The Defense: Eye test vs. the Numbers

[ivanvamp]

I know what my eyes tell me: the Patriots defense is better now than it was earlier in the year. I certainly have more confidence in them now than I did earlier. The switch to Talib/Dennard at CB with McCourty as S really has helped matters out a ton.

At least that's what it looks like to my eyes.

Do the numbers back it up? Unfortunately, no. I broke the season up into two halves: weeks 1-8, then weeks 9-18 (including the playoff game vs. Houston). Here's what I found.

First, the raw numbers:

Points allowed per game:
- Wks 1-8: 21.3
- Wks 9-18: 21.0

(so that's good, right? a little improvement)

Yards allowed per game:
- Wks 1-8: 369.8
- Wks 9-18: 382.1

(so that's bad, right? more yards allowed per game)

But we might want to focus on that 0.3 ppg improvement and say, hey, they're doing better. Well, maybe, maybe not. Let's look at the quality of offenses they faced in those periods of time. I'm going by those offenses' final rankings and averages. Too much of a pain to go through and figure out where those teams were at that point in the season.

Avg Offensive Ranking by Points:
- Wks 1-8: 18.6
- Wks 9-18: 19.8

Avg Offensive Points Scored:
- Wks 1-8: 21.9
- Wks 9-18: 21.1

So they faced slightly worse offenses in terms of scoring, by 0.8 ppg. What about yards produced?

Avg Offensive Ranking by Yards:
- Wks 1-8: 20.9
- Wks 9-18: 18.6

Avg Offensive Yards Gained:
- Wks 1-8: 331.0
- Wks 9-18: 337.0

So their 2nd half opponents tended to gain a few more yards per game than their 1st half opponents did over the course of the entire season.

Now, let's look at the Patriots' performance relative to their opponents' offensive production.

Weeks 1-8:
- The average points scored by their opponents over the course of their (opponents') season was 21.9. The Patriots held them to an average of 21.3. That's a +0.6 figure for the Patriots.

- The average yards gained by their opponents over the course of their (opponents') season was 331.0. The Patriots held them to an average of 369.8. That's a -38.8 figure for the Patriots.

Weeks 9-18:
- The average points scored by their opponents over the course of their (opponents') season was 21.1. The Patriots held them to an average of 21.0. That's a +0.1 figure for the Patriots.

- The average yards gained by their opponents over the course of their (opponents') season was 337.0. The Patriots held them to an average of 382.1. That's a -45.1 figure for the Patriots.

So in both points allowed and yards allowed, the Patriots have been *worse* during the 2nd half of the season than they were during the 1st half of the season, relative to the offensive capability of their opponents.

The numbers here do not support what my eyes tell me.

Now let's see how the Pats have done this year against above-average offenses. They've only played 7 games against offenses in the top 16 in the league: Bal, Den, Sea, Ind, Hou, SF, and Hou again. Here's what they've done in those games:

Avg Rank (Pts): 9.4
Avg Rank (Yds): 10.3

Avg PPG: 25.7
Pats allowed: 26.1
Pats +/-: -0.4

Avg YPG: 367.1
Pats allowed: 407.0
Pats +/-: -39.9

Long story shorter, we have good reason, statistically, to believe that Baltimore will put up both yards and points on the Patriots. It *feels* like the Pats' D is better now than it was earlier in the year, but the numbers suggest that's just not true, unfortunately.

Baltimore averages 24.9 points and 352.5 yards per game this year (regular season numbers). I fully expect them to have around 24-27 points against the Pats on Sunday. I just hope the Pats can put up 28-31 on the Ravens' old and slow defense.

 

[ThatllMoveTheChains!!!]

Talib didn't play until week 10; splitting up the season by before and after yields:

• 22.3 ppg before and 19.75 ppg after
• 382 ypg before and 370 ypg after
• 285 passing ypg before and 263 passing ypg after

Pretty much every metric either improved or stayed about the same after adding Talib. If you drop the SF game which was skewed pretty heavily by offensive TOs and poor STs the ppg number improves to 16.7. So the stats actually back up the eyeballs if you split them based on the arrival of Talib rather than the halfway point.

 

[randomk1]

What D do you want to go against Ravens with? The one from week 3 or the current lineup? We've been relatively stout against the run all year but the coverage in the secondary seems far better and you don't have guys constantly catching balls unchallenged. The only bad game they had was against the 49ers imo but the entire team had a bad game.

 

[ALP]

I was also going to mention taking out the SF game, and to equal that take out worst game from group 1 as well...

