Expected Completion Percentage and Defensive Personnel Faced

| April 15, 2014

Untitled-2Last week, I took a look at the impact defensive personnel has on yards-per-carry.

Today, I’m going to flip over to passers to see how the combination of depth of throw and defensive personnel impacts completion percentage.

Before I get started, I want to give a shout out to one of the top statistical analysts in the football business, Chase Stuart, who gets an assist on this one. Check out Chase’s work over at FootballPerspective.com.

This article will be broken into two sections. First, I’m going to compare the actual and expected adjusted completion percentages of each relevant quarterback from last season, showing my work along the way. Second, I’ll look at the quarterbacks who saw the most/least Base, Nickel, and Dime defenses last season.

Part 1 – Actual vs. Expected

For the purpose of this study, I’ll be using an adjusted version of completion percentage (aC%). The math is as follows:

(Completions + Drops) / (Pass Attempts – Batted Balls – Throw Aways – Throws Disrupted by a Hit – Spikes)

Our first chart shows NFL-wide aimed throws and aC% split into the number of defensive backs on the field.

# of DBs Aimed aC%
1 89 69%
2 170 66%
3 716 67%
4 34793 74%
5 47234 72%
6 16553 68%
7 971 65%
8 37 54%
Total 100563 72%

No shocker here, but we see that completion percentage tends to drop off as the number of defensive backs increase. The “poor” completion percentage against packages with fewer than four defensive backs may seem odd but note that these are mostly throws in and around the end zone.

We see here that 98 percent of throws come against Base, Nickel, or Dime defenses. As a result, I omitted from part one the study throws against packages that included fewer than four or more than six defensive backs. That worked out to a sample of 1,983 throws, which seems like a lot, but, like I said, is only two percent of our total sample.

My next step was to further tweak our “expected” aC% by accounting for depth of throw. I already did a piece on the impact of aDOT on completion percentages this year, so today we’re simply taking the next step and accounting for defense.

To come up with said expected aC%, I ran three different regressions, which looked at depth of target’s relationship to four, five, and six defensive back packages. In the end, I was left with the following

Four Defensive Backs

Untitled-1

 

 

 

 

 

 

 

 

 

 

Five Defensive Backs

Untitled-2

 

 

 

 

 

 

 

 

 

 

Six Defensive Backs

Untitled-3

 

 

 

 

 

 

 

 

 

 

The lines are relatively linear, as you can see. With a polynomial order of three or four (shown), we get slightly better results. The r-squared for each is between .98 and .99, which is extremely useful.

The next step is to apply the formula to each player and examine the results. The next chart shows the quarterbacks who exceeded their expected aC% last season. Only quarterbacks who accrued at least 100 aimed passes are included.

Rk Passer Aimed aC% Exp aC% Diff
1 Aaron Rodgers 283 79% 72% 7.0%
2 Philip Rivers 536 79% 73% 6.0%
3 Josh McCown 207 78% 72% 5.3%
4 Peyton Manning 756 77% 72% 5.0%
5 Jay Cutler 316 75% 71% 4.4%
6 Drew Brees 665 77% 73% 3.9%
7 Nick Foles 323 75% 71% 3.9%
8 Matt Ryan 596 79% 75% 3.8%
9 Ben Roethlisberger 541 75% 72% 2.5%
10 Ryan Fitzpatrick 320 74% 72% 2.0%
11 Cam Newton 449 72% 71% 1.1%
12 Andy Dalton 573 73% 72% 0.8%
13 Jake Locker 171 71% 70% 0.8%
14 Russell Wilson 423 72% 71% 0.7%
15 Matthew Stafford 592 72% 72% 0.3%
16 Tony Romo 503 73% 73% 0.3%
17 Carson Palmer 520 71% 70% 0.1%
18 Matt Flynn 181 74% 74% 0.0%
19 Tom Brady 638 72% 72% 0.0%

Aaron Rodgers paces the league with an actual aC% (79 percent) well above his expected (72 percent). Rodgers’ sample (283 aimed) is a lot lower than that of most starting quarterbacks, but consider that he has a league-high +6.5 percent mark in this category over the past six seasons. Only Michael Vick saw more Nickel defense than Rodgers did last season. In fact, Vick and Rodgers were the only two quarterbacks to see Nickel on more than 80 percent of their drop backs.

