Writers of Pro Football Prospectus 2008

05 Jan 2017

ALEX: Season Review

by Scott Kacsmar

In case you missed it, the 2016 NFL season set all-time records for the highest completion percentage (63.0 percent), lowest interception percentage (2.3 percent), and lowest sack rate (5.8 percent) in league annals. The passes keep coming out faster and shorter, which has placed an emphasis on tackling well.

You do not have to watch many games to see that checkdowns are prevalent in today's NFL too, especially on crucial third downs, which is why we have a stat like ALEX. This is a review of the regular-season results, and as always the numbers are subject to change after we clean up the game charting more in the offseason.

For those new to this metric, it is 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 5 yards behind the line of scrimmage on third-and-15, then that would be minus-20 ALEX. The best application of ALEX is to look at third and fourth downs, when it's really crucial to get 100 percent of the need yards to extend the drive.

In this piece, we will look at both defenses and quarterbacks for the 2016 season.

Defenses: 2016 ALEX Rankings Through Week 17

The following numbers are for third down only. Since this is defense, the teams are ranked from lowest to highest for the various stats. Lower DVOA is better. A lower ALEX means the defense is keeping the play short of the sticks. Short% is the percentage of third-down attempts that were thrown short of the sticks (negative ALEX). ALEX and Short% generally have strong correlation (typically near minus-0.85), but Short% is a good way to minimize the effects of outliers.

2016 Defenses: Third-Down ALEX Splits
Rk Team ALEX 2015 ALEX 2015 Rk CONV% Rk DVOA Rk Avg. Need Rk Short% Rk Passes
1 MIA -1.2 0.7 10 36.7% 8 8.5% 6 8.2 3 51.3% 2 150
2 DEN -0.4 1.0 15 35.1% 2 -38.9% 1 7.6 16 46.2% 6 171
3 DET -0.3 1.5 24 45.4% 29 50.0% 29 8.2 4 47.5% 5 141
4 HOU 0.0 1.9 27 40.8% 17 10.6% 7 7.4 23 40.8% 18 147
5 NYG 0.1 2.7 29 34.6% 1 -2.8% 3 8.3 2 49.2% 3 191
6 ARI 0.1 1.3 21 41.3% 19 33.0% 24 8.1 5 52.3% 1 155
7 ATL 0.1 0.4 6 42.7% 23 25.8% 21 7.6 19 43.3% 8 164
8 JAC 0.3 0.4 7 35.7% 4 22.8% 19 7.7 13 41.1% 14 168
9 SD 0.3 1.1 16 41.7% 20 11.9% 8 7.6 20 42.4% 10 144
10 LARM 0.6 -0.6 2 35.7% 3 20.0% 16 7.9 9 40.9% 15 171
11 NO 0.7 1.0 14 43.8% 25 41.2% 28 7.9 8 41.2% 13 153
12 TEN 0.7 0.4 5 40.9% 18 18.5% 15 7.7 14 34.4% 30 154
13 CLE 1.0 1.2 18 48.0% 31 40.6% 27 7.7 15 37.8% 23 148
14 SEA 1.0 0.2 3 38.2% 11 39.0% 25 8.0 6 47.9% 4 144
15 CAR 1.0 0.6 8 38.3% 12 15.6% 10 8.4 1 40.9% 16 154
16 PIT 1.0 1.3 22 36.5% 7 17.5% 12 7.8 10 41.7% 11 156
Rk Team ALEX 2015 ALEX 2015 Rk CONV% Rk DVOA Rk Avg. Need Rk Short% Rk Passes
17 WAS 1.1 0.6 9 51.6% 32 77.5% 32 7.5 21 43.4% 7 159
18 CIN 1.4 0.4 4 42.1% 22 27.4% 22 7.2 27 40.2% 19 164
19 PHI 1.4 2.2 28 39.8% 16 24.6% 20 7.8 11 42.9% 9 161
20 OAK 1.6 1.2 19 39.2% 14 21.4% 18 7.4 25 41.2% 12 148
21 SF 1.8 4.6 32 45.3% 28 52.0% 30 7.4 24 36.5% 26 159
22 DAL 1.9 1.1 17 36.1% 6 12.4% 9 7.4 22 38.2% 22 144
23 GB 2.0 1.3 20 42.0% 21 39.4% 26 7.9 7 39.9% 20 143
24 TB 2.1 0.9 13 35.8% 5 4.3% 5 7.8 12 36.6% 25 134
25 BUF 2.2 0.8 11 45.9% 30 54.9% 31 7.6 17 36.8% 24 133
26 NE 2.4 3.1 31 37.2% 9 18.2% 13 7.6 18 40.9% 17 164
27 BAL 2.5 1.4 23 38.7% 13 -8.1% 2 7.1 31 36.2% 27 163
28 MIN 2.6 0.9 12 37.8% 10 3.2% 4 7.1 30 31.4% 31 156
29 KC 2.8 2.9 30 45.1% 27 18.3% 14 6.9 32 30.2% 32 162
30 IND 2.9 1.7 26 44.7% 26 32.4% 23 7.2 29 36.0% 28 150
31 NYJ 3.0 1.6 25 39.6% 15 16.9% 11 7.3 26 39.0% 21 154
32 CHI 3.4 -0.6 1 43.8% 24 21.3% 17 7.2 28 35.4% 29 144

