Writers of Pro Football Prospectus 2008

Most Recent FO Features

SmithAle05-QB.jpg

» 2017 ALEX: Midseason Report

The latest ALEX update looks at the recent draft class that is struggling, the unusual Chicago strategy, and what's gotten into Alex Smith? We also looked at Tyrod Taylor's declining ALEX, but rising conversion rate that Buffalo just sent to the bench.

19 Jan 2016

Divisional Round Quick Reads

by Vincent Verhei

The leading rusher of the divisional round, Jonathan Stewart of Carolina, gained 106 yards on 19 carries against Seattle. That's an average of 5.6 yards per carry, which is very good, but not necessarily an accurate description of Stewart's day. It's not as if he was reeling off 5-plus yards every carry. In fact, only two of his runs gained more than 5 yards: a 59-yard scamper on his first carry of the game, and an 8-yard gain later in the first quarter. The rest of his 17 carries all gained 5 yards or less. That's in large part because the Panthers spent most of the second half willing to let Stewart plow into a wall of Seahawks for short, clock-killing gains, but for right now we are interested only in the distribution of Stewart's yardage, not the cause of that distribution.

Regardless, the distribution of Stewart's carries was hardly unique. The week's second-leading rusher, Eddie Lacy of Green Bay, gained 89 yards on 12 carries, an average of 7.4 yards per rush, but he actually only gained 7 yards twice -- back-to-back carries of 14 and 61 yards in the third quarter.

The third-leading rusher? That would be Denver's C.J. Anderson, who gained 72 yards on 15 carries against Pittsburgh. That's an average of 4.8 yards per carry. Anderson actually had four runs that gained 4.8 yards or more, the longest a 34-yarder.

My point here is that while average yards per carry can be a good measure of a running back's total production, it's not always an indicator of consistency. Yes, we can use success rate to measure that sort of thing, but I wanted to try something new this week. Specifically, I wanted to look at each player's median yards per carry -- not the longest run, nor the shortest, but the carry exactly in the middle -- and see if that revealed any useful information. And in this case, the answer is yes. Though Stewart only had the two long runs against Seattle, he was usually achieving at least moderate production -- his median carry gained 3 yards. Anderson had a tougher time against Pittsburgh, with a median gain of 2. Lacy, meanwhile, mostly struggled against Arizona, with a median carry of just 1 yard.

While we find different results in individual games, things quickly get muddled over the course of an entire season. I found the median yards per carry of all 44 running backs with at least 100 carries this season. Thomas Rawls, Giovani Bernard, and Charles Sims tied for the NFL lead with a median gain of 4 yards per carry. Lamar Miller, Isaiah Crowell, and Antonio Andrews tied for last place at 2; Matt Jones was just ahead of them at 2.5. The other 37 runners all had a median gain of 3 yards. So no, there wasn't much useful here. There's just not enough granularity in yardage to distinguish most running backs from the pack. Perhaps if we went back through the records, we would find Adrian Peterson or Terrell Davis or Barry Sanders broke 5 in a season, and just maybe if we had the data for Jim Brown at his best we might hit 6. Otherwise, everyone comes out looking the same.

However, we have more than just the yardage gained on each carry available to us. We also know the DYAR gained on each run, and there the differences from one carry to the next are fine enough to show which runners really were having consistent success, and which were falling below standards over and over again.

The following table lists each qualifying running back along with their average gain, average DYAR per carry, and median DYAR per carry. DYAR per carry is not exactly the same as DVOA, because the baselines are different, but it is very close -- the correlation between the two is .999. Players are sorted from best median DYAR (Ryan Mathews) to worst (Matt Jones). Not surprisingly, there is a high correlation between median DYAR and success rate -- .785 -- but it's not a perfect match.


