ALEX

A weekly look at the most aggressive and conservative quarterbacks on third downs.

2018 ALEX: The Early Results

by Scott Kacsmar

As we head into Week 6 of this NFL season, it's time to take a first look at this year's ALEX. No, that's not referring to Washington quarterback Alex Smith, but we may spend more time on him here than ESPN did on Monday night during the Drew Brees marathon.

We're talking about 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 4-yard pass on third-and-12, then that would be -8 ALEX. A pass thrown beyond the first-down marker would earn a positive ALEX figure.

As you will see, we have identified some quarterbacks with a strong year-to-year tendency for aggressive or ultra-conservative play. Since 2006, quarterbacks with at least 200 passes in a season have seen a year-to-year correlation of 0.36 in ALEX for all downs, which is right between the year-to-year correlation for stats such as passer rating (0.35) and yards per attempt (0.38). That's also very close to aDOT (0.40) and passing DVOA (0.40). Most quarterbacks are wise enough to attack the sticks on fourth down, which is why ALEX is best applied on third downs to gauge a quarterback's aggressiveness.

When we looked at ALEX on third down through Week 4 last season, the league-wide average was +0.7, the lowest since 2009. By season's end, the average ALEX was +1.1, which still tied 2009 for the lowest in the last decade. Through Week 5 of 2018, ALEX is again at +0.7, so not much has changed in this regard from 2017.

How about the quarterbacks themselves? Quite a few changed teams, several returned from injury, and four rookies are starting now. The following table, which is now sortable, looks at every quarterback this season with at least 20 third-down passes. Why 20? We wanted to make sure both San Francisco quarterbacks (Jimmy Garoppolo and C.J. Beathard) made the cut since neither was at 25 attempts yet. We also were able to include Sam Bradford so that Arizona is represented.

The 34 quarterbacks are ranked from highest to lowest ALEX. Also included are their conversion rate, DVOA, and average need yards for a first down. Finally, Short% is the percentage of third-down attempts that were thrown short of the sticks (negative ALEX). ALEX and Short% generally have strong correlation, but Short% is a good way to account for outlier plays like screens on third-and-forever or bombs on third-and-1. This is especially important early in the season when the quarterbacks have only averaged 40 third-down passes through Week 5. For that reason, we will not include a split by distance table until the midseason report. All 2018 numbers are subject to change upon review of game charting and the removal of passes thrown away or batted down at the line.