No. 2 in the category, Philip Rivers was very good last season. But was he much better than in years past? A bit, but not as much as you might think, especially when you think about San Diego’s offensive struggles in 2012. Rivers’ marks since 2008 in the category are as follows:

2008: +3.4 percent      2009: +5.4 percent      2010: +3.8 percent      2011: +4.7 percent       2012: +2.5 percent      2013: +6.0 percent

We see that obvious dip in 2012, but even the worst mark he’s had over the past six seasons was good enough to rank him pretty well in the category. With San Diego expected to focus on the run in 2014, Rivers is in position to have one of his most efficient campaigns.

Next, we have the quarterbacks who had an actual aC% below their expected.

Rk Passer Aimed aC% Exp aC% Diff
44 Josh Freeman 136 55% 69% -13.6%
43 Kirk Cousins 141 63% 73% -9.6%
42 Jason Campbell 271 69% 74% -5.1%
41 E.J. Manuel 289 68% 73% -4.7%
40 Matthew McGloin 191 66% 70% -4.4%
39 Thaddeus Lewis 150 68% 72% -4.2%
38 Terrelle Pryor 236 69% 73% -4.0%
37 Kellen Clemens 224 68% 72% -3.9%
36 Joe Flacco 575 68% 71% -3.8%
35 Geno Smith 401 67% 71% -3.7%
34 Case Keenum 217 69% 72% -3.4%
33 Brandon Weeden 240 68% 71% -3.1%
32 Eli Manning 505 68% 71% -3.1%
31 Colin Kaepernick 440 68% 71% -2.7%
30 Matt Schaub 322 70% 72% -2.3%
29 Chad Henne 450 73% 74% -1.6%
28 Mike Glennon 376 70% 71% -1.5%
27 Andrew Luck 599 71% 72% -1.5%
26 Alex D. Smith 488 74% 75% -1.1%
25 Michael Vick 128 66% 67% -1.0%
24 Ryan Tannehill 543 71% 72% -1.0%
23 Sam Bradford 238 74% 75% -0.8%
22 Robert Griffin III 415 72% 73% -0.6%
21 Matt Cassel 231 71% 72% -0.6%
20 Christian Ponder 219 73% 73% -0.3%

Josh Freeman was exceptionally poor last season, completing a miserable 63 of 138 (45.7 percent) aimed throws. Freeman was in the upper third of the league in the percent of his throws that came against Base defenses, as well, which means he was even worse than his numbers show on the surface.

Thanks mostly to a Super Bowl title, Joe Flacco gets a pretty good rap. But he really shouldn’t when you consider how he struggles to complete passes. Flacco’s -3.8 percent mark last season placed him in the basement of the league among passers with at least 300 aimed throws. A few players ahead of him include Geno Smith, Eli Manning, Case Keenum, and Brandon Weeden. Many will point to the Raven’s poor receiver situation, but that excuse goes out the window (for the most part) when you consider his career -2.5 percent mark in the category. Of the 16 quarterbacks with at least 2,000 throws over the past six years, Flacco’s mark is the worst. Ryan Fitzpatrick is next closest at -1.4 percent. If there’s one saving grace for Flacco, he did face a lot of Dime defense last season, which certainly makes life a bit harder.

Part 2 – Defensive Personnel Splits

The next part of our study will focus on defensive personnel splits. To kick off, we’ll look at the passers who faced fewer than five defensive backs (99 percent of which is Base) on the highest percentage of their 2013 drop backs. As we learned earlier, it becomes easier for quarterbacks to complete passes as defensive backs are removed from the field. Quarterbacks who dropped back at least 100 times last season are included.

Rk Passer DB <5 >4 <4 Base Nickel Dime >6
46 Colin Kaepernick 601 60% 40% 5% 56% 26% 13% 0%
45 Matthew McGloin 219 46% 54% 2% 44% 20% 34% 0%
44 Terrelle Pryor 336 45% 55% 1% 44% 32% 19% 4%
43 Christian Ponder 291 43% 57% 1% 42% 43% 13% 1%
42 Cam Newton 588 42% 58% 2% 40% 46% 12% 1%
41 Case Keenum 274 41% 59% 1% 40% 44% 12% 3%
40 Matt Cassel 281 40% 60% 1% 38% 56% 4% 0%
39 Jason Campbell 344 39% 61% 1% 38% 34% 23% 4%
38 Andy Dalton 688 38% 62% 1% 37% 45% 16% 0%
37 Kellen Clemens 272 38% 63% 1% 37% 53% 10% 0%
36 Robert Griffin III 530 37% 63% 0% 36% 50% 11% 2%
35 Carson Palmer 609 36% 64% 1% 36% 45% 16% 2%
34 Josh Freeman 156 36% 64% 1% 35% 52% 12% 0%
33 Matt Schaub 383 35% 65% 1% 34% 42% 21% 2%
32 Russell Wilson 583 34% 66% 3% 32% 56% 9% 1%
31 Alex D. Smith 644 34% 66% 1% 33% 50% 13% 3%
30 E.J. Manuel 359 33% 67% 1% 33% 52% 15% 0%
29 Drew Brees 774 33% 67% 2% 31% 52% 15% 1%
28 Blaine Gabbert 105 32% 68% 2% 30% 38% 23% 7%
27 Tom Brady 735 32% 68% 1% 32% 43% 23% 2%
26 Geno Smith 517 32% 68% 1% 31% 47% 21% 0%
25 Jake Locker 213 31% 69% 2% 30% 44% 24% 0%
24 Josh McCown 243 31% 69% 0% 31% 64% 5% 0%