Running Backs by Median DYAR Per Carry, 2015 (minimum 100 rushes)
Name Runs Yds Avg Rk DYAR Avg.
DYAR
Rk Median
DYAR
Rk Suc% Rk
24-R.Mathews 108 539 4.99 2 133 1.23 3 1.06 1 53% 5
23-R.Jennings 195 863 4.43 16 117 0.60 13 0.86 2 56% 4
34-T.Rawls 147 830 5.65 1 217 1.47 1 0.68 3 62% 1
34-C.Sims 107 529 4.94 3 49 0.46 16 0.65 4 57% 2
31-D.Johnson 125 581 4.65 9 131 1.05 4 0.58 5 56% 3
26-L.Bell 113 556 4.92 4 160 1.42 2 0.46 6 50% 10
25-G.Bernard 154 730 4.74 7 129 0.84 7 0.39 7 49% 13
33-J.Langford 148 537 3.63 41 120 0.81 8 0.28 8 47% 20
22-M.Forte 218 898 4.12 23 194 0.89 6 0.28 9 48% 17
28-C.Hyde 115 470 4.09 24 61 0.53 14 0.27 10 49% 14
24-M.Lynch 111 417 3.76 35 44 0.40 19 0.24 11 50% 11
29-D.Johnson 104 379 3.64 39 30 0.29 25 0.21 12 45% 26
29-J.Forsett 151 641 4.25 19 117 0.78 9 0.17 13 46% 22
35-C.West 160 634 3.96 31 78 0.49 15 0.16 14 45% 27
34-De.Williams 200 907 4.54 13 184 0.92 5 0.14 15 50% 9
Name Runs Yds Avg Rk DYAR Avg.
DYAR
Rk Median
DYAR
Rk Suc% Rk
25-L.McCoy 202 896 4.44 15 140 0.69 11 0.08 16 47% 21
20-D.McFadden 239 1089 4.56 12 83 0.35 22 0.01 17 50% 8
32-J.Hill 223 792 3.55 42 85 0.38 21 0.01 18 49% 12
21-A.Abdullah 143 597 4.17 20 -2 -0.01 37 -0.06 19 51% 7
37-J.Allen 137 518 3.78 34 29 0.21 29 -0.08 20 47% 19
29-L.Blount 165 703 4.26 18 64 0.39 20 -0.10 21 52% 6
28-M.Gordon 184 641 3.48 43 -67 -0.36 43 -0.12 22 43% 32
28-A.Peterson 327 1490 4.56 11 143 0.44 17 -0.12 23 45% 25
22-D.Martin 288 1402 4.87 5 80 0.28 26 -0.20 24 48% 16
22-M.Ingram 166 769 4.63 10 108 0.65 12 -0.22 25 45% 30
24-T.Yeldon 182 740 4.07 26 39 0.22 28 -0.22 26 42% 38
23-C.Johnson 196 812 4.14 22 25 0.13 31 -0.23 27 47% 18
27-E.Lacy 187 758 4.05 28 3 0.01 35 -0.26 28 49% 15
44-J.Starks 148 601 4.06 27 -46 -0.31 42 -0.39 29 43% 37
28-J.Stewart 242 988 4.08 25 23 0.10 34 -0.40 30 43% 33
Name Runs Yds Avg Rk DYAR Avg.
DYAR
Rk Median
DYAR
Rk Suc% Rk
28-A.Blue 183 698 3.81 33 61 0.33 24 -0.41 31 45% 29
30-T.Gurley 229 1102 4.81 6 171 0.75 10 -0.44 32 43% 36
24-D.Freeman 264 1062 4.02 29 89 0.34 23 -0.49 33 46% 23
34-I.Crowell 185 706 3.82 32 36 0.20 30 -0.55 34 41% 40
23-F.Gore 260 968 3.72 36 0 0.00 36 -0.56 35 40% 42
26-A.Andrews 143 520 3.64 40 -6 -0.04 38 -0.60 36 42% 39
26-L.Miller 194 872 4.49 14 81 0.42 18 -0.69 37 43% 34
33-C.Ivory 247 1073 4.34 17 -31 -0.13 39 -0.72 38 43% 35
23-R.Hillman 207 863 4.17 21 22 0.10 33 -0.74 39 45% 28
28-L.Murray 266 1066 4.01 30 33 0.12 32 -0.81 40 39% 43
29-D.Murray 193 704 3.65 38 -29 -0.15 40 -0.89 41 45% 31
22-C.Anderson 152 720 4.74 8 34 0.23 27 -0.96 42 41% 41
46-A.Morris 202 751 3.72 37 -54 -0.27 41 -1.10 43 39% 44
31-M.Jones 144 490 3.40 44 -95 -0.66 44 -1.21 44 46% 23

As always, the most interesting players here are those who rank much higher are lower in one category than they do in others. Take Chicago's Jeremy Langford -- 41st in yards per carry and 20th in success rate, but eighth in both average and median DYAR per carry. He had zero fumbles, which affects his high DYAR rankings, and he was outstanding near the goal line, with six carries inside the 3-yard line resulting in five touchdowns and a first-and-goal at the 1. He converted seven of eight power opportunities on the year.

It was a similar story for Marshawn Lynch. The Seahawks' veteran wasn't as explosive as he has been in the past, but he never fumbled, and he still had a high success rate.

Cincinnati's Jeremy Hill had a terrible average, but was above average in all other categories. That mostly shows a lack of explosiveness -- he had just one 20-yard run all year, and even that gained just 23 yards.

Detroit's Ameer Abdullah was good in most metrics, especially success rate, where he was seventh. His yards per carry and median DYAR were also good, but his average DYAR was just 37th. That's because when Abdullah did have a bad play, they were very bad, with four fumbles in only 143 carries. (Also, a random observation: Abdullah needs more short-yardage carries. He converted all five of his power opportunities, and more than that, those successes include gains of 24, 27, and 30 yards!)

Melvin Gordon was next to last in both average gain and average DYAR, but his success rate and median DYAR were much better. This is the profile of a runner who moves the ball fairly well, but combines a lack of home-run speed with a tendency to put the ball on the ground, with five fumbles in 184 runs.

Todd Gurley is interesting -- top ten in both average gain and average DYAR, but bottom ten in success rate and not much better in median DYAR. In essence, he is the opposite of Gordon -- if you feed him the rock, you will have to wait a long while for him to break a big run, but those big runs will be big. And you don't need to worry about ball security -- he had only three fumbles in 229 runs.

Lamar Miller is an interesting case. As we mentioned, he was one of the very few runners with a median gain of less than 3 yards per carry. However, his average yardage and average DYAR per carry were quite good, even if his median DYAR and success rate were lacking. This is because, while he didn't have as many good runs as his peers, his good runs usually went longer. Yes, his median gain was only 2 yards, but on those 94 runs where gained 3 yards or more, he averaged 8.9 yards a carry. Among qualifying runners, only Ryan Mathews (9.1) had a higher average on runs of 3 yards or more.

Chris Ivory ranked 17th in average yardage, but 35th or worse in each of the other three categories. He had five runs of 30 yards or more -- only Gurley (seven), Adrian Peterson (six), and Doug Martin (six) had more.

It was a similar but even more extreme story for C.J. Anderson -- eighth in average yards, but 27th in average DYAR and in the bottom three in both median DYAR and success rate. Anderson had great explosiveness, with 21 10-yard runs in only 152 carries. For comparison's sake, Peterson led the league with 43 10-yard runs, and he had 327 total carries.

And finally there is the bizarre case of Matt Jones, who was dead last in average yards, average DYAR, and median DYAR -- and yet somehow 23rd in success rate. Jones was hit for no gain or a loss 38 times in 144 carries. Peterson also led the league in this category, with 71 -- but again, he carried the ball 183 times more than Jones did.