I believe the SF skews the data quite a bit

 

[Kontradiction]

Talib didn't play until week 10; splitting up the season by before and after yields:

• 22.3 ppg before and 19.75 ppg after
• 382 ypg before and 370 ypg after
• 285 passing ypg before and 263 passing ypg after

Pretty much every metric either improved or stayed about the same after adding Talib. If you drop the SF game which was skewed pretty heavily by offensive TOs and poor STs the ppg number improves to 16.7. So the stats actually back up the eyeballs if you split them based on the arrival of Talib rather than the halfway point.

Yup. That San Fran game skewed things quite a bit by giving their offense a short field to work with a lot of the time. Further, after the offense stopped turning the ball over, the defense held them in check for a good amount of time to help the offense get us back into the game.

 

[Kontradiction]

What D do you want to go against Ravens with? The one from week 3 or the current lineup? We've been relatively stout against the run all year but the coverage in the secondary seems far better and you don't have guys constantly catching balls unchallenged. The only bad game they had was against the 49ers imo but the entire team had a bad game.

You obviously go with the more current version since the Week 3 version sucked gorilla balls.

 

[BradyManny]

Very interesting breakdown ivanamp - appreciate the hard work and taking a closer look. But I think as others stated, the improvement of this D started later. It really started with the Jets game, I would say, and I think as ThatllMoveTheChains posted, the numbers are much more strikingly different starting there.

The big thing is that our opposing QB passer rating has improved dramatically from the midseason switch (acquiring Talib, McCourty to safety), I believe.

The bottom line for me is, you got Vince Wilfork in the middle, you got 3 absolute beast linebackers behind him, you have one of our most versatile edge players we've ever had in Ninko out there, and in the secondary, you have McCourty playing at a Pro Bowl level, a young, feisty corner in Dennard playing well, and Talib who is at least good in man coverage. That's a whole lot more talent than a lot of D's have, so why should we be surprised they are good now?

 

[JJDChE]

The eye test tells me they are far and away better. Not just a little better.

Outside of the 9ers game, which was just an egg all around, teams have been have big time trouble putting up more than ~14-17 points against this defense since Talib came in.

These are the opponents points coming into the 4th qtr (except SF) since the Buffalo game:

17
12
10
7
13
0
13

Most games were decided by that point and whatever happened in the 4th was garbage time scoring.

That scoring rate equates to about 13.7 ppg (the same analysis yields 18.6 ppg for the first 9 games.). To me, that's about the level where this defense is playing, ~5 ppg better.

 

[PATSYLICIOUS]

Talib didn't play until week 10; splitting up the season by before and after yields:

• 22.3 ppg before and 19.75 ppg after
• 382 ypg before and 370 ypg after
• 285 passing ypg before and 263 passing ypg after

Pretty much every metric either improved or stayed about the same after adding Talib. If you drop the SF game which was skewed pretty heavily by offensive TOs and poor STs the ppg number improves to 16.7. So the stats actually back up the eyeballs if you split them based on the arrival of Talib rather than the halfway point.

^ This post is a much more accurate analysis if you truly want to compare old vs new defense. The numbers should be even more in favor of the new defense considering:

1. The 24 in the Indy game could be alot lower if the new defense was in midseason form, which they weren't obviously being the very first time Talib was on the field and judging by the first few drives. But once they adjusted they were dominant in that game. Not a big deal but something to consider.

2. The 49er game kills the average too and Talib/Dennard were banged up in that one. So in actuality that game would be better suited as grouped in with the old defense.

3. The garbage points Houston scored this week and the ST disaster that game was. I can't remember a single game in the first group where we gave up that many points based off prevent defense and multiple scores off ST miscues. I guess the AZ game is somewhat comparable with the blocked punt and INT in our territory but it's still a good 7-10 extra points from that.

So I'd say the new defense trounces the old one in the eye test and numbers test.

 

[patsfanincleveland]

I know what my eyes tell me: the Patriots defense is better now than it was earlier in the year. I certainly have more confidence in them now than I did earlier. The switch to Talib/Dennard at CB with McCourty as S really has helped matters out a ton.

At least that's what it looks like to my eyes.

Do the numbers back it up? Unfortunately, no. I broke the season up into two halves: weeks 1-8, then weeks 9-18 (including the playoff game vs. Houston). Here's what I found.

First, the raw numbers:

Points allowed per game:
- Wks 1-8: 21.3
- Wks 9-18: 21.0

(so that's good, right? a little improvement)

Yards allowed per game:
- Wks 1-8: 369.8
- Wks 9-18: 382.1

(so that's bad, right? more yards allowed per game)

But we might want to focus on that 0.3 ppg improvement and say, hey, they're doing better. Well, maybe, maybe not. Let's look at the quality of offenses they faced in those periods of time. I'm going by those offenses' final rankings and averages. Too much of a pain to go through and figure out where those teams were at that point in the season.

Avg Offensive Ranking by Points:
- Wks 1-8: 18.6
- Wks 9-18: 19.8

Avg Offensive Points Scored:
- Wks 1-8: 21.9
- Wks 9-18: 21.1

So they faced slightly worse offenses in terms of scoring, by 0.8 ppg. What about yards produced?

