Article: PROTRADE - Lynch and Baily top 30 DB's - The Orange Mane

El Minion · 12-02-2005, 12:55 PM

http://www.protrade.com/insight/Insi...jardins%2F&x=x

Rating the Defensive Backs
Dec 01 2005 11:40 AM PST
GABRIEL DESJARDINS

Though we might argue about the relative merits of quarterback ratings, yards-per-carry and Expected Scoring ("ES"), football offers no lack of offensive statistics. For every single play, there exists troves of widely available data that describes the every accomplishment and failure of players at the skill positions.

But there's but next to nothing that lets us examine what defenders do.

Let's compare the 2001-04 statistics of two Pro Bowl cornerbacks: Dre' Bly of the Detroit Lions and Champ Bailey of the Denver Broncos.

Games Tackles Assists INTs Pass Def
Bly 59 160 21 18 45
Bailey 64 245 25 11 49

The official NFL stats tell us that Bailey makes tackles at a much higher rate than and has been credited with more pass defenses. Bly's only advantage is in interceptions.

This data reads like Bailey is much more involved defensively. But what exactly do we make of his extra tackles? Did he get them because he allowed a pass completion and then tackled the receiver? Or did he make the tackle because was he helping out a teammate, tackling a safety or linebacker's man?

Thanks to PROTRADE's NFL play-by-play database, we can look a little deeper at their defensive statistics. For 2001-04, we've tracked defensive opportunities for every defensive back:

Completions Incomplete INTs Penalties Broken Tackles
Bly 176 160 18 23 17
Bailey 207 162 11 16 9

There is obviously some subjectivity in these measures -- crediting a defensive back with forcing a pass incompletion is not as straightforward as crediting a running back with a carry or a quarterback with a pass attempt.

But it is clear that when a pass is thrown near Bly, it was much more likely to end up incomplete than if Bailey was the defender.

Using PROTRADE's ES system, we can quantify the point value of each player's performance. However, we need to make a few preliminary adjustments on a per-play basis.

Passing plays that involve defensive backs are typically very positive for the offense. While your usual incomplete pass equates to around -0.25 ES, a 10-yard completion might be worth 0.50 ES.

Seeing as 57% of passes thrown into the secondary are completed, we would expect the average such play to earn +0.18 ES, an amount we would need to credit back to defenders so as not to reward sitting on the bench over playing.

We adjust for pass location and distance and we actually credit slightly more than this amount per play, since we want to compare these defensive backs to replacement level players, not the league average.

The top 30 NFL defensive backs by expected scoring above replacement level (ESAR), 2001-04:

Athlete Pos ESAR Pro Bowls
1 Ronde Barber CB -54.7 3
2 Dre' Bly CB -52.1 3
3 Deshea Townsend CB -48.7 -
4 Darren Sharper S -45.0 2
5 Ty Law CB -44.8 3
6 Brian Dawkins S -39.0 5
7 Ahmed Plummer CB -38.7 -
8 Chad Williams S -38.4 -
9 John Lynch S -38.3 6
10 Dwight Smith S -38.0 -
11 Eric Warfield CB -37.8 -
12 Tony Parrish S -35.9 -
13 Patrick Surtain CB -35.4 2
14 Sheldon Brown CB -34.3 -
15 Rodney Harrison S -32.4 2

Athlete Pos ESAR Pro Bowls
16 Mike McKenzie CB -32.3 -
17 Nate Clements CB -30.6 2
18 Tod McBride CB -30.0 -
19 Roy Williams S -30.0 3
20 Marcus Coleman S -29.9 -
21 Brian Kelly CB -29.9 -
22 Lawyer Milloy S -28.9 4
23 Tyrone Poole CB -28.3 -
24 Champ Bailey CB -27.8 5
25 Earl Little S -27.2 -
26 Ray Mickens CB -27.1 -
27 Ashley Ambrose CB -26.9 1
28 Bobby Taylor CB -26.8 1
29 Terence Newman CB -26.6 -
30 Troy Vincent S -26.5 4

The way to read this: as the top defensive back, Ronde Barber prevented 54.7 points more than a replacement level player.