2018 Third Down ALEX (Weeks 1-5; Min. 20 Passes)

Rk Player Team ALEX CONV% Rk DVOA Rk Passes Avg. Need Rk Short% Rk
1 Tom Brady NE 3.8 41.5% 15 43.6% 13 41 6.7 30 36.6% 10
2 Patrick Mahomes KC 3.6 50.0% 1 64.9% 5 38 8.1 14 36.8% 11
3 Marcus Mariota TEN 3.4 36.7% 24 23.4% 25 30 6.1 34 26.7% 2
4 Ryan Fitzpatrick TB 3.0 41.9% 13 73.9% 3 31 8.4 10 25.8% 1
5 Tyrod Taylor CLE 2.8 19.2% 34 -12.8% 29 26 8.6 4 57.7% 33
6 Joe Flacco BAL 2.8 37.7% 21 52.0% 9 53 7.4 23 32.1% 5
7 Sam Bradford ARI 2.2 28.6% 32 -32.5% 33 21 7.0 28 42.9% 18
8 Russell Wilson SEA 2.0 32.5% 28 23.4% 24 40 7.8 19 35.0% 6
9 Sam Darnold NYJ 2.0 39.5% 17 70.9% 4 43 8.5 7 37.2% 12
10 Cam Newton CAR 1.9 38.7% 19 -19.3% 31 31 6.5 32 35.5% 8
11 Ryan Tannehill MIA 1.8 29.7% 31 14.3% 27 37 9.0 3 54.1% 30
12 Josh Allen BUF 1.7 33.3% 26 -22.9% 32 39 9.8 1 38.5% 13
13 Alex Smith WAS 1.7 37.5% 22 37.6% 15 40 7.5 21 40.0% 15
14 Matthew Stafford DET 1.7 45.8% 6 25.7% 22 48 7.1 26 31.3% 4
15 Case Keenum DEN 1.3 31.3% 29 24.2% 23 48 8.2 12 41.7% 17
16 Philip Rivers LAC 1.0 39.1% 18 11.4% 28 46 7.0 27 47.8% 25
17 Andy Dalton CIN 0.9 43.6% 10 36.0% 17 39 8.4 9 35.9% 9
18 Deshaun Watson HOU 0.8 37.3% 23 49.6% 11 51 7.6 20 35.3% 7
19 Aaron Rodgers GB 0.8 41.7% 14 110.7% 1 48 9.2 2 47.9% 26
20 Jared Goff LAR 0.5 42.9% 11 27.2% 20 35 8.3 11 28.6% 3
21 Derek Carr OAK 0.1 45.2% 7 55.6% 7 42 6.8 29 47.6% 24
22 Carson Wentz PHI 0.1 33.3% 27 25.8% 21 27 8.1 13 44.4% 20
23 Ben Roethlisberger PIT 0.1 34.7% 25 30.3% 19 49 8.0 17 46.9% 23
24 Matt Ryan ATL -0.3 49.0% 2 62.5% 6 49 8.4 8 38.8% 14
25 Baker Mayfield CLE -0.4 38.2% 20 52.9% 8 34 8.5 6 41.2% 16
26 Kirk Cousins MIN -0.5 48.3% 4 48.7% 12 58 6.3 33 55.2% 31
27 Andrew Luck IND -0.6 48.5% 3 36.4% 16 66 7.2 25 43.9% 19
28 Mitchell Trubisky CHI -0.6 40.0% 16 14.8% 26 30 6.6 31 60.0% 34
29 Blake Bortles JAX -0.7 43.6% 9 33.1% 18 55 7.3 24 52.7% 29
30 Drew Brees NO -1.5 42.4% 12 39.3% 14 33 7.4 22 48.5% 27
31 Eli Manning NYG -1.7 43.8% 8 93.7% 2 48 8.1 15 56.3% 32
32 Jimmy Garoppolo SF -1.9 30.0% 30 -13.9% 30 20 8.6 5 45.0% 21
33 C.J. Beathard SF -2.5 47.6% 5 49.9% 10 21 8.0 18 52.4% 28
34 Dak Prescott DAL -2.6 23.1% 33 -43.7% 34 39 8.1 16 46.2% 22

Aaron Rodgers and Eli Manning Lead Unusual Results

Now it has only been five weeks and some of these quarterbacks have only started three games for various reasons. However, I cannot remember running these numbers and seeing such odd results as far as conversions and DVOA go relative to ALEX. Out of the 16 quarterbacks who are at least +1.0 in ALEX, only Patrick Mahomes and Matthew Stafford rank in the top 12 in conversion rate. When we ran these numbers early last year, the top six quarterbacks (and 11 of the top 15) in conversion rate were at least +1.0 in ALEX. The correlation between ALEX and DVOA here is 0.02. The correlation between ALEX and conversion rate is -0.12. At the league-wide level from 2006 to 2017, the correlation between ALEX and conversion rate was 0.70, so this is an odd beginning to the season.

You know something's off when Aaron Rodgers is only 19th in ALEX (+0.8), but first in DVOA (110.7%). Rodgers has ranked in the top six in ALEX in every season since 2008. Maybe he will still get there by season's end, but he has taken a less aggressive approach and has had decent results in the context that his average third-down throw is coming with 9.2 yards to go (second highest).