No player had it “easier” than Colin Kaepernick last season. In fact, Kaepernick leads this category for the second consecutive season. San Francisco’s signal caller saw five or more defensive backs just 39.4 percent of the time in 2012, which is just below his 39.6 mark from this past season. No player saw fewer than four defensive backs (five percent) or Base defense (56 percent) more than Kaepernick last season. In fact, it wasn’t even close. He took advantage, putting up a 19:6 TD:INT ratio against four or fewer defensive backs, compared to a 5:5 mark against five or more. Consider this your official Colin Kaepernick red flag.

On the other hand, we have the players who had it tough. These players saw five-plus defensive backs (mostly Nickel and Dime) on the highest percentage of their drop backs.

Rk Passer DB <5 >4 <4 Base Nickel Dime >6
1 Michael Vick 171 4% 96% 0% 4% 85% 11% 0%
2 Aaron Rodgers 361 10% 90% 0% 10% 84% 5% 0%
3 Peyton Manning 807 11% 89% 1% 10% 68% 21% 1%
4 Joe Flacco 675 15% 85% 1% 14% 65% 20% 1%
5 Nick Foles 396 16% 84% 0% 16% 78% 6% 0%
6 Philip Rivers 639 17% 83% 0% 17% 61% 20% 2%
7 Matthew Stafford 668 19% 81% 0% 19% 60% 21% 0%
8 Matt Flynn 236 22% 78% 0% 22% 65% 13% 0%
9 Ben Roethlisberger 639 22% 78% 1% 21% 67% 10% 0%
10 Thaddeus Lewis 183 22% 78% 1% 22% 60% 18% 0%
11 Ryan Tannehill 660 22% 78% 0% 22% 51% 26% 0%
12 Kirk Cousins 162 23% 77% 1% 22% 65% 10% 1%
13 Matt Ryan 701 23% 77% 1% 22% 54% 22% 1%
14 Ryan Fitzpatrick 398 25% 75% 2% 24% 53% 21% 1%
15 Mike Glennon 468 27% 73% 1% 26% 62% 11% 0%
16 Andrew Luck 737 27% 73% 1% 26% 53% 18% 1%
17 Brian Hoyer 104 28% 72% 2% 26% 67% 5% 0%
18 Jay Cutler 387 28% 72% 1% 28% 52% 16% 3%
19 Brandon Weeden 299 28% 72% 0% 28% 54% 18% 0%
20 Eli Manning 596 29% 71% 0% 29% 49% 19% 2%
21 Sam Bradford 284 30% 70% 1% 28% 64% 7% 0%
22 Tony Romo 578 30% 70% 1% 30% 51% 18% 2%
23 Chad Henne 550 31% 69% 1% 29% 44% 24% 2%

Vick paces the field, but he didn’t play a ton last season, so I won’t spend much time on him. He dropped back to pass only six times against a Base defense, compared to 146 against Nickel and 19 against Dime. That’s rough sledding, but the good news is that life will be easier when under center for the run-heavy Jets.

No player saw more Dime than Peyton Manning last season. Big surprise, right? Manning put up a stellar 10:4 TD:INT mark on 167 tries against Dime. He also dropped back six times against Quarter (seven defensive backs). Manning did most of his damage against Nickel, racking up 4,442 yards and a 43:9 TD:INT mark. Manning completed 61 of 83 throws and had a 7:0 TD:INT mark against packages with fewer than four defensive backs.

Our final chart shows the same chart we just saw, but for teams instead of players. The numbers shown are for the 2013 season, including the playoffs.