If Pro Bowl appearances are an indication of defensive back skill, then the system performs quite well-- fully half of the Top 30 players listed here have made All-Pro.

So what about the Bottom 30?

Athlete Pos ESAR Pro Bowls
Ronnie Heard S 1.9 -
David Barrett CB 2.1 -
Derek Ross CB 2.2 -
Chris McAlister CB 2.5 3
Andre' Goodman CB 2.9 -
Jeff Burris CB 3.0 -
Kevin Mathis CB 3.3 -
Todd Lyght CB 3.6 1
Omar Stoutmire S 5.3 -
Fernando Bryant CB 5.4 -
Tyrone Carter S 5.6 -
Matt Stevens S 5.7 -
Pierson Prioleau S 5.8 -
Mike Rumph CB 5.9 -
Jerry Wilson S 5.9 -

Athlete Pos ESAR Pro Bowls
Rogers Beckett S 6.0 -
Ken Irvin CB 6.1 -
Cory Hall S 6.6 -
William Bartee CB 6.7 -
Denard Walker CB 7.0 -
Chris Cash CB 8.1 -
Alex Molden CB 9.8 -
Eric Kelly CB 10.0 -
Corey Chavous S 10.0 1
Ryan McNeil CB 10.3 -
Andre Dyson CB 11.7 -
Quentin Jammer CB 12.8 -
Bryan Scott S 15.7 -
Jay Bellamy S 15.7 -
Phillip Buchanon CB 17.6 -

Most of the bottom 30 defensive backs are the definition of replacement players: they saw from 20 to 50 plays and performed very poorly on most of them, giving up huge gains on a per-play rate.

Our Bottom 30 is actually drawn from the bottom 114 players, but it only includes semi-regular players who had at least 100 defensive chances over the last four years. Only three of the 114 lowest-ranked defensive backs have been to the Pro Bowl, and Todd Lyght's only appearance was in 1999, two years to prior to those covered by the dataset.

Baltmore's Chris McAlister is the most significant outlier. He also had the third-most penalties of any defensive back, costing his team more than 10 points.

At the same time, as a cornerback, McAlister covers a lot of ground that would traditionally belong to the safety. He may have been credited with pass defenses (and completions) in the play-by-play that he wouldn't have had he stayed in a much smaller zone.

Defensive play-by-play remains inexact - for the most part, we don't know who the other 10 players on the field are, nor do we know where they lined up. Some of the most obvious questions about defense, such as how far a player ran to make a tackle or whether he beat a block to get there, are impossible to answer today.

More in-depth play charting may someday offer clues. But for now, PROTRADE's play-by-play and ES system seem to agree with what coaches, players and fans see when they vote for the Pro Bowl rosters.

Bob's your Information Minister · 12-02-2005, 01:03 PM

Quote:

Originally Posted by El Minion

11 Eric Warfield CB -37.8 -

13 Patrick Surtain CB -35.4 2
1

How you like them apples?

Lidderer · 12-02-2005, 01:30 PM

Quote:

Originally Posted by Bob's your Information Minister

How you like them apples?

"Defensive play-by-play remains inexact - for the most part, we don't know who the other 10 players on the field are, nor do we know where they lined up."

in other words: "apple drink", not "juice"

Bob's your Information Minister · 12-02-2005, 01:31 PM

The same applies to your DBs then.

Rock Chalk · 12-02-2005, 01:36 PM

Quote:

Originally Posted by Bob's your Information Minister

How you like them apples?

They also had McAllister in the bottom 30 and id take mcallister over any DB on KCs roster.

How you like them apples?

Willynowei · 12-02-2005, 01:38 PM

most useless crap ever.