If you're trying to make sense of how his DVOA can be the highest, but his conversion rate is only 14th, consider the 75-yard touchdown to Randall Cobb to beat Chicago's No. 1 defense in Week 1. Without the opponent adjustment and that huge play, Rodgers' DVOA is down to 91.1% (VOA down to 79.7%). That would put Rodgers behind Eli Manning in DVOA, which is another oddity this season. Manning has the fourth-lowest ALEX (-1.7) and third-highest Short%, but he's still eighth in conversion rate and second in DVOA. Rookie running back Saquon Barkley has helped a little with four double-digit YAC catches on third down, but it helps that Manning is the only quarterback who is completing over 70 percent of his passes on third down and doesn't have an interception yet.

Tom Brady and Patrick Mahomes

What a matchup we have coming on Sunday night between the Chiefs (5-0) and Patriots (3-2). Tom Brady and Patrick Mahomes are the top two in ALEX, but Mahomes is the league's only quarterback converting 50 percent of his third-down passes into first downs. For Brady, his trend since 2016 has been to throw deeper, especially while under pressure and on third down. He has been in the top 10 in ALEX since 2016 after not placing that high from 2010 through 2015.

Last year in this article, I wrote that "we're unlikely to see a big change in third-down ALEX for Kansas City's passing game until Patrick Mahomes takes over, which may not be for quite a while now." Well, despite a career year from Alex Smith, it was time to go to Mahomes this season, and he has been fantastic in 2018. It's really not a surprise to see him be a high-ALEX quarterback. I wrote this in his player comment from Football Outsiders Almanac 2017:

The Anti-Alex should be an interesting quarterback to watch when he gets his chance. In using air yards data from ESPN Stats & Info, we found that Mahomes' 2016 ALEX on third down was +4.2, the type of number that would lead the NFL in most seasons. He's an aggressive playmaker, but still avoided interceptions fairly well despite the number of shootouts in which he was involved at Texas Tech. Mahomes had only one game in college with more than two interceptions.

Mahomes actually just threw his first two picks of the season against Jacksonville, but has been very low on turnovers despite all the high-scoring games the Chiefs have played in the season's first five weeks. The most common comparison for Mahomes has been Rodgers with his ability to make accurate throws from inside and outside the pocket. We'll see if Mahomes can sustain Rodgers' career level of balancing a high ALEX with few turnovers, but so far, so good.

Mr. Smith Goes to Washington

We are obligated to mention Alex Smith in any ALEX article, but his results are among the most mundane on the table. His first season in Washington ranks from 13th to 22nd in the five listed stats. Smith made a couple of nice plays on third down against the Saints in the first quarter on Monday night, but overall he has produced mediocre results while trying to be a little more aggressive than his usual standards.

Marcus Mariota and Tampa Bay's Duo

Marcus Mariota and Jameis Winston have both been in the top 10 in ALEX the last few seasons. Their latest shared trait is not being available every week to their teams. Mariota has missed time with injuries again, giving way to Blaine Gabbert, who actually didn't have a horrifically low ALEX like he always has in the past. Sure, it was -0.9, but he's usually more than 2 full yards short of the sticks. Mariota has had some room to be more aggressive as he faces the shortest third-down throws in the league (average need: 6.1 yards).

In Tampa Bay, Ryan Fitzpatrick stepped up while Winston was suspended and started the year with three 400-yard passing games before promptly giving the job back to Winston. Fitzpatrick only threw a league low 25.8 percent of his passes short of the sticks on third down. Last season, Winston finished second in Short% at 22.3 percent. I think we'll continue to see Tampa Bay attack down the field, but the next time we update this table, it will be with Winston's name near the top and Fitzpatrick nowhere to be found. The only question is whether Winston will keep up Fitzpatrick's efficiency, as the Harvard alum was third in DVOA. Winston finished fifth in DVOA on third-down throws last season. It's an underrated aspect of his game.