Avg Offensive Ranking by Yards:
- Wks 1-8: 20.9
- Wks 9-18: 18.6

Avg Offensive Yards Gained:
- Wks 1-8: 331.0
- Wks 9-18: 337.0

So their 2nd half opponents tended to gain a few more yards per game than their 1st half opponents did over the course of the entire season.

Now, let's look at the Patriots' performance relative to their opponents' offensive production.

Weeks 1-8:
- The average points scored by their opponents over the course of their (opponents') season was 21.9. The Patriots held them to an average of 21.3. That's a +0.6 figure for the Patriots.

- The average yards gained by their opponents over the course of their (opponents') season was 331.0. The Patriots held them to an average of 369.8. That's a -38.8 figure for the Patriots.

Weeks 9-18:
- The average points scored by their opponents over the course of their (opponents') season was 21.1. The Patriots held them to an average of 21.0. That's a +0.1 figure for the Patriots.

- The average yards gained by their opponents over the course of their (opponents') season was 337.0. The Patriots held them to an average of 382.1. That's a -45.1 figure for the Patriots.

So in both points allowed and yards allowed, the Patriots have been *worse* during the 2nd half of the season than they were during the 1st half of the season, relative to the offensive capability of their opponents.

The numbers here do not support what my eyes tell me.

Now let's see how the Pats have done this year against above-average offenses. They've only played 7 games against offenses in the top 16 in the league: Bal, Den, Sea, Ind, Hou, SF, and Hou again. Here's what they've done in those games:

Avg Rank (Pts): 9.4
Avg Rank (Yds): 10.3

Avg PPG: 25.7
Pats allowed: 26.1
Pats +/-: -0.4

Avg YPG: 367.1
Pats allowed: 407.0
Pats +/-: -39.9

Long story shorter, we have good reason, statistically, to believe that Baltimore will put up both yards and points on the Patriots. It *feels* like the Pats' D is better now than it was earlier in the year, but the numbers suggest that's just not true, unfortunately.

Baltimore averages 24.9 points and 352.5 yards per game this year (regular season numbers). I fully expect them to have around 24-27 points against the Pats on Sunday. I just hope the Pats can put up 28-31 on the Ravens' old and slow defense.

The problem revolves around the utilization of statistics.

Look at last Sunday. A 28 point, 425 yard surrender would statistically indicate "poor" defense. What did your eyes tell you?

The reality is Houston entered Q4 with a 38-13 deficit and 269 yards.

That prorates to 17 points and 358 yards. All 13 points were of short fields.

The final 15 minutes of "garbage time" was 15 points 156 yards. That prorates to 60 points and 624 yards.

Your stats don't care when yards or points were scored.

 

[ivanvamp]

Talib didn't play until week 10; splitting up the season by before and after yields:

• 22.3 ppg before and 19.75 ppg after
• 382 ypg before and 370 ypg after
• 285 passing ypg before and 263 passing ypg after

Pretty much every metric either improved or stayed about the same after adding Talib. If you drop the SF game which was skewed pretty heavily by offensive TOs and poor STs the ppg number improves to 16.7. So the stats actually back up the eyeballs if you split them based on the arrival of Talib rather than the halfway point.

Talib's first game with NE was against Indy (he had a pick six in that game for the Pats). So let's break it up first 9 and then last 8.

First 9
Opp Avg PPG: 21.8
NE PPG allowed: 22.3
NE +/-: -0.5

Opp Avg YPG: 332.4
NE YPG allowed: 382.1
NE +/-: -49.7

Last 8
Opp Avg PPG: 21.1
NE PPG allowed: 19.8
NE +/-: +1.3

Opp YPG allowed: 336.2
NE YPG allowed: 369.8
NE +/-: -33.6

So to compare:

PPG +/-
W/O Talib: -0.5
W/ Talib: +1.3

YPG +/-
W/O Talib: -49.7
W/ Talib: -33.6

So I guess if that's where the split is, then yes, the defense has improved some. Not a lot, though I'd suggest the overall difference of +1.8 in PPG differential is actually something substantial.

For the record, if they gave up 19.8 ppg and 369.8 ypg for the season, here's where they would rank:

#7 points allowed
#25 yards allowed

 

[Nunchucks]

I think that the important stat to look at is 3rd down efficiency. Are they getting off the field more often?

 

[Marqui]

Don't you have to take out the Jacksonville game to? Talib nor Dennard played in that game. And McCourty wasn't at safety.

 

[mayoclinic]

Look at last Sunday. A 28 point, 425 yard surrender would statistically indicate "poor" defense. What did your eyes tell you?