Lidderer · 12-02-2005, 01:39 PM

Quote:

Originally Posted by Bob's your Information Minister

The same applies to your DBs then.

agreed. I mean, c'mon, Lynch?!

Bob's your Information Minister · 12-02-2005, 01:39 PM

Quote:

Originally Posted by AlecRaenos

They also had McAllister in the bottom 30 and id take mcallister over any DB on KCs roster.

How you like them apples?

I suppose you'd take one of your rookies over him, too.

Rock Chalk · 12-02-2005, 01:41 PM

Quote:

Originally Posted by Bob's your Information Minister

I suppose you'd take one of your rookies over him, too.

Id take both our rookies over anything in KCs secondary yes. 1) Surtain is not as good as advertised, nor is he young anymore. 2) Sammy Knight who?

Lidderer · 12-02-2005, 01:45 PM

Quote:

Originally Posted by AlecRaenos

2) Sammy Knight who?

I believe you're referring to Sammy "the ballhawk" Knight. He of 1 INT fame.

Rock Chalk · 12-02-2005, 01:47 PM

Yeah the "Red" Knight. As in "I thought I red the play right" (pun n the bad spelling intended)

Mile High Shack · 12-02-2005, 01:50 PM

Quote:

Originally Posted by AlecRaenos

Yeah the "Red" Knight. As in "I thought I red the play right" (pun n the bad spelling intended)

yeah

none shall pass or some D&D crap Bob posted

Bob's your Information Minister · 12-02-2005, 01:56 PM

Quote:

Originally Posted by Mile High Shack

yeah

none shall pass or some D&D crap Bob posted

This was back when KC signed him...I photoshopped one of his Miami pics.

Taco John · 12-02-2005, 02:01 PM

Quote:

Originally Posted by Bob's your Information Minister

I suppose you'd take one of your rookies over him, too.

I'd take Darrent Williams over Warfield without question.

Lidderer · 12-02-2005, 02:01 PM

Quote:

Originally Posted by Bob's your Information Minister

This was back when KC signed him...I photoshopped one of his Miami pics.

if you pan out further the whole line reads "none shall pass go without collecting 20 receiving yards."

Drek · 12-02-2005, 02:02 PM

Crappy stat analysis. Completion percentages against are very speculative at best. I'd dare call it outright stat mining but without any particular goal. Just throwing individual numbers into equations used for team stats doesn't work well.

This study should at least normalize the numbers by team stats along with filtering some of the noise by making a comparative evaluation based on what teammates have done relative to one another, etc..

Lidderer · 12-02-2005, 02:05 PM

Quote:

Originally Posted by Drek

Crappy stat analysis. Completion percentages against are very speculative at best. I'd dare call it outright stat mining but without any particular goal. Just throwing individual numbers into equations used for team stats doesn't work well.

This study should at least normalize the numbers by team stats along with filtering some of the noise by making a comparative evaluation based on what teammates have done relative to one another, etc..

I'd say they go pretty far in making it clear that their numbers aren't at all definitive.

OrangeShadow · 12-02-2005, 07:23 PM

Darrent Williams
Depth	Att	Comp	Yds	TD	INT	Pen
Short	33	22	198	0	1	0
Medium	12	7	109	0	1	0
Deep	9	0	0	0	0	0
Totals	54	29	307	0	2	0

Domonique Dominique Foxworth
Depth	Att	Comp	Yds	TD	INT	Pen
Short	30	20	137	1	0	0
Medium	16	8	111	1	0	0
Deep	12	3	101	1	1	18
Totals	58	31	349	3	1	18

Champ Bailey
Depth	Att	Comp	Yds	TD	INT	Pen
Short	20	12	83	0	2	0
Medium	11	3	47	1	1	28
Deep	9	2	37	1	1	0
Totals	40	17	167	2	4	28