Rk Team DB <5 >4 <4 Base Nickel Dime >6
1 DEN 826 11% 89% 1% 10% 68% 20% 1%
2 PHI 620 12% 88% 0% 12% 81% 7% 0%
3 BLT 682 15% 85% 1% 14% 65% 19% 1%
4 GB 681 15% 85% 0% 15% 76% 8% 0%
5 SD 639 17% 83% 0% 17% 61% 20% 2%
6 DET 668 19% 81% 0% 19% 60% 21% 0%
7 MIA 666 22% 78% 0% 22% 51% 26% 0%
8 PIT 641 22% 78% 1% 21% 67% 11% 0%
9 ATL 708 23% 77% 1% 22% 54% 21% 1%
10 TEN 611 27% 73% 2% 26% 50% 22% 1%
11 IND 749 28% 72% 1% 26% 54% 18% 1%
12 BUF 605 29% 71% 0% 28% 54% 16% 1%
13 NYG 613 29% 71% 0% 29% 50% 19% 2%
14 CHI 630 30% 70% 0% 29% 57% 12% 2%
15 TB 573 30% 70% 1% 29% 58% 13% 0%
16 NYJ 561 31% 69% 1% 30% 49% 20% 1%
17 JAX 658 31% 69% 1% 30% 43% 24% 3%
18 DAL 630 32% 68% 0% 31% 51% 16% 1%
19 NE 735 32% 68% 1% 32% 43% 23% 2%
20 NO 774 33% 67% 2% 31% 52% 15% 1%
21 CLV 747 33% 67% 1% 32% 47% 18% 2%
22 KC 689 33% 67% 1% 32% 50% 15% 3%
23 WAS 692 33% 67% 0% 33% 53% 11% 2%
24 SL 557 34% 66% 1% 32% 58% 8% 0%
25 SEA 597 35% 65% 3% 32% 56% 9% 1%
26 ARZ 611 36% 64% 1% 36% 46% 16% 2%
27 CIN 690 39% 61% 1% 37% 45% 16% 0%
28 HST 681 39% 61% 1% 38% 42% 17% 2%
29 MIN 627 40% 60% 1% 39% 52% 8% 0%
30 CAR 588 42% 58% 2% 40% 46% 12% 1%
31 OAK 598 46% 54% 1% 45% 28% 23% 3%
32 SF 603 61% 39% 5% 56% 26% 13% 0%
NFL 20950 30% 70% 1% 29% 53% 16% 1%

No team saw fewer than five defensive backs more than Denver, but Philadelphia was just behind. In fact, Eagles’ passers faced Nickel 81 percent of the time, which easily paced the league. Nick Foles will need to get used to facing an extra cornerback on a majority of his attempts.

I included the league average in the bottom row. Teams had five-plus defensive backs on the field for 70 percent of pass plays last season. That includes 53 percent Nickel and 16 percent Dime. Note that only one percent of plays, most of which happen at the goal line, come with fewer than four defensive backs on the field.

Follow Mike Clay on Twitter: @MikeClayNFL

  • Fusue Vue

    Is facing more DBs that much more of a “negative” when the opposing defense is simply matching up to your own offense having the same amount of receivers on the field?

    • Scott C

      I’m not sure. I don’t see anything in this analysis that accounts for the massive selection bias that teams that have good passing personnel place more pass catchers on the field.

      Comparing the defensive personnel to the offensive would help: how often does an offense face Dime when they have 3 receivers? Passing against a dime with 2 RBs 1 TEs (actually lined up as a TE) and 2 WR is different than when in shotgun with 4 WR spread out.

      • Mike Clay

        Hey guys – I did consider this (In fact, I studied it as part of the YPC version of this article – See link below). I didn’t see much of a difference in the YPC article. I wanted to do it for C%, as well, but I only had personnel available, not the # of route runners. Eg. Jimmy Graham in the slot would show up as a TE, the same as an in-line blocking TE who didn’t run a route. Once I have the correct data, I can look at that a bit closer. Thanks for reading/commenting.

        https://www.profootballfocus.com/blog/2014/04/09/the-impact-of-defensive-packages-on-yards-per-carry/

  • Robin Strick

    I wonder if the teams like Clev, NO, Hou, NE who are near the top in regular season pass attempts but still seeing base packages, 30+% of the time should perhaps get a more dynamic WR2. For as much as these teams are passing if defensive teams feel they can match up in a base package the offensive running game and pass blocking will be constant pressure. This may tie in to Ratcliffe’s article showing the QB pressure % that Brees/Brady/Schaub faced last year.

  • jack_sprat2

    I’d definitely like to see what correlates strongly with %-of-DIME coverages faced. A-Rod, in particular, seems to be getting a free ride there. Is it relative success on early downs, higher passing % on early downs, strong breakaway running threats, low number of credible receiving threats, or what? And, how do my Lions go about getting that for Staff_9?