The correlation between ALEX from 2015 to 2016 was only 0.16, and nothing exemplified this more than the Bears going from first to last despite employing defensive coordinator Vic Fangio in both seasons. Of course, defensive inconsistency from year to year is nothing new in the NFL, but we would have to study the older seasons to see if there's ever any consistency among certain defenses in ALEX. The type of quarterbacks faced may drive these numbers more than any defensive scheme could, though the Bears did play Sam Bradford -- twice.

Miami was one of three defenses to have a negative ALEX on the season, which might not help much against a typically aggressive Ben Roethlisberger this weekend in the playoffs. Miami and Arizona were the only two defenses to force throws short of the sticks more than half of the time on third down. Denver's coverage was also outstanding with by far the best DVOA (-38.9%) of all 32 defenses on third down, and remember, this is for pass attempts only and does not include sacks -- which means everyone's DVOA is going to be higher. The Giants are the best defense left standing in the playoffs, and they also have some strong metrics here, ranking in the top five in ALEX, conversion rate, DVOA, average need yards, and Short%. On the flip side is the Washington team that the Giants eliminated from the playoffs in Week 17. The Redskins had the worst conversion rate and DVOA on third-down passes.

Kansas City is a most interesting team when it comes to ALEX, and we're not even talking about the inspiring Alex Smith yet. In 2015, the Chiefs had a nearly identical plus-2.9 ALEX, but still allowed the fourth-lowest conversion rate despite the lowest Short% (30.2 percent) in the NFL. This season, the Chiefs are at plus-2.8 ALEX, but 27th in conversion rate and have the same league-low 30.2 percent Short%. Some of this can likely be explained by the pass rush. In 2015, the Chiefs had the league's fifth-highest pressure rate (premium subscription required), but they have the fourth lowest in 2016. Not having a healthy Justin Houston for much of the season is a big problem.

Quarterbacks: 2016 ALEX Rankings Through Week 17

The following table looks at every quarterback this season with at least 50 third-down passes through Week 17. The quarterbacks are ranked from highest to lowest ALEX.