Quarterbacks
Rk
Player
Team
CP/AT
Yds
TD
INT
Sacks
Total
DYAR
Pass
DYAR
Rush
DYAR
Opp
1.
Tom Brady NE
28/42
302
2
0
0
170
157
13
KC
First four third downs: four completions, four conversions, 60 total yards, one touchdown. Third downs after that: five incompletions, one conversion -- a 6-yard gain on third-and-4. He had three plays with more than 10 yards to go for a first down and converted all three, gaining 86 yards in the process. He didn't have much luck throwing to his right, where Marcus Peters usually lines up, going 9-of-13 for 65 yards, but only three first downs. Two of those completions lost yards.
2.
Ben Roethlisberger PIT
24/36
339
0
0
3
137
137
0
DEN
Roethlisberger averaged 9.2 yards per pass against Denver, nearly a full yard higher than any other quarterback against Denver this season (Jay Cutler averaged 8.3 in Week 11). He had some remarkably dramatic splits in this game. Let's start with field position. Inside the Pittsburgh 25, he went 9-of-13 for 231 yards and eight first downs; a 14th throw resulted in a 22-yard DPI. And then from the Pittsburgh 47 to midfield, he had completions for 22 and 13 yards, plus another DPI for 12 more yards. Over the rest of the field, he went 13-of-21 for 73 yards and only one first down, with three sacks. And then there's the matter of pass distance. Anything within 7 yards of the line of scrimmage, he went 15-of-21 for 95 yards and only one first down. Deeper than that, he went 9-of-15 for 244 yards and nine first downs, plus DPIs of 12 and 22 yards.
3.
Russell Wilson SEA
31/46
366
3
2
5
128
117
11
CAR
DYAR by quarter: -55 (last among starters this week), -12 (fifth), 65 (fourth), 119 (first). And a big part of this ranking is due to opponent adjustments -- Wilson had only 35 unadjusted YAR, which would have been sixth this week. (It's the playoffs. Most players go up after opponent adjustments.) He had most of his consistent success attacking the middle of the field, going 5-of-7 for 60 yards, with every completion going for a first down. Of course, one of those incompletions was returned for a rather costly pick-six.
4.
Cam Newton CAR
16/22
161
1
0
1
109
122
-13
SEA
The key number here isn't Newton's DYAR, it's the number of pass plays he had, only 22. Every other starter had at least 37. And in such limited action, Newton didn't have much chance to rack up DYAR. His passing DVOA, though, was a stellar 71.5% -- the next best for a quarterback this week was Tom Brady at 46.2%. Despite passing so infrequently, Newton threw for 12 first downs -- that's tied with Ben Roethlisberger and Aaron Rodgers for fewest this week, but again, they had 41 and 45 dropbacks, respectively. And the amazing thing is, none of those first downs came on first down -- he went 4-of-6 for 16 yards on first down. Meanwhile, he was nearly flawless on second and third downs, going 12-of-16 for 145 yards with a sack -- and all of those completions went for first downs. He also converted on one of two third-down runs, and had another first down on second-and-2.
5.
Carson Palmer ARI
25/41
349
3
2
3
70
70
0
GB
Remember when we said Newton had only 21 plays? Palmer had 16 plays against Green Bay in the red zone alone. Mind you, he didn't play particularly well in the red zone. Yes, he had touchdowns of 5, 8, and 9 yards, but otherwise he went 5-of-12 for 29 yards and one first down, with an interception and a sack-fumble.
6.
Alex Smith KC
29/49
246
1
0
1
49
32
17
NE
In addition to his 49 pass attempts we list here, Smith also spiked a ball to stop the clock. (I don't remember him doing this, but I can only assume he let 20 seconds roll off the clock first.) That means officially, this was the 14th time this century a quarterback has thrown at least 50 passes in a playoff game. (Tom Brady has done it five times.) Smith's 246 yards and 4.92 yards per pass are the worst of that group. In fact, it has been more than two decades since a quarterback threw 50 passes in the playoffs with fewer yards than this -- Drew Bledsoe (who has a lot in common with Smith, when you think about it) threw 50 times for 235 yards in a loss to Cleveland in his second season.
7.
Aaron Rodgers GB
24/44
261
2
1
1
47
39
8
ARI
The 60-yarder on fourth-and-20 was worth 39 DYAR, and the Hail Janis was worth 35. Before those two throws, he had gone 0-for-10 on deep balls. He also had 11 passes on first down, with terrible results: seven completions (three of which lost yards) for 17 yards with no first downs and an interception.
8.
Peyton Manning DEN
21/37
222
0
0
1
16
16
0
PIT
Manning was the worst quarterback to make the playoffs this year, and as long as the Broncos keep winning, he'll be the worst quarterback in the playoffs. On third and fourth downs, he went 6-of-12 for 55 yards and only one first down, with a fumbled snap. (He did pick up another conversion on a 9-yard DPI.)


