vick was our most aggressive ALEX quarterback in last week's ESPN Insider study. He led the league in ALEX on third down in 2006 and 2010. The Steelers mostly used him conservatively before Vick started unleashing bombs in the fourth quarter. This was a first-down play right out of the two-minute warning, but Vick underthrew the pass. Vick also had the week's third-highest ALEX play (plus-40) when he threw a Hail Mary to Darrius Heyward-Bey that was intercepted before halftime.

Speaking of Hail Mary throws, Brian Hoyer had the week's highest ALEX (plus-37) on a third or fourth-down play when he threw a 42-yard Hail Mary touchdown on fourth-and-5 to Jaelen Strong before halftime against the Colts.

The Best Play of the Week

Teams: Pittsburgh at San Diego
Situation: first-and-10 at own 28, trailing 17-10 in fourth quarter with 7:55 left
Play: Michael Vick to Markus Wheaton for a 72-yard touchdown
Air yards: 42
ALEX: plus-32

It's just a coincidence our plays came from the same game again. Earlier in the quarter, Vick's first great deep ball of the night connected with Markus Wheaton on a double move to beat Brandon Flowers for a stunning 72-yard touchdown. The Steelers had produced three points on 10 drives prior to that point, and Ben Roethlisberger reportedly was the one who called this play for Vick. Of all the unexpected one-play touchdown drives, this would rank up there.

2015 ALEX Rankings Thru Week 5

Finally, here is our first weekly table to show where each qualified quarterback (minimum 20 passes) ranks in ALEX on third downs only (does not include fourth down like some other FO stats for third down do). There are also rankings for DYAR, average need yards (ranked from highest to lowest) and conversion rate.

Rk Quarterback Team ALEX DYAR Rk Passes Avg. Need Rk CONV% Rk
1 Ben Roethlisberger PIT 9.4 284 2 27 7.5 20 63.0% 1
2 Ryan Fitzpatrick NYJ 6.3 102 16 39 6.7 29 41.0% 17
3 Carson Palmer ARI 6.2 173 10 33 6.6 31 45.5% 12
4 Brian Hoyer HOU 5.4 95 17 25 7.7 18 48.0% 9
5 Matt Hasselbeck IND 4.8 44 24 21 6.7 28 38.1% 22
6 Blake Bortles JAC 4.4 128 12 45 7.4 22 40.0% 20
7 Russell Wilson SEA 4.2 113 15 39 6.0 35 46.2% 10
8 Andy Dalton CIN 2.5 189 6 37 8.5 7 48.6% 7
9 Tyrod Taylor BUF 2.5 78 21 37 8.4 9 40.5% 19
10 Cam Newton CAR 2.3 92 18 34 8.6 6 38.2% 21
11 Matt Ryan ATL 2.3 243 4 56 6.4 34 51.8% 5
12 Andrew Luck IND 2.0 120 14 26 9.2 3 46.2% 11
13 Joe Flacco BAL 1.8 -74 31 53 7.3 23 32.1% 29
14 Drew Brees NO 1.5 176 8 52 7.6 19 44.2% 14
15 Josh McCown CLE 1.0 299 1 41 7.2 25 58.5% 3
16 Aaron Rodgers GB 0.9 82 20 37 6.6 30 40.5% 18
17 Nick Foles STL 0.5 -91 34 53 8.4 8 30.2% 31
Rk Quarterback Team ALEX DYAR Rk Passes Avg. Need Rk CONV% Rk
18 Colin Kaepernick SF 0.4 47 23 42 7.9 17 42.9% 15
19 Marcus Mariota TEN 0.4 86 19 35 8.0 16 42.9% 16
20 Jay Cutler CHI 0.4 249 3 36 6.6 32 61.1% 2
21 Peyton Manning DEN 0.3 -59 30 57 8.1 14 36.8% 24
22 Ryan Mallett HOU 0.2 -78 32 42 7.2 24 35.7% 25
23 Derek Carr OAK 0.0 189 7 50 7.1 26 48.0% 8
24 Kirk Cousins WAS -0.2 167 11 60 7.4 21 50.0% 6
25 Ryan Tannehill MIA -0.2 -40 27 46 8.3 10 23.9% 33
26 Tom Brady NE -0.2 176 9 28 6.5 33 53.6% 4
27 Teddy Bridgewater MIN -0.2 124 13 34 8.2 11 32.4% 28
28 Eli Manning NYG -0.4 215 5 53 7.0 27 45.3% 13
29 Jameis Winston TB -0.5 -87 33 52 8.9 4 34.6% 27
30 Sam Bradford PHI -0.8 -48 29 45 9.3 2 20.0% 34
31 Matthew Stafford DET -0.8 -140 35 49 8.2 12 34.7% 26
32 Michael Vick PIT -1.7 16 25 20 8.2 13 20.0% 35
33 Philip Rivers SD -2.2 -47 28 46 8.7 5 37.0% 23
34 Brandon Weeden DAL -2.4 73 22 25 8.0 15 32.0% 30
35 Alex Smith KC -6.3 -4 26 44 9.8 1 29.5% 32