The 2016 Draft Class' Wild Ride

Remember when Jared Goff was drafted ahead of Carson Wentz, but fourth-round pick Dak Prescott was the best rookie quarterback in 2016? Last year, Wentz seemed to take over in that trio, and now Goff is the one leading the NFL in passing DVOA while Prescott looks like a fourth-round pick and Wentz is trying to acclimate to his surgically repaired ACL. All three seem heavily affected by the quality of their supporting cast, and they're all interesting ALEX cases right now.

Goff has gone up in ALEX each season, moving from dead last as a rookie (-2.6) to 32nd in 2017 (+0.1), and he's currently 20th this year (+0.5). What's interesting is Goff has the third-lowest Short% (28.6 percent) despite only ranking 20th in ALEX. That's because he has two plays that were -24 ALEX or worse; there have only been 11 such plays this season. He faced a third-and-30 against the Chargers and threw a pass 3 yards behind the line of scrimmage to Robert Woods for -33 ALEX, the second-most negative ALEX play this season. You may remember that Goff got a 52-yard touchdown to Woods on a -34 ALEX pass last season against the Giants to convert a third-and-33. No such luck this season, but Goff has been more phenomenal on early downs in Sean McVay's offense.

Wentz was only 27th in ALEX (+0.0) as a rookie, but took a much more aggressive approach in 2017 with great results on third down: second in ALEX (+3.3) and conversion rate (49.2 percent) and first in DVOA (103.8%). It has only been three games post-ACL this year, but thus far Wentz has gone back to his rookie level in ALEX (+0.1) and is 27th in conversion rate. At least the numbers are up from when Nick Foles started the first two games. Foles' ALEX was -2.3 and his average need was 10.4 yards. Part of that was the outlier of the year when Foles faced a third-and-41 in Tampa Bay and threw a -45 ALEX pass to put the drive out of its misery.

Speaking of misery, these numbers look terrible for Prescott and Dallas. He's last in ALEX and DVOA and next to last in conversion rate. Prescott was trending this direction last season when he was 27th in ALEX (+0.2), but this year he's not even trying to be aggressive on third down. His average pass has traveled 5.6 yards, the lowest in the league. No one has yet stepped up to replace Jason Witten and Dez Bryant, but running back Ezekiel Elliott doesn't seem like the answer either as a receiver. On 13 third-down targets, Elliott only has 40 yards and two first downs.

Mitchell Trubisky and Deshaun Watson: Take Two

Deshaun Watson has still been a vertical passer for Houston this year, but his ALEX has dropped from +2.1 as a rookie to the middle of the pack this year at +0.8. His conversion rate has also dropped from 41.9 percent to 37.3 percent. Like Wentz, he is still in the early stages of a return from an ACL injury.

As for Mitchell Trubisky, he was dead last in ALEX (-2.4) as a rookie. Even with all the new weapons the Bears added, I did not think Trubisky would make huge strides in ALEX since Matt Nagy comes from the Andy Reid coaching tree, and even Alex Smith only ranked 22nd in his best season last year. Through four games this year, Trubisky is 28th in ALEX (-0.6) and has the highest Short% in the league (60.0 percent). Anything over 60 percent gets into the scary Blaine Gabbert Zone. Trubisky's first three games this season weren't encouraging at all, but at least the Bears schemed receivers open for a six-touchdown game against Tampa Bay in Week 4. We'll see where this one goes, but odds are Trubisky finishes on the lower end on most of these stats.

The 2018 Rookies

It's still very early for the rookies. Josh Rosen wasn't included in the table since he only has 15 third-down passes in Arizona, but his ALEX is +0.5. Sam Darnold was the only Week 1 rookie starter, and he fares the best on this table. Darnold has proven the preseason criticism that he would be a dink-and-dunk passer in the NFL wrong with the eighth-highest aDOT (8.7 yards) with the Jets. He's one of nine quarterbacks with an ALEX over +2.0 and he ranks fourth in DVOA.