Rk Quarterback Team ALEX CONV% Rk DVOA Rk Passes Avg. Need Rk Short% Rk
1 Ben Roethlisberger PIT 4.0 45.9% 7 52.0% 6 133 7.1 31 33.8% 6
2 Matt Barkley CHI 4.0 50.0% 3 -0.9% 29 50 7.4 21 24.0% 1
3 Aaron Rodgers GB 3.5 51.0% 1 71.1% 3 153 7.1 27 29.4% 4
4 Cam Newton CAR 3.3 38.3% 21 26.4% 15 149 8.2 5 28.9% 3
5 Jameis Winston TB 3.0 47.8% 5 31.5% 12 161 7.6 18 28.6% 2
6 Eli Manning NYG 2.8 35.5% 30 -1.5% 30 169 7.1 28 40.2% 14
7 Tyrod Taylor BUF 2.3 41.5% 15 41.6% 8 118 8.0 8 36.4% 10
8 Kirk Cousins WAS 2.2 42.5% 13 37.1% 10 153 7.1 29 39.9% 13
9 Tom Brady NE 2.1 50.4% 2 94.4% 1 121 7.2 25 33.9% 7
10 Marcus Mariota TEN 2.1 45.4% 8 56.8% 5 141 7.5 19 31.9% 5
11 Matthew Stafford DET 1.5 44.4% 10 17.9% 21 151 7.6 17 37.7% 12
12 Ryan Fitzpatrick NYJ 1.4 29.7% 32 -32.8% 32 111 7.3 24 46.8% 28
13 Ryan Tannehill MIA 1.4 35.9% 29 14.7% 25 117 7.5 20 41.9% 17
14 Andrew Luck IND 1.4 44.4% 11 15.6% 24 124 7.4 22 36.3% 9
15 Drew Brees NO 1.3 49.7% 4 57.8% 4 161 6.9 33 43.5% 21
16 Blake Bortles JAC 1.3 36.4% 28 -24.0% 31 176 7.2 26 34.7% 8
17 Dak Prescott DAL 1.2 43.7% 12 44.8% 7 126 8.0 11 37.3% 11
Rk Quarterback Team ALEX CONV% Rk DVOA Rk Passes Avg. Need Rk Short% Rk
18 Philip Rivers SD 1.1 45.0% 9 30.9% 14 151 7.1 30 42.4% 18
19 Trevor Siemian DEN 1.0 37.9% 25 16.7% 22 132 8.0 10 43.9% 23
20 Andy Dalton CIN 0.8 38.6% 19 32.8% 11 158 7.8 13 41.1% 16
21 Brock Osweiler HOU 0.5 38.0% 23 0.9% 27 142 7.0 32 45.1% 24
22 Derek Carr OAK 0.5 37.9% 24 37.6% 9 153 8.1 7 40.5% 15
23 Russell Wilson SEA 0.4 39.0% 17 31.3% 13 146 8.3 4 42.5% 19
24 Alex Smith KC 0.4 38.9% 18 24.4% 17 131 7.4 23 49.6% 30
25 Joe Flacco BAL 0.4 37.6% 26 24.9% 16 178 8.1 6 48.3% 29
26 Colin Kaepernick SF 0.1 31.1% 31 23.0% 18 90 8.8 2 45.6% 25
27 Carson Wentz PHI 0.0 38.5% 20 0.4% 28 156 7.7 16 43.6% 22
28 Case Keenum LARM 0.0 38.0% 22 16.7% 23 92 6.8 34 53.3% 32
29 Carson Palmer ARI -0.2 42.2% 14 19.3% 19 154 8.6 3 42.9% 20
30 Matt Ryan ATL -0.6 46.4% 6 71.9% 2 125 7.9 12 45.6% 26
31 Cody Kessler CLE -0.8 37.5% 27 19.0% 20 56 8.0 9 46.4% 27
32 Blaine Gabbert SF -1.2 26.0% 33 -51.0% 34 50 7.8 15 50.0% 31
33 Sam Bradford MIN -1.3 40.1% 16 13.9% 26 152 7.8 14 55.9% 33
34 Jared Goff LARM -2.6 23.1% 34 -45.8% 33 65 8.8 1 63.1% 34

The correlation between ALEX and conversion rate was 0.61, which is higher than usual (2015 was 0.46). Again, much like with performance under pressure, any time YAC and broken tackles are involved, the game can get pretty unpredictable.