Ben Roethlisberger's league-leading plus-9.4 ALEX is certainly not sustainable, but he is first in conversion rate while Michael Vick is currently last. Alex Smith is naturally last in ALEX, but he is really outdoing himself this year at minus-6.3. He has also been in the longest third-down situations, which only feeds Kansas City's conservatism.

Posted by: Scott Kacsmar on 14 Oct 2015

30 comments, Last at 16 Oct 2015, 12:16pm by Scott Kacsmar

Comments

1
by PaddyPat :: Thu, 10/15/2015 - 1:17am

Since the ALEX discussion debuted this season (I first read about it in FOA this year), I have been watching Tom Brady, and the consistency with which the Patriots' offense throws behind the line to gain on key downs. It makes me wonder if crossing routes and YAC opportunities should play into this discussion. I wouldn't be surprised, for example, if Brady similarly had a poor ALEX number in the Super Bowl, and yet, his receivers were picking up a lot of the 3rd downs (57%). Is it conceivable that there is a systematic element to this--that an offense predicated on YAC-type plays can get away with poor ALEX based on scheme and design? Just a thought. There's certainly a lot of grist in this metric and in attempting to determine how accurately it reflects performance and skill.

3
by Jerry :: Thu, 10/15/2015 - 3:08am

All the factors you're looking for are in the last column, where Brady is fourth in conversion percentage.

Like a lot of numbers, ALEX is fun to look at, and does indeed tell us something, but nobody is claiming that it's a great way to evaluate passers in and of itself. It tells us that both Brady and Tannehill throw, on average, the same distance short of the sticks. It doesn't pretend to tell us that Brady is much more successful with his passes. Just enjoy (or ignore) ALEX for what it is, and don't try to attach too much meaning to it.

11
by Noah Arkadia :: Thu, 10/15/2015 - 9:29am

I think there's more meaning than you think. For example, Brady is 33 in distance to go (meaning one of the lowest numbers) which means that throwing short and letting the receiver get it is more feasible -a strategy emulated by... Cutler, of all people. And, to a degree, Cousins and Carr. However, there's a bunch of QBs who have high ALEXs and short distances to go, with the corresponding high conversion percent.

Meanwhile, the QBs that draw my attention are Dalton and Luck, who both face among the longest distances to go on 3rd and both feature some of the highest conversion percents (and nice healthy ALEXs). None of the other QBs in the top 15 do well at converting 3rds (which is only natural, as if you can't get it done on first and second, why should 3rd be any easier?). In the opposite camp (short to go, low conversion percent) you have Hasselbeck, Bortles, Mallet... and Flacco.

But maybe you are right after all, this is mostly about avg to go and conversions, not so much about ALEX per se.

------
Who, me?

14
by Hoodie_Sleeves :: Thu, 10/15/2015 - 10:23am

It's an interesting stat, but like you said - it's key to not read too much into it.

There's a huge difference in expected result between checking down to a RB 3 yards past the LOS on 3rd and 10 and throwing to a WR in the middle of a carefully executed rub play with no one near him. And ALEX has no way of knowing that.

And I do agree with one of the below posters that it might make more sense to look at the 'success yards' rather than 'to-go yards' - a 7 yard pass on 1st and 10 shouldn't be a negative event.

19
by duh :: Thu, 10/15/2015 - 2:26pm

Isn't the Cutler number just a reflection of Forte?

23
by Scott Kacsmar :: Thu, 10/15/2015 - 3:17pm

Forte makes up 8/36 plays for Cutler. 3/8 in converting, and the ALEX is minus-6.9.

27
by duh :: Fri, 10/16/2015 - 11:30am

Thanks for the data on Forte I appreciate it. is that a total-6.9 ALEX? (I'd think so but want to be sure I understand)

29
by Scott Kacsmar :: Fri, 10/16/2015 - 12:14pm

The ALEX is an average.

13
by tuluse :: Thu, 10/15/2015 - 10:23am

This doesn't tell you how good a QB is, it describes their style of play.

24
by Scott Kacsmar :: Thu, 10/15/2015 - 4:24pm

Just looking at the QBs we know to be the best in 2006-2014...

Drew Brees is the most consistent at converting, ranked in the top 7 for each season (2006-2014). His ALEX rankings are really all over the place: 28th, 28th, 10th, 12th, 10th, 25th, 1st, 19th, 3rd.