In Cleveland, Baker Mayfield (-0.4 ALEX) is ranked 20 spots lower than Tyrod Taylor (+2.8) in ALEX. However, Mayfield's conversion rate is almost double that of Taylor's, which is last in the league at 19.2 percent. Taylor is also a great example of why Short% is worth noting. Taylor is fifth in ALEX, but next to last in Short% due mainly to two plays. Before he was replaced by Mayfield against the Jets in Week 3, Taylor threw a 48-yard bomb on third-and-6 to Antonio Callaway for the biggest ALEX play of the season so far on third down (+42). Taylor also had a +39 ALEX throw to Jarvis Landry against the Steelers in Week 1, so he has two of the three biggest ALEX plays this season. Only one of those throws (the Landry bomb) was complete.

As is usually the case, a poor rookie quarterback in a terrible offense has the longest average distance to go on third down. That's Josh Allen in Buffalo this year, who has thrown his average third-down pass with 9.8 yards to go. Allen has been a vertical passer this year, but just not a very accurate one, as any college scouting report should have noted. He's currently 26th in conversion rate and 32nd in DVOA.

Sam Bradford and Joe Flacco: Old Dogs, New Tricks?

It only took three games for Sam Bradford to lose his job in Arizona to rookie Josh Rosen. There's a pectoral injury listed for Bradford, but the reality was he led the offense to 20 points in three games. The surprising result is that his ALEX was actually +2.2 as Bradford has almost always ranked in the bottom quarter of the league in ALEX. However, his new strategy did not work, as he ranked 32nd in conversion rate and 33rd in DVOA. Maybe teams will finally wise up before offering him another $20 million to not produce.

The Ravens clearly had some concerns about Joe Flacco's production, which is why they drafted Lamar Jackson in the first round. Jackson is not about to take the job from Flacco, who actually ranks sixth in ALEX (+2.8) and ninth in DVOA so far. Flacco was 25th in ALEX in each of the previous two seasons, and a +2.8 season would be the highest of his career if he sustains it.

Obligatory Jimmy Garoppolo Mention

It's only 41 plays total, but there are interesting numbers from Kyle Shanahan's 49ers again. Jimmy Garoppolo (-1.9) and C.J. Beathard (-2.5) are in the bottom three in ALEX and are close together again like they were in 2017 -- albeit they were around +1.4 then. But Beathard actually has a much better conversion rate and DVOA. After being the toast of the town last year, Garoppolo's brief 2018 season is frowned upon by virtually all advanced metrics. He'll have to wait until 2019 after a torn ACL recovery to prove his value again.

Conclusion: Should We Be Including Sacks?

We'll wrap up with some food for thought. This was Drew Brees' week as he set the NFL record for career passing yardage, but he has the fifth-lowest ALEX (-1.5) after finishing 2017 with the fifth-lowest ALEX (-0.2). Brees has been all over the map with ALEX in New Orleans, and obviously it hasn't hurt the Saints from being the No. 1 scoring offense this year. However, I noted last year that Brees finished 14th in conversion rate after ranking in the top eight in every season from 2006 to 2016. He's 12th this year, so there has been some decline in moving the chains on third down with this more conservative approach.

The other thing about Brees' 2018 that won't show up here is that he has a league-high sack rate of 17.1 percent on third down. Shockingly, Alex Smith (2.2 percent) and Goff (2.7 percent) have the lowest sack rates on third down. While we're always looking at the aggressiveness of throws on third down, should we be factoring in sacks too? By not releasing the ball, we could give the quarterback a "0" for his air yards, making a sack on third-and-10 a -10 ALEX play. That would make Brees' season look even more conservative while making Smith appear more aggressive.

Then again, it's all about perception. If Ben Roethlisberger or Andrew Luck hold the ball a long time and take a sack, they are lauded for not giving up on the play in an effort to make something happen. They were actually being aggressive. However, if Smith or Derek Carr gets sacked, they would be viewed as being too passive and too scared to throw a pick.

Perception can be really hard (if not impossible) for players to shake, but hopefully we'll continue gathering data to accurately present differences in playing styles among quarterbacks.