Ben Roethlisberger (plus-4.00) just nudged out Matt Barkley (plus-3.98) to lead the league in third-down ALEX for the second year in a row, and the third time since 2011. The fact that Barkley is near the top with some of the usual suspects (Aaron Rodgers, Cam Newton) is a little surprising, but then again, how much did we really know about Barkley as an NFL passer before this season? He was a turnover-prone mess as a rookie for the 2013 Eagles, and that part of his game continued this year. However, his "YOLO" style of play also benefitted Chicago at times, and you can see that Barkley actually finished right behind Rodgers and Tom Brady in conversion rate, and Barkley also had the lowest Short% in 2016. Still, his DVOA was terrible, with three picks on 50 passes.

It is no surprise to see Brady, Rodgers, and Matt Ryan were the top three in third-down passing DVOA, but Ryan was surprisingly just 30th in ALEX this season. Meanwhile, Brady's ALEX is the highest it has been since 2009, his last full season with Randy Moss, when he ranked seventh at plus-2.5. He fell off from his torrid midseason pace, but still finished as one of the top third-down quarterbacks this year. He had the highest average YAC (7.5), and Julian Edelman's 77-yard touchdown in Miami on Sunday on a minus-1 ALEX throw did a lot to boost that average. Rodgers has ranked in the top six in ALEX every season since 2008, but this was the first time he ranked higher than third in conversion rate. His opponent this week, Eli Manning, had good intentions with the sixth-highest ALEX this season, but ranked 30th in conversion rate and DVOA -- a sign of the offensive struggles for the Giants.

We of course have to mention how Alex Smith did this year in what has been a more offensive season for the Chiefs as of late. After Week 10, Smith's ALEX was ranked 11th at plus-1.7, but he eventually settled into a 24th-place finish at plus-0.4. That's still the first time he ranked higher than 30th since 2007, and this might be the team with which he has the best chance to win a Super Bowl yet.

Carson Palmer fell a bit in love with the David Johnson targets this season, while also not hitting the deep balls like he did in 2015, but it's still surprising to see any quarterback in a Bruce Arians offense sit at 29th with minus-0.4 ALEX. Most of the bottom is filled with the passers you would expect, such as Sam Bradford, Blaine Gabbert and a few of the rookies.

Then there is the case of the Los Angeles Rams. Case Keenum had the shortest third-down distance to go, but still struggled to move the ball or attack the sticks. Then when No. 1 overall pick Jared Goff took over, we were treated to an even worse offense, as Goff finished dead last in ALEX, conversion rate, and Short%. He also had the longest situations to convert, and was next to last in DVOA. It was just 65 passes, so we'll wait and see on the progress next year, but a number like minus-2.6 ALEX is pretty alarming already. You're well into the Gabbert Zone at that range.

Finally, ALEX is presented in splits by distance: short (1 or 2 yards to go), medium (3 to 6 yards), and long (7-plus yards). The colors indicate where a player is well above average (darker green) versus below average (darker red) based on standard deviations. Those conversion rates are also shown with a ranking.

Rodgers is the only quarterback to rank in the top 10 in each category, and he's never lower than fifth. Through five games played, Brady was on top in third-and-long conversion rate at 56.5 percent, but he slipped to 11th by season's end. He only slipped from first to second in the other two splits though.

This does illustrate just how bad Goff's season was, and how inefficient Manning has been at times. Meanwhile, Barkley did in fact convert half of his third-and-long passes, but we are talking about 26 throws. We may not get much more Barkley data ever again, but 2017 is going to be very interesting for Goff. His 63.1 percent Short% is the second-highest season since 2006, only behind Blaine Gabbert's 64.9 percent Short% from 2015.

Posted by: Scott Kacsmar on 05 Jan 2017

13 comments, Last at 06 Jan 2017, 2:41pm by Aaron Brooks Good Twin

Comments

1
by Aaron Brooks Go... :: Thu, 01/05/2017 - 4:45pm

I still don't see a compelling argument here that ALEX correlates with success.