Aaron Rodgers is the most consistent at ALEX. His ALEX rankings since 2008: 4th, 2nd, 3rd, 3rd, 2nd, 3rd, 2nd. Again, that makes his 2015 start pretty interesting. He's usually top 9 in converting, though was 15th in 2008 and 18th in 2010.

Ben Roethlisberger has five top-5 finishes in CONV%, but his ALEX has gone down under Todd Haley outside of his big September start this year. He was usually top 10 in ALEX under Bruce Arians, but has ranked 14th, 21st and 13th under Haley.

Philip Rivers was top 7 in ALEX every year from 2007-2013. He slipped to 14th last year and is currently 33rd, so we've seen him dink and dunk more than ever this season as I've mentioned in other articles. He has 8 top 10 finishes in CONV%, including twice leading the league.

Tony Romo is usually top 10 in ALEX and CONV%, though he's had some low years such as 30th in CONV% (2013) and 28th in ALEX (2009).

Peyton Manning consistently had the best CONV% before last year: 1st, 2nd, 4th, 1st, 9th, 3rd, 3rd, 17th. And he's continued that fall this year. He's never been lower than 14th in ALEX with five seasons in the top 8.

Tom Brady has the numbers that support a style change in 2010 when they went 2 TE and traded Randy Moss. From 2006-2009, Brady was 5th-7th in ALEX and 7th-17th in CONV%. Starting in 2010, he's been 26th, 13th, 22nd, 23rd and 19th in ALEX (around league average). And 14th, 7th, 5th, 20th, 4th in CONV%.

I don't think anyone has been able to stay consistently high in CONV% while having one of the lowest ALEX averages in the league. Alex Smith ranked 10th in CONV% in 2014, but that was a big outlier in his career where he's always been pretty piss poor in CONV%.

28
by Noah Arkadia :: Fri, 10/16/2015 - 11:57am

Interesting, thanks.

------
Who, me?

2
by Ben :: Thu, 10/15/2015 - 2:30am

What was the number on the 3rd and 6 throw by Hasselbeck to ice the game versus the Texans?

22
by Scott Kacsmar :: Thu, 10/15/2015 - 2:53pm

Plus-31. That's really the best choice for a play of the week for ALEX on third down. Great throw, and Hilton beat the Texans deep again.

4
by wrbrooks :: Thu, 10/15/2015 - 3:17am

You might think about changing the "expected" part of ALEX for 1st and 2nd downs. Maybe use the threshold for a successful play (40% of the distance to go)?

17
by Aaron Schatz :: Thu, 10/15/2015 - 11:59am

Honestly, not much of an issue. We're not really using ALEX for first and second downs at this point. We're bringing it up to talk about the validity of certain strategies to convert on third down.

Edit: Um, Scott did talk about other downs up there, didn't he? My mistake.

21
by Scott Kacsmar :: Thu, 10/15/2015 - 2:52pm

Next week we will highlight only third and fourth-down plays. I really don't use ALEX for 1st and 2nd down. I brought it up here since there weren't many interesting examples this week of high ALEX plays, and I didn't really want to highlight a Hail Mary. You can calculate ALEX on any play, but it works best for the later downs.

5
by Bright Blue Shorts :: Thu, 10/15/2015 - 4:23am

Given the Derek Carr scenario I'm wondering has football become so averse to turnovers?

My initial weighing up of that is wouldn't it be better to throw the ball down the field for a jump ball and either hope one of your WRs makes the catch or a defender is stupid enough to intercept it thereby at least improving field position for your defense by 20-yds.

On the other hand, I wonder how often does the screen work out on 3rd&long or 4th&long? Is the possibility of a defender returning the interception significantly too much of a downside.

It's easy to question what Carr was doing because it look so ridiculous when it fails (see also the Kansas City screen pass returned for a TD against the Broncos).

6
by ammek :: Thu, 10/15/2015 - 4:46am

Another useful statistic to add to the FO collection. Thanks for doing the work on this, in an era of short passing it's an important field of inquiry.

Would it be possible to add completion percentage to the table? It would help us to differentiate between the QBs who suffer with inaccuracy/bad receivers and those who simply throw a lot of checkdowns. In future weeks, I hope you'll look at the targets of these passes – how many are running backs, how many are thrown outside the hashmarks, possibly even at conversion rates and DYAR for receivers. There is a lot of potential with this statistic.

I say this because, looking at the 2015 numbers so far, Aaron Rodgers is exactly average in conversion rate and a little below average in DYAR. That's very much a function of the lack of receiving options; in previous season, his third-down DVOA has been very good. But the Packers are thin at WR this year, the only healthy TE does not get open consistently, and the RBs aren't breaking any tackles. Rodgers is on pace for something like 70 rush attempts this year, largely because there's nobody open downfield. If he wasn't mobile, I think his ALEX would be even worse.