Five Best Running Backs by DYAR (Total)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Jonathan Stewart CAR
19
106
2
1/1
5
0
52
51
2
SEA
In addition to his 59-yard gain, Stewart had touchdowns of 1 and 4 yards, and also had a 1-yard gain on third-and-1. Those were his only first downs; he was hit for no gain or a loss three times. He was the first player to run for 100 yards against Seattle, regular season or playoffs, since Jamaal Charles did it in November of 2014 That 59-yarder (which was worth 22 DYAR) was Stewart's longest run of the season, and the longest run Seattle had allowed all year. In fact, Pete Carroll's Seahawks have only given up one longer run in the regular season or playoffs -- Adrian Peterson had a 74-yarder in 2012.
2.
Charcandrick West KC
17
61
1
2/4
15
0
25
23
2
NE
West had five first downs on the ground: a 1-yard touchdown; gains of 11 and 17 yards; and conversions on third-and-2 and third-and-4. He was hit for no gain or a loss four times.
3.
C.J. Anderson DEN
15
72
1
2/3
11
0
25
29
-4
PIT
Anderson had four first downs on the ground: a 1-yard touchdown; and gains of 7, 11, and 34 yards. He was hit for no gain or a loss three times.
4.
Eddie Lacy GB
12
89
0
2/2
2
0
14
25
-11
ARI
Lacy's 61-yarder was worth 22 DYAR. So yes, that is almost all he did against Arizona on Sunday. His two receptions: a 4-yard gain on second-and-9, and a 2-yard loss on first-and-10.
5.
Marshawn Lynch SEA
6
20
0
2/3
15
0
1
6
-4
CAR
Lynch's two receptions were an 11-yard gain on third-and-24 and a 4-yard gain on second-and-7. His only first down was an 8-yard gain on second-and-4 in the fourth quarter, his only carry of the second half. Look, it's the playoffs. Pickings for running backs are slim.


Five Best Running Backs by DYAR (Rushing)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Jonathan Stewart CAR
19
106
2
1/1
5
0
52
51
2
SEA
2.
C.J. Anderson DEN
15
72
1
2/3
11
0
25
29
-4
PIT
3.
Eddie Lacy GB
12
89
0
2/2
2
0
14
25
-11
ARI
4.
Charcandrick West KC
17
61
1
2/4
15
0
25
23
2
NE
5.
Knile Davis KC
6
30
0
2/3
13
0
-7
12
-18
NE
Davis had a very good day rushing, with all six carries gaining at least 2 yards. His three targets, though: a fumble on what would have been a first down on second-and-8; a 4-yard gain on first-and-10; and an incompletion on third-and-8.


Worst Running Back by DYAR (Total)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Ronnie Hillman DEN
16
38
0
0/1
0
0
-15
-8
-8
PIT
Yes, Hillman had two short first downs, but his longest run gained only 9 yards, and he was hit for no gain or a loss five times.


Worst Running Back by DYAR (Rushing)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
David Johnson ARI
15
35
0
6/9
43
0
-15
-25
10
GB
It's a good thing Johnson picked up three first downs as a receiver, because his running numbers were ugly. Two short first downs, a long gain of only 8 yards (which came on third-and-18), and three times hit for no gain or a loss.


Five Best Wide Receivers and Tight Ends by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Jeff Janis GB
7
11
145
20.7
2
77
ARI
The fact that Janis had 101 yards on the Packers' final drive of the season is sure to be one of the best stats of these playoffs. Those catches were worth 28 and 29 DYAR -- a healthy chunk, to be sure, but it's not like he was having a bad day before that. His three other first downs included a pair of third-down conversions.
2.
Jermaine Kearse SEA
11
15
110
10.0
2
74
CAR
The latest big game from Seattle's unlikely playoff hero. Kearse is now the 14th player with at least a half-dozen touchdown grabs in the postseason this century. (Julian Edelman has four and Danny Amendola has three, so they could join the list with another big game or two.) Of those 14 players, none has fewer regular-season touchdowns than Kearse's 10. David Givens had 12, and Brandon Stokley had 38; every other name on the list scored at least 40 times in the regular season, up to 98 for Larry Fitzgerald.
3.
Martavis Bryant PIT
9
15
154
17.1
0
60
DEN
Bryant gets 39 DYAR receiving, 21 rushing for his two carries, a gain of 40 and a failed run on third-and-2. He had three catches for 20 or more yards, including a 52-yarder, and five total first downs in the air.
4.
Larry Fitzgerald ARI
8
12
176
22.0
1
57
GB
You notice how there were about five plays this weekend I have to keep referencing in these comments? Yes, it was a weekend of big plays. Fitzgerald's two overtimes catches were worth 26 and 16 DYAR, so no, he would not have been on this list if Patrick Peterson had knocked the Hail Mary out of Jeff Janis' hands. Regardless, Fitzgerald had four other first downs on the day, on catches of 32, 19, 22, and 13 yards.
5.
Greg Olsen CAR
6
6
77
12.8
1
48
SEA
All of Olsen's catches gained first downs, including conversions on third-and-6, third-and-7, third-and-14, and second-and-15.


Worst Wide Receiver or Tight End by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Chris Conley KC
5
9
33
6.6
0
-32
NE
Conley's sole first down was a 16-yard gain on third-and-8. He had three other failed third-down targets.

Posted by: Vincent Verhei on 19 Jan 2016

69 comments, Last at 20 Jan 2016, 9:38pm by panthersnbraves

Comments

1
by Jerry :: Tue, 01/19/2016 - 8:11am

Proofreading:
Under Palmer: "Remember when we said Newton had only 21 plays?" You actually said 22.
"Fitzgerald's two overtimes catches"

I have a vague recollection of Mike Tanier doing a Walkthrough where he looked at distribution of running plays some years back. I think this covers much of what he did in much less space.

2
by Whatev :: Tue, 01/19/2016 - 8:56am

Dunno if you'd believe the result or not but you could try kernel smoothing the player's run yardage distribution and estimating the 50th percentile off of that.

36
by Scott C :: Tue, 01/19/2016 - 4:43pm

or take the average of the median quintile. Unless the distributions are highly irregular this should be a decent median-like metric that is more fine-grained.