12-02-2005, 12:55 PM	#1
El Minion Ring of Famer Join Date: Apr 2005 Posts: 1,551	Article: PROTRADE - Lynch and Baily top 30 DB's http://www.protrade.com/insight/Insi...jardins%2F&x=x Rating the Defensive Backs Dec 01 2005 11:40 AM PST GABRIEL DESJARDINS Though we might argue about the relative merits of quarterback ratings, yards-per-carry and Expected Scoring ("ES"), football offers no lack of offensive statistics. For every single play, there exists troves of widely available data that describes the every accomplishment and failure of players at the skill positions. But there's but next to nothing that lets us examine what defenders do. Let's compare the 2001-04 statistics of two Pro Bowl cornerbacks: Dre' Bly of the Detroit Lions and Champ Bailey of the Denver Broncos. Games Tackles Assists INTs Pass Def Bly 59 160 21 18 45 Bailey 64 245 25 11 49 The official NFL stats tell us that Bailey makes tackles at a much higher rate than and has been credited with more pass defenses. Bly's only advantage is in interceptions. This data reads like Bailey is much more involved defensively. But what exactly do we make of his extra tackles? Did he get them because he allowed a pass completion and then tackled the receiver? Or did he make the tackle because was he helping out a teammate, tackling a safety or linebacker's man? Thanks to PROTRADE's NFL play-by-play database, we can look a little deeper at their defensive statistics. For 2001-04, we've tracked defensive opportunities for every defensive back: Completions Incomplete INTs Penalties Broken Tackles Bly 176 160 18 23 17 Bailey 207 162 11 16 9 There is obviously some subjectivity in these measures -- crediting a defensive back with forcing a pass incompletion is not as straightforward as crediting a running back with a carry or a quarterback with a pass attempt. But it is clear that when a pass is thrown near Bly, it was much more likely to end up incomplete than if Bailey was the defender. Using PROTRADE's ES system, we can quantify the point value of each player's performance. However, we need to make a few preliminary adjustments on a per-play basis. Passing plays that involve defensive backs are typically very positive for the offense. While your usual incomplete pass equates to around -0.25 ES, a 10-yard completion might be worth 0.50 ES. Seeing as 57% of passes thrown into the secondary are completed, we would expect the average such play to earn +0.18 ES, an amount we would need to credit back to defenders so as not to reward sitting on the bench over playing. We adjust for pass location and distance and we actually credit slightly more than this amount per play, since we want to compare these defensive backs to replacement level players, not the league average. The top 30 NFL defensive backs by expected scoring above replacement level (ESAR), 2001-04: Athlete Pos ESAR Pro Bowls 1 Ronde Barber CB -54.7 3 2 Dre' Bly CB -52.1 3 3 Deshea Townsend CB -48.7 - 4 Darren Sharper S -45.0 2 5 Ty Law CB -44.8 3 6 Brian Dawkins S -39.0 5 7 Ahmed Plummer CB -38.7 - 8 Chad Williams S -38.4 - 9 John Lynch S -38.3 6 10 Dwight Smith S -38.0 - 11 Eric Warfield CB -37.8 - 12 Tony Parrish S -35.9 - 13 Patrick Surtain CB -35.4 2 14 Sheldon Brown CB -34.3 - 15 Rodney Harrison S -32.4 2 Athlete Pos ESAR Pro Bowls 16 Mike McKenzie CB -32.3 - 17 Nate Clements CB -30.6 2 18 Tod McBride CB -30.0 - 19 Roy Williams S -30.0 3 20 Marcus Coleman S -29.9 - 21 Brian Kelly CB -29.9 - 22 Lawyer Milloy S -28.9 4 23 Tyrone Poole CB -28.3 - 24 Champ Bailey CB -27.8 5 25 Earl Little S -27.2 - 26 Ray Mickens CB -27.1 - 27 Ashley Ambrose CB -26.9 1 28 Bobby Taylor CB -26.8 1 29 Terence Newman CB -26.6 - 30 Troy Vincent S -26.5 4 The way to read this: as the top defensive back, Ronde Barber prevented 54.7 points more than a replacement level player. If Pro Bowl appearances are an indication of defensive back skill, then the system performs quite well-- fully half of the Top 30 players listed here have made All-Pro. So what about the Bottom 30? Athlete Pos ESAR Pro Bowls Ronnie Heard S 1.9 - David Barrett CB 2.1 - Derek Ross CB 2.2 - Chris McAlister CB 2.5 3 Andre' Goodman CB 2.9 - Jeff Burris CB 3.0 - Kevin Mathis CB 3.3 - Todd Lyght CB 3.6 1 Omar Stoutmire S 5.3 - Fernando Bryant CB 5.4 - Tyrone Carter S 5.6 - Matt Stevens S 5.7 - Pierson Prioleau S 5.8 - Mike Rumph CB 5.9 - Jerry Wilson S 5.9 - Athlete Pos ESAR Pro Bowls Rogers Beckett S 6.0 - Ken Irvin CB 6.1 - Cory Hall S 6.6 - William Bartee CB 6.7 - Denard Walker CB 7.0 - Chris Cash CB 8.1 - Alex Molden CB 9.8 - Eric Kelly CB 10.0 - Corey Chavous S 10.0 1 Ryan McNeil CB 10.3 - Andre Dyson CB 11.7 - Quentin Jammer CB 12.8 - Bryan Scott S 15.7 - Jay Bellamy S 15.7 - Phillip Buchanon CB 17.6 - Most of the bottom 30 defensive backs are the definition of replacement players: they saw from 20 to 50 plays and performed very poorly on most of them, giving up huge gains on a per-play rate. Our Bottom 30 is actually drawn from the bottom 114 players, but it only includes semi-regular players who had at least 100 defensive chances over the last four years. Only three of the 114 lowest-ranked defensive backs have been to the Pro Bowl, and Todd Lyght's only appearance was in 1999, two years to prior to those covered by the dataset. Baltmore's Chris McAlister is the most significant outlier. He also had the third-most penalties of any defensive back, costing his team more than 10 points. At the same time, as a cornerback, McAlister covers a lot of ground that would traditionally belong to the safety. He may have been credited with pass defenses (and completions) in the play-by-play that he wouldn't have had he stayed in a much smaller zone. Defensive play-by-play remains inexact - for the most part, we don't know who the other 10 players on the field are, nor do we know where they lined up. Some of the most obvious questions about defense, such as how far a player ran to make a tackle or whether he beat a block to get there, are impossible to answer today. More in-depth play charting may someday offer clues. But for now, PROTRADE's play-by-play and ES system seem to agree with what coaches, players and fans see when they vote for the Pro Bowl rosters.

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

12-02-2005, 01:31 PM	#4
Bob's your Information Minister Chiefs > Broncos Join Date: Apr 2004 Posts: 25,931 Adopt-a-Bronco: CHRIS KUPER!!!	The same applies to your DBs then.

12-02-2005, 01:38 PM	#6
Willynowei Some dude Football is a wonderful thing. Join Date: Oct 2004 Location: NJ Posts: 2,848 Adopt-a-Bronco: Eddie Royal	most useless crap ever.

12-02-2005, 01:47 PM	#11
Rock Chalk Cheeky Bastards Laus Deo Join Date: Oct 2001 Location: The Backside of the Internet Posts: 29,942 Adopt-a-Bronco: Chris Harris	Yeah the "Red" Knight. As in "I thought I red the play right" (pun n the bad spelling intended)

12-02-2005, 02:02 PM	#16
Drek Ring of Famer Join Date: Mar 2004 Posts: 12,359	Crappy stat analysis. Completion percentages against are very speculative at best. I'd dare call it outright stat mining but without any particular goal. Just throwing individual numbers into equations used for team stats doesn't work well. This study should at least normalize the numbers by team stats along with filtering some of the noise by making a comparative evaluation based on what teammates have done relative to one another, etc..