Comments

4 comments, Last at 12 Oct 2018, 1:02am

1 Re: 2018 ALEX: The Early Results

by Mountain Time ---- formerly Ninjalectual // Oct 11, 2018 - 6:09pm

For anyone else who hadn't heard of aDOT: https://www.profootballfocus.com/news/fantasy-introducing-%25e2%2580%259...

Login or Register to post comments

2 Re: 2018 ALEX: The Early Results

by techvet // Oct 11, 2018 - 9:04pm

1-2 years ago, the NY Times ran a study that showed ranked teams by WR speed. I remember because the Packers finished dead last. This was based on the chips that are now embedded in the shoes. Is there a correlation between WR speed and ALEX?

Login or Register to post comments

4 Re: 2018 ALEX: The Early Results

by RobotBoy // Oct 12, 2018 - 1:02am

I found the article:
https://www.nytimes.com/2018/01/04/sports/football/nfl-speed-leonard-fou...
Chips were embedded in the shoulder pads.
Don't know how a single speed metric would be all that helpful as an analytic tool. Ability to rapidly change speeds, and direction, would seem to be far more important for receivers (and then the ability to catch the damn ball). If straight-line speed, even straight-line game speed, mattered all that much, different guys would be getting the big checks.
Last season, the Patriots had three of the fastest gameday DB's in Malcolm Butler, Patriots (22.07); Johnson Bademosi, Patriots (21.72); Devin McCourty, Patriots (21.66). Note that the list doesn't even include their #1 corner. (Granted, McCourty's ability to cover ground has been an asset).
This is the list you mentioned for 'team average speed at receiver'.

Los Angeles Rams* 13.32 m.p.h.
Jacksonville Jaguars* 13.09
Cleveland Browns 13.08
Buffalo Bills* 12.97
Los Angeles Chargers 12.93
Chicago Bears 12.90
New England Patriots* 12.84
Seattle Seahawks 12.83
Indianapolis Colts 12.83
Kansas City Chiefs* 12.71
Washington Redskins 12.66
Dallas Cowboys 12.64
San Francisco 49ers 12.58
Tennessee Titans* 12.58
Pittsburgh Steelers* 12.56
Miami Dolphins 12.52
Baltimore Ravens 12.51
Minnesota Vikings* 12.50
Philadelphia Eagles* 12.50
Atlanta Falcons* 12.49
Arizona Cardinals 12.48
Cincinnati Bengals 12.47
New Orleans Saints* 12.46
Carolina Panthers* 12.40
Detroit Lions 12.38
Houston Texans 12.33
Tampa Bay Buccaneers 12.29
New York Jets 12.25
Denver Broncos 12.24
New York Giants 12.20
Oakland Raiders 11.96
Green Bay Packers 11.74
* playoff teams

Login or Register to post comments

3 Re: 2018 ALEX: The Early Results

by lightsout85 // Oct 12, 2018 - 12:12am

When sorting by conversion% I saw a lot of short-throwing QBs, so I thought I'd plot the other measures (ALEX, Short% & AvgNeed) vs Conv% and while non had good correlation (as we'd expect), AvgNeed actually was relatively much more correlated (R^2=0.08 to 0.01 for ALEX & 0.02 for Short%). It'd be interesting (once the sample sizes are larger) to see it split by QB &/or yards-to-go, conversion% when throwing short vs throwing past the sticks. (I'd have to guess that despite lower comp%, throwing farther would result in more conversion, since median YAC gains are very small).

I also thought maybe QBs could be throwing shorter *because* they didn't have as far to go (some magical distance where they feel safer trusting YAC than a deeper completion), but R^2 here (need vs short%) was 0.0004.

Edit: I actually started reading after the table & see conversion correlations was the first thing you talked about, haha (different numbers, but still the conclusion that weirdly deeper passers aren't converting well so far).

Login or Register to post comments