3
by Will Allen :: Tue, 01/19/2016 - 9:39am

Running back performance is so much more dependent on o-line performance, relative to qb performance, that it gets hard to evaluate running backs , especially in 1 game sample sizes. Example 1 is Eddie Lacy this weekend. Somebody just looking at the stat sheet, even advanced stat sheet, might think Lacy had a really good night rushing the football; after all, 25 DYAR per game makes you the number 1 DYAR rusher for the year, by a huge margin. Anybody who watched the game, of course, saw a doughnut gorging tub of lard, who left yards and yards off the stat sheet, which was provided to him by blocking and opposing scheme. He sucked, stats be damned. Jonathan Stewart wasn't bad, but clearly his numbers reflected more on his blocking in the 1st half than it did his play.

It would be fascinating to have a very knowledgable running game evaluater go back and analyze film from the Barry Sanders and Emmitt Smith era. This is no rip on Smith, who was a great, great, player. Man, did he ever have the benefit of great teammates, however.

Evaluating individual football player performance is really hard.

6
by Hoodie_Sleeves :: Tue, 01/19/2016 - 10:59am

I feel like statistical analysis of running backs is almost a fools errand.

Between the influence of the offensive line (which may be a bigger factor than the influence of most RBs), the influence of scheme and quarterback (we know QBs who can run significantly increase runningback performance), and the fact that any sort of traditional analysis is going to be screwed up by the presence of goalline and other short-yardage situations, it seems like there's a lot more noise than there is data.

8
by Will Allen :: Tue, 01/19/2016 - 11:14am

I think Adrian Peterson might have had 200-plus yards rushing in the 1st half against the Seahawks on Sunday, if he had been provided the space that Stewart was afforded. Of course, even allowing for how thoroughly they were blocked by the Panthers, the Seahawks never would have allowed Peterson that much space, which maybe would have meant Newton would have run for 150 yards, to go with 150 passing yards.

In my wishes for obtaining football insight, the Vikings decide to get Peterson off of their cap space, so they cut him, Peterson decides that maximizing income isn't his highest priority (HA!) compared to winning a championship, so he signs a team friendly deal with the Panthers, and we see teams try to defend the read option with Newton placing the ball in Peterson's belly. I know Peterson likes to start deep behind a qb under center, but I think it still would afford all manner of fascinating match-ups.

27
by justanothersteve :: Tue, 01/19/2016 - 1:59pm

I think Peterson more likely goes to New England if maximizing his income isn't a priority and drives those who really hate NE even more mad. Belichick will stick with a RB when there is a clear #1 back (see Corey Dillon and, to a lesser extent, pre-injury Dion Lewis).

8
by Will Allen :: Tue, 01/19/2016 - 11:14am

I think Adrian Peterson might have had 200-plus yards rushing in the 1st half against the Seahawks on Sunday, if he had been provided the space that Stewart was afforded. Of course, even allowing for how thoroughly they were blocked by the Panthers, the Seahawks never would have allowed Peterson that much space, which maybe would have meant Newton would have run for 150 yards, to go with 150 passing yards.

In my wishes for obtaining football insight, the Vikings decide to get Peterson off of their cap space, so they cut him, Peterson decides that maximizing income isn't his highest priority (HA!) compared to winning a championship, so he signs a team friendly deal with the Panthers, and we see teams try to defend the read option with Newton placing the ball in Peterson's belly. I know Peterson likes to start deep behind a qb under center, but I think it still would afford all manner of fascinating match-ups.

11
by MilkmanDanimal :: Tue, 01/19/2016 - 11:36am

Is the analysis any worse for running backs than other players? QB play is just as dependent on the offensive line, and you have to factor in whether a back does his blitz pickup right and if the WR and QB are on the same page regarding coverage and which option route to run and press vs. zone and everything else. That's just the basic issue with trying to do analysis of a very team-based sport.

12
by Will Allen :: Tue, 01/19/2016 - 11:41am

I really think a qb can frequently compensate for bad blocking, to an extent that a running back cannot, and I don't think a qb can flat out stink, while putting together an impressive stat sheet, like Eddie Lacey did Sunday.

14
by dank067 :: Tue, 01/19/2016 - 11:55am

Carson Palmer in the same game almost fits that bill! (He made enough good throws that you can say he didn't flat-out stink, and while the rest of his numbers were impressive, the 2 INTs are enough such that you wouldn't mistake it for a truly great game.)

17
by Will Allen :: Tue, 01/19/2016 - 12:05pm

I wouldn't say "almost". Palmer had his alarmingy bad moments, balanced by some good ones. Lacy was just uniformly awful. My garbage man, absent his truck, could have covered the distance Lacey did on his biggest gain.

22
by MilkmanDanimal :: Tue, 01/19/2016 - 12:38pm

Yeah, Palmer came terrifyingly close to losing that game several times. His pick in the end zone was an utterly terrible throw, and he came back later to almost throw what would have been a pick-six to Shields. He was extremely shaky at best at least from the eyeball test point of view.

23
by tuluse :: Tue, 01/19/2016 - 12:46pm

During the game I decided Palmer had turned into a slow Jay Cutler. Repeatedly confused by a defense designed to confuse and the kicker was that lob off his back foot into the endzone. Prime example of just trusting your arm way too much.

Still, you can win games (even playoff games) with a slow Jay Cutler. You just hope for more.

26
by big10freak :: Tue, 01/19/2016 - 12:59pm

Palmer should have had 4 interceptions and with some great athleticism possibly a 5th. He could have easily been pegged as the new Jake Delhomme because he almost lost a fumble on the sack by Perry. (Who for reasons unknonwn basically just pushed the qb over versus going in hard where if he had been there might have been able to get the resulting fumble)

Palmer did make some great throws. But mostly the GB defense had him struggling.

Toss in the fact that the officials swallowed their whistles on offensive holding in the second half and Palmer was spared a truly disasterous outing. Because Peppers in particular was killing the Cards left tackle.