The reality is Houston entered Q4 with a 38-13 deficit and 269 yards.

That prorates to 17 points and 358 yards. All 13 points were of short fields.

The final 15 minutes of "garbage time" was 15 points 156 yards. That prorates to 60 points and 624 yards.

Your stats don't care when yards or points were scored.

The Houston game is even more misleading:

1. Houston started out with a 94 yard kickoff return to the 12 yard line. The defense held the Texans to 3 yards, forcing a field goal. Result: defense "gives up" 3 points.

2. In the next 4 drives the defense held the Texans to 118 yards, 6 first downs and 0 points, while the Pats put up 17 unanswered points to take a 17-3 lead:

- 38 yards, 1 first down, punt
- 16 yards, 1 first down, punt
- 27 yards, 2 first downs, punt
- 37 yards, 2 first downs, punt

So with 3:38 left in the first half the Pats' D had given up 121 yards, 6 first downs, and 3 points to the Texans, the points coming on a long kickoff return. I'd call that a very good defensive performance.

3. Another nice kickoff return by Manning plus a 15 yard penalty on Gostkowski for a touchdown-saving horse collar tackle results in good field position at the NE 47. The run defense disappears, and Adrian Foster gets all of the Texans' yardage in 5 plays leading to a TD, including runs of 21 and 19 yards.

4. The Pats failed to run off the clock to end the half and gave Houston decent starting field position at their own 38, allowing them to get in range for a very long (55 yard) field goal. So the Texans end up with 194 yards and 13 points on the half, even though they were pretty much shut down for 26 minutes of the half while the Pats build a commanding lead.

5. In the first 3 drives of the 2nd half the defense held the Texans to 71 yards of offense, 4 first downs and 0 points, during which time the Patriots put up 21 unanswered points to take a decisive 38-13 lead with 13 minutes left to play in the game:

- 5 yards, 0 first down, punt
- 47 yards, 3 first downs, interception
- 19 yards, 1 first down, change of possession on loss of downs

6. Another long Manning kickoff return gave the Texans good field position at the NE 37, allowing them to score to make the game 38-20. The Patriots went to a "prevent" defense, allowing the Texans to put together their only sustained long drive of the game (11 plays, 73 yards 4 first downs including the TD on 4th and goal).

The stats are very misleading. Among the things that the stats don't clearly show:

- For almost 43 minutes of the game - nearly 3 full quarters worth of play - the defense held the Texans to 0 points, 188 yards and 10 first downs on 7 drives, while the offense put up 38 unanswered points.
- The Texans sustained only 1 drive of greater than 50 yards, and 4 drives of 40 yards or more (including the last drive in garbage time).
- The Texans' longest drive produced only 4 1st downs, and that required a 4th and goal. They had drives of 4, 3, 2, 2, 2, 2, 2, 1, 1, 0 and 0 first downs.
- The longest Houston offensive play was 28 yards, so the defense did not give up any really huge plays.
- The defense made a key interception and a key 4th down stop leading to change of possession and 14 points by the offense, allowing the Pats to put the game pretty much out of reach with 13 minutes to play.
- 17 of Houston's 28 points came after long kickoff returns by Danieal Manning, with field position starting at the NE 12, 47 and 35 yard line.

I'd call that a pretty solid defense performance, despite how the box score ended up.

 

[BSR]

According to Football Outsiders their defense was ranked 18th after their 8th game of their season and 15th by the end of the year so a slight improvement there as well. After these couple of playoff games, however, they moved to 12th.

In the end the improvement I think has been significant but not exponential. The difference is that playoff football tends to be different than regular season football. I would expect them to be able to play much tighter than they did in the regular season.

 

[Nunchucks]

Patriots defenses always get better in the playoffs.

 

[mayoclinic]

I think that the important stat to look at is 3rd down efficiency. Are they getting off the field more often?

That's clearly one. Some recent games:

- Houston (playoffs): 4-15 (27%)
- Miami: 5-13 (38%)
- Jacksonville: 4-15 (27%)
- San Francisco: 5-14 (36%)
- Houston: 4-14 (29%)
- Miami: 3-13 (23%)
- New York: 3-10 (30%)
- Indianapolis: 8-14 (57%)

That's pretty dramatic, when you consider some of our recent 3rd down woes. That works out to 36-108 (33%), which would have ranked in the top 5 of the NFL if projected over the full 16 games. And that number was skewed by the Indy game, which was Talib's first game.

So I'd say the defense has improved dramatically in terms of doing the critical job of getting opposing offenses off the field on 3rd down.

Another key thing to look at is big plays:

- Houston (playoffs): longest play was 28 yards. 5 plays of greater than 20 yards for the game. One scoring play of greater than 20 yards (the Posey TD catch).
- Miami: longest play was 22 yards. 3 plays of greater than 20 yards, 0 for