Both DET and NYG have a great defensive Alex. Both Rodgers and Eli have great ALEX scores. Ryan has a terrible one. For all of its correlation, you might as well be evaluating hair color.

2
by Scott Kacsmar :: Thu, 01/05/2017 - 5:12pm

I can elaborate on this tomorrow after the playoff preview is finished, but I've never pretended there is a hard correlation between offensive success and ALEX. But there is value to stats that show a player's style of play, and this delivers on that front. I also don't know how much correlation you expect to see for a stat that is showing how far a pass was thrown relative to the marker. Even if it was only in the 0.4-0.6 range, is that really that bad when you're not even accounting for things like accuracy, coverage, pressure or YAC/broken tackles? A lot of the stats we look at can't do much better than that.

And the "career" (back to 2006) ALEX rankings without a doubt show that it's almost impossible to be a good QB and be negative in ALEX. You might pull it off for a season, but it's not a sustainable strategy for success.

3
by doktarr :: Thu, 01/05/2017 - 5:38pm

Can you give a short explanation why Matt Ryan could have a career year with negative ALEX then? Is it just "the receivers he was throwing to (namely Jones and the RBs) were really good after the catch, so he just fed that"?

5
by LyleNM :: Thu, 01/05/2017 - 6:30pm

Lots of success on first and second down.

8
by Bobikus :: Thu, 01/05/2017 - 8:01pm

Look at the distance splits. He has a high conv% rank on short and medium where he has small but positive ALEX, and negative ALEX and low conv% ranking on 3rd and long.

Ryan also was incredibly successful on first down passing, with over 10.6 YPA first downs, compared to 8.36 for Brady and 6.66 for Rodgers (the other two QBs in the MVP conversation). Ryan has 76 less total attempts than Rodgers but only 10 less first down attempts.

To put that first down YPA in perspective, I've looked through every high YPA season by an active player, plus Peyton Manning. Rivers in '10 and Foles in '13 are the only seasons in the '10s to be even within 1.00 of that (both had in the upper 9 range.) To find another QB with 10+ YPA on first downs I had to look all the way back to Ben in '05, who did that on about half the first down attempts.

4
by Aaron Brooks Go... :: Thu, 01/05/2017 - 6:03pm

I'm fine with presenting ALEX as an "aggression" metric. That has value.

But you tend to conflate ALEX with some variant on talent and/or success in your write-ups.

6
by brambo :: Thu, 01/05/2017 - 6:32pm

Agreed. This stat doesn't make a lot of sense to me. The successful QBs are just all over the place with as many on the top half as the bottom.

7
by bravehoptoad :: Thu, 01/05/2017 - 6:50pm

Those in negative ALEX territory seem to be mostly pretty bad, though.

10
by Bobikus :: Fri, 01/06/2017 - 3:02am

It's a stat that looks like it would create a threshold a QB should be over moreso than it has correlation across its full range.

9
by Dales :: Thu, 01/05/2017 - 8:50pm

If Paul Perkins can turn some of those 3rd and 7s into 3rd and 3s then all of a sudden the Giants offense might go from putrid to somewhat less putrid.

11
by tg :: Fri, 01/06/2017 - 11:11am

I'm having difficulty figuring out why this is a more useful stat than 3rd down conversion %. The defensive stat seems illustrative, but there are so many situational factors that would make a throw to an open guy a yard short of the sticks a better throw than one into double coverage one yard past. Game score, location on the field, offensive playcalling, pass-blocking efficiency and receiving personnel matchups would have a huge impact on this, so it seems to me that it should be a team stat for offense. But maybe I'm wrong.

12
by Hoodie_Sleeves :: Fri, 01/06/2017 - 11:30am

I don't see it addressed anywhere (although it's possible I missed it) - but how does ALEX treat incomplete passes?

13
by Aaron Brooks Go... :: Fri, 01/06/2017 - 2:41pm

The stat is called "ALEX", not "BLAKE".