37
by Duff Soviet Union :: Tue, 01/19/2016 - 4:49pm

Was Palmer that much worse than Tom Brady? Brady had 5 dropped picks by my count (admittedly one of them was completely not his fault as the ball hit an open Edelman in the hands before he deflected it to a Chiefs defender) and nobody seems to be talking about this.

Palmer was ok. He was either very good or very bad on any given throw with nothing in between.

64
by RBroPF :: Wed, 01/20/2016 - 1:10pm

So are we just counting all pass breakups as dropped picks now?

If you give me the game times, I'll go check the tape. But off the top of my head I don't recall any times where a defender even got two hands on the ball at the same time.

65
by Anon Ymous :: Wed, 01/20/2016 - 1:15pm

The one that bounced off Edelman's hands was clearly interceptable. I'm not sure about any of the others. Hali's "drop" at the end would have been a difficult catch by any player, let alone a LB, but it is the only other time I felt any pass was close to being picked.

67
by RBroPF :: Wed, 01/20/2016 - 1:28pm

I agree Edelman's drop was interceptable, but only because Brady's perfectly thrown pass was bobbled and then tapped up in the air directly in front of the defender. (Exactly like the only pick-6 Brady threw this year).

28
by ChrisS :: Tue, 01/19/2016 - 2:02pm

After watching all the games I expected to see Palmer as the worst QB of the week, maybe ahead of Peyton.

15
by Hoodie_Sleeves :: Tue, 01/19/2016 - 11:56am

Quarterback is dependent on other players of course, but nowhere to the degree as RB.

A quarterback who can get the ball out quickly makes offensive line quality significantly less important (assuming he has options that can get open quickly).

There's definitely still interference, I just think that at RB, the interference is a more significant factor than RB skill, whereas QB skill is a bigger factor in most cases than the interference.

38
by Duff Soviet Union :: Tue, 01/19/2016 - 4:52pm

Pretty much every back in the league is interchangeable with every other back in the league. Running backs don't cause good offenses, they benefit from them.

I've said this before, but the consensus best three backs in the league in the preseason were Lynch, Charles and Bell. All three got injured, all three were replaced by garden variety backups and their offenses weren't effected in the slightest.

40
by tuluse :: Tue, 01/19/2016 - 6:15pm

Seattle went from 29% rushing DVOA with an effective Lynch to 7.4% rushing DVOA this year.

Just sayin'

41
by Duff Soviet Union :: Tue, 01/19/2016 - 6:28pm

They still had Lynch for most of this year, and Rawls was better than Lynch anyway.

Most of their reduced effectiveness was losing Unger and Wilson having a fluke great season as a runner last year, which he couldn't repeat. He was still good as a runner, but not amazing.

42
by tuluse :: Tue, 01/19/2016 - 6:39pm

There's having 2015 Lynch and having 2014 Lynch, and they're not the same thing. The Broncos had Peyton Manning this year and their passing offense still sucked.

31
by Hang50 :: Tue, 01/19/2016 - 2:17pm

Watching Hillman and Anderson is an interesting exercise. Hillman is faster, but Anderson has better vision. If there's a hole in view, Hillman is likely to make more out of it than Anderson. Trouble is, Kubiak's offense is built so the RB needs to find the cutback lanes, a task more in line with Anderson's skills.

49
by tunesmith :: Tue, 01/19/2016 - 8:59pm

Yeah, it seems like Hillman's success is entirely dependent on the offensive line's success. He was "least valuable running back" two weeks in a row earlier this season. Although it seemed that this time, the offensive line played pretty well.

4
by craigj1971 :: Tue, 01/19/2016 - 10:08am

Reading the comments about Jeremy Hill reminded me of the fantastic quote from Leroy Hoard, "If you want three yards, I'll get you three yards; if you want ten yards, I'll get you three yards." Classic!

5
by PirateFreedom :: Tue, 01/19/2016 - 10:08am

Bledsoe and Smith are both #1 overall picks who got traded away, but unless Smith owns a vinyard that is pretty much all the similarity I can think of ;)

7
by big10freak :: Tue, 01/19/2016 - 11:04am

Lost in the late game heroics is that even with good protection and yes, really, guys getting open Rodgers was missing throws.

There were multiple ways for GB to win that game and one of them was the all world qb not just playing ok but playing well.

And Will understates the disgusting nature of Lacy's run Saturday night. Why McCarthy stuck with him versus Starks is completely baffling. Especially since Starks can catch the ball better than Lacy and AZ struggled with running backs catching the ball most of the season.

10
by drobviousso :: Tue, 01/19/2016 - 11:25am

"Is there a good way to evaluate running backs without letting a small number of big runs skew the results?"

Yes, it's called a bar graph. I know they are images and don't sort into tables and you have to interpret them, but they paint a really detailed picture.

Someone did this years ago, and I can't find it. This would have been around AP's second or third year in the league. His distribution looked like a normal running back with lots and lots of runs over 20 yards tacked on. I think it was Marshall Faulk who's distribution was skewed right because he almost never lost yardage and you could see that those bad runs turned into 1 and 2 yard gains.

You should give it a try this off season.