Your numbers support the Bears fans on this site who say that Cutler has been playing well, especially as his receiving options are no better than Rodgers'.

7
by Snoth :: Thu, 10/15/2015 - 6:36am

This seems like a semantics statistic.

9
by dbostedo :: Thu, 10/15/2015 - 8:51am

What do you mean by "semantics statistic"?

16
by Snoth :: Thu, 10/15/2015 - 10:50am

It's suppose to show the most aggressive/conservative QBs on 3rd down but a QB who on average only has a few yards to go on 3rd down and has a conversion rate in the 60s doesnt mean he's aggressive nor conservative

20
by Scott Kacsmar :: Thu, 10/15/2015 - 2:49pm

Re: ammek and Snoth

I've included the conversion rate to show more of the big picture (i.e. is this offense's strategy working or not?). You should be looking at the ALEX column if you want to note the level of aggression. A QB who often is throwing deep on third-and-short is definitely aggressive.

I don't think completion percentage ever adds much value on third down. I prefer to keep that for failed completions where we can pick out which QBs are boosting the stats by checking down.

I could add the average air yards and YAC gained in the table on a weekly basis, if people want to see that. See how Alex Smith is throwing by far the shortest passes despite being in the longest third-down situations.

We could break it down by position thrown to (RB/WR/TE), but that would be more of a midseason and final season review instead of something I put up weekly.

I have to remind everyone that this is just 5 weeks of data, or 4 games for some people (fewer if you're injured like Roethlisberger). With that said, Aaron Rodgers is an interesting one so far because he is so consistent with ALEX. He has ranked in the top 4 in every year since 2008. He is the most aggressive QB on third-and-short and third-and-medium, but actually quite conservative on third-and-10+. Overall we've seen him shorten his game this year, and maybe that is because of no Jordy Nelson. Definitely one to keep an eye on as the weeks go by. I think the last two games give us some pause about this GB offense, being held to 17 points in consecutive weeks.

8
by aces4me :: Thu, 10/15/2015 - 8:33am

Am I reading that correctly? KC's third downs average almost 3rd and 10. Yikes!

10
by dbostedo :: Thu, 10/15/2015 - 8:52am

It appears hail-mary's are included? Should they be filtered out? Do they have any significant impact?

12
by andrew :: Thu, 10/15/2015 - 9:42am

I would also like to see defensive Alex against.

i.e., what defenses encourage (either via pressure or airtight coverage) quarterbacks to throw short.

Because this is all leading to..... Defense-Adjusted ALEX, or DALEX.

This will be a better measure of which quarterbacks.... Exterminate! their team's hopes on third down.....

15
by TimK :: Thu, 10/15/2015 - 10:47am

Next step, work out a way to make DAVROS the DALEX equivalent of DAVE?

18
by BlueStarDude :: Thu, 10/15/2015 - 12:33pm

Nevermind. I get it now.

25
by nat :: Fri, 10/16/2015 - 11:03am

In a way, this whole ALEX exercise is a step backwards for FO. It ignores actual effectiveness of the QBs in favor of getting yards a "politically correct" (aka "aggressive") way. It would be as if we decided to focus on QB runs because handing the ball off is cowardly and demonstrates weakness on the part of the QB.

But we can look at the table above and ask "which QBs are most effective at passing on third down?". DVOA would be ideal for that, but DYAR per pass is pretty close. It handles the trade off between a few first downs and gaining a lot of extra yards pretty well.

Your top ten in DYAR per pass are

Roethlisberger
McCown
Cutler
Brady
Palmer
Dalton
Luck
Ryan
Manning (Eli)
Hoyer

That they range from 1st to 28th in ALEX simply shows that there are a lot of ways to be a great passer on third down. How a QB attacks the chains is largely a matter of the personnel he has and the coverage he faces. The best QBs are able to attack the whole field. It would be foolish to only attack the area beyond the sticks. It would be just as foolish to only attack with dump offs and screens. The mix should be dictated by the players and the coverage faced, not some misguided concept of throwing longer passes being more "aggressive" and therefore (I guess) more "manly".

26
by tuluse :: Fri, 10/16/2015 - 11:21am

You're presuming high ALEX = good. It's a descriptive stat, it tells how a player played, not whether said play was good or bad. That's why conversation rate is listed on the table.

30
by Scott Kacsmar :: Fri, 10/16/2015 - 12:16pm

Exactly, and quantifying the differences in playing style in this era is just as important as anything, but haters gonna hate.