16
by JMM :: Tue, 01/19/2016 - 11:56am

For more on the value of graphing, see Anscomb's Quartet.

https://en.m.wikipedia.org/wiki/Anscombe%27s_quartet

20
by billsfan :: Tue, 01/19/2016 - 12:32pm

What might also be informative would be to bin the carries into quintiles (or deciles), so you can exclude long runs (and stuffs, which might be the o-line's fault). So if you're just looking at the average YPC of the middle 60% (or 80%) of the RB's runs, it may say something different than aggregate YPC.

e.g., for Rawls, you'd get 3.9 YPC for the average of the middle 3 bins, with each bin averaging -0.2, 2.2, 3.7, 6.0, 16.6, vs. that aggregate 5.6 YPC

Ultimately, I suspect any complicated analysis of carry data might lead you to some absurd conclusion like "RBs are fungible"

24
by Will Allen :: Tue, 01/19/2016 - 12:46pm

Yeah, I don't know if you last sentence was made ironically or not, but the issue I've had, stretching back to the analysis of Barry Sanders' metrics, stretching through Peterson's, is that I think excluding, or discounting, the long runs of consistently extremely explosive running backs has the effect of inaccurately representing how much such a player increases the odds of winning. When a running back dictates terms to a defensive coordinator, because the coordinator lives in fear of his unit yielding a td every time the running back takes a handoff, anywhere on the field, that's just extremely valuable. It is what makes Christian Ponder and Scott Mitchell look like viable NFL starting qbs, or makes a 40 year old Brett Favre look like he is as good as he was at age 28.

Yes, there is a lot more fungibility at rb, than there is at receiver or qb. At the highest level of performance, this isn't nearly the case.

29
by billsfan :: Tue, 01/19/2016 - 2:03pm

Little of both, I'm afraid. I think it's pretty fair to say that you can likely get median YPC performance from a street free agent RB.

But I certainly agree, and I think it's a little clearer in the WP/EP models, that a DVOA dog like AP does what all our eyes can plainly see. I think there have been attempts in the last few years at probabilistic modeling of a boom-or-bust back, i.e. what frequency or length of long runs is necessary to sustain a drive or simply not go three-and-out. Can't remember how that worked out, but I seem to recall someone did it (Burke, maybe?).

And you might even be able to tease a metric or two out by something as simple as binning the carries and fitting a line. (negative y-intercept means more stuffs, higher slope means more explosive plays, etc.)

35
by Dan :: Tue, 01/19/2016 - 4:29pm
61
by armchair journe... :: Wed, 01/20/2016 - 9:56am

Now that's pretty cool.

//AJMQB

13
by RickD :: Tue, 01/19/2016 - 11:49am

I'm interested in the difference between Success Rate and Median DYAR. Consider LeGarrette Blount, who has a high success rate with a low Median DYAR. His success rate of 52% puts him at 6th highest on the table, and yet his median DYAR is -0.10. This means that (because math) at least 2% of his runs are considered successes and yet have negative DYAR of -0.10 or worse.

On the other side we have runners like David Johnson. His success rate is lower, at 45%, and yet his median DYAR is 0.21. That means at least 5% of his rushes are not successes and yet have DYAR of at least 0.21.

Clearly this means that success rate isn't the sole determinant of DYAR. Anybody want to talk about this a bit?

19
by dank067 :: Tue, 01/19/2016 - 12:26pm

Don't know if these were frequent occurrences for NE and ARI this year, but plays that could lead to these situations:

Case one: something like a one yard gain on 2nd and 1. It's a successful play, but without the defense selling out to stop the conversion like they would on third down, you'd be expected to gain more yards

Case two: maybe a draw play for 5-10 yards with very long distances to go on 2nd and 3rd down. Don't get the 60%/100% of required yards, but enough teams throw incomplete or for dumpoffs that maybe you get positive DYAR.

21
by tuluse :: Tue, 01/19/2016 - 12:36pm

"Clearly this means that success rate isn't the sole determinant of DYAR. Anybody want to talk about this a bit?"

It's degree of success compared to a baseline (replacement level). So if the expected outcome of a play is unsuccessful (image a draw on 3rd and 27), you can be better than replacement level while being unsuccessful.

18
by DavidL :: Tue, 01/19/2016 - 12:12pm

Wait a second -- 60 yards on 4th and 20 is 39 DYAR...meaning a replacement-quality QB is expected to get that first down? That doesn't speak well for the Cards defense.

25
by tuluse :: Tue, 01/19/2016 - 12:47pm

The conversion itself is worth DYAR. A replacement level QB is probably expected to get ~15 yards in such a situation. Remember each yard in a big play is counted a little bit less than the yard before it. There's not a huge difference in getting a 60 yard gain over a 55 yard gain (compared to say a 10 yard gain over a 5 yard gain).

30
by RickD :: Tue, 01/19/2016 - 2:11pm

The longer a completion or a run is, the lower the rate of DYAR/yard for the extra yards.

32
by TomC :: Tue, 01/19/2016 - 2:48pm

Drew Bledsoe (who has a lot in common with [Alex] Smith, when you think about it).

Yup: mobile, noodle-armed, never throws deep, hasn't won a Super Bowl...

33
by PatsFan :: Tue, 01/19/2016 - 2:53pm

One out of four ain't bad. :)

43
by TomC :: Tue, 01/19/2016 - 7:29pm

I had completely forgotten that Brady got knocked out of the AFCC and Bledsoe played the 2nd half (and threw the game's only TD). He actually earned the ring more than I realized.

44
by Will Allen :: Tue, 01/19/2016 - 7:31pm

I bet Bledsoe hasn't forgotten.

47
by PatsFan :: Tue, 01/19/2016 - 8:46pm

Bledsoe did a nice job of getting that TD before the first half ended. IIRC he threw the TD only a couple of plays after he took a Mo Lewis-like hit, too. But that was about it. And Brady had moved them into PIT territory before Lee Flowers conveniently rolled into Brady's ankle.

The second half was all about gritting one's teeth and praying that Bledsoe wouldn't do some standard Bledsoe-esque game-losing thing.

He almost did, too. While being sacked he threw the ball backwards over his head. Thankfully the PIT linebacker he hit in the hands with the ball dropped it.

He really didn't do much at all in the 2nd half. NE's defense and STs won the game for them at that point.

53
by Athelas :: Tue, 01/19/2016 - 10:10pm

That's an accurate recap of that game.

I remember thinking Brady couldn't make that throw in the corner of the endzone--how wrong I was.

55
by PatsFan :: Tue, 01/19/2016 - 10:13pm

Especially since he hit one exactly one week later.

56
by Athelas :: Tue, 01/19/2016 - 10:18pm

Exactly.

63
by Anon Ymous :: Wed, 01/20/2016 - 11:02am

This is very accurate with the exception of one thing: Bledsoe's hook shot was not the pass that was almost intercepted. That was a wholly separate terrible throw. :)

Drew should be complimented for a beautiful rainbow that dropped right into Brown's hands for a key first down, though. That was exquisite. The rest of the second half... not so much.

34
by duh :: Tue, 01/19/2016 - 3:59pm

All of what you say is true I think what he was getting at was they are in many ways heads and tails of the same coin on a more meta level.

66
by RBroPF :: Wed, 01/20/2016 - 1:20pm

"they are in many ways heads and tails of the same coin on a more meta level."

Wow. I don't have any idea what this means but I can't wait to find a way to use it.

39
by Brendan Scolari :: Tue, 01/19/2016 - 6:03pm

Referring to several comments above: I think what gets lost in the "running backs are fungible" argument is that of course an O-line should have a larger influence on the running game effectiveness than a single running back. We're talking about 5 guys playing at once versus one. They constitute almost half of the offensive players.

In terms of positional value, a more accurate comparison would be, "Who is more important to the running game, a running back or a left guard?" Substitute other O-line positions in as you please. Also interesting, "Who is more important to overall offensive efficiency, a running back or a wide receiver?" One this one I'd go with the wide receiver but the gap is probably a lot smaller than the gap in importance between the quarterback and every other position.

45
by t.d. :: Tue, 01/19/2016 - 7:46pm

I mean it's not scientific, but PFF gave Manning a higher grade than Brady this week (he was killed by drops, but there was a dropped pick, too)

46
by Hummingbird Cyborg :: Tue, 01/19/2016 - 8:35pm

I wonder if they gave Sanders a positive grade for the pass defense of that pick.

50
by blan :: Tue, 01/19/2016 - 9:19pm

I don't think that's true.

I see Manning -0.9 and Brady +2.6.

But I'm not paying for all their numbers, so maybe you're looking at something else.

54
by Athelas :: Tue, 01/19/2016 - 10:11pm

I think you're thinking of "Who's wrong more than" Charlie Casserly.

48
by lightsout85 :: Tue, 01/19/2016 - 8:48pm

RB measures seem to either favor OL play too much, or frequency & length of very big plays. If only things like missed/broken tackles & yards-after-contact were more wildly available. They seem to more accurately reflect a RB "creating" on their own.

One thing I do like looking at, is the average of carries that are at least 11 yards long (and outside the redzone), the area FO considers outside the influence of the OL. I like it better than big-play-frequency as a measure of a big-play back, because it eliminates the variables involved in getting the RB into space in the first place. (However, I have heard the theory that backs who see more 8-man fronts will tend to produce longer gains than those facing defenses w/2 deeper safeties).

59
by billsfan :: Wed, 01/20/2016 - 9:32am

Chicken and egg with those 8-man fronts--you could just as easily say they're seeing more guys in the box *because* they tend to produce longer runs.

52
by Kal :: Tue, 01/19/2016 - 9:49pm

Interesting double post

51
by Kal :: Tue, 01/19/2016 - 9:49pm

Something else to consider is whether or not measuring median yards is what you care about at all. What ends up being more successful for a team - big plays or consistent yards?

Was it more important that Carolina got 4 yards instead of 3 per carry, or was that 60 yard run really hugely important to the overall result of the game?

I don't know the answer in the NFL. My suspicion is that it correlates strongly to college, where big play ability is a massive predictor of future success and one of the best descriptors of who won a game (per Bill Connelly).

58
by tuluse :: Wed, 01/20/2016 - 3:16am

The question isn't "does a 60 yard run help a team win the game it happens in". That is clearly true. The question is "does a 60 yard run followed by a bunch of failed runs predict that a team will be successful in the future".

60
by Will Allen :: Wed, 01/20/2016 - 9:39am

And that depends on how the defense responds to the 60 yard run. It's the same thing as the 60 yard pass, really. When Randy Moss runs past two normally good defenders, easily grabs the pass as they flail pointlessly at the ball, and runs into the end zone, it has a lot more positive effect, in terms of the offensive team winning, than a 60 yard scoring pass that occurs because a single, normally good, defender slipped on a wet field. A 60 yard play which forces a defense to become predictable is a lot more valuable than a 60 yard play which doesn't.

68
by Kal :: Wed, 01/20/2016 - 1:40pm

That, and 'does a 60 yard run combined with a bunch of failed runs contribute more to a win than a bunch of successful runs with otherwise no real gain'.

From Connelly's work it seems like the former is more true, though both contribute.

57
by greybeard :: Wed, 01/20/2016 - 1:54am

But they have plenty high DYAR. As good as the Carolina and GB defenses are and whatever adjustments they bring, I will trust my eyes in this case.

62
by Arkaein :: Wed, 01/20/2016 - 10:40am

Palmer's DYAR is good because:

1. DYAR doesn't know that Sam Shields dropped a couple of easy INTs, and just sees these as a pair of incompletions.
2. DYAR doesn't know that one of Palmer's TD passes came off of a well defended pass deflected by a DB.
3. Two of Palmers most productive plays, a 75 yard pass and a 5 yard shovel pass TD, came in OT. Before that point I have to think he's behind Rodgers at least.

69
by panthersnbraves :: Wed, 01/20/2016 - 9:38pm

I love the write-up on Cam - too bad 99% of the world is just going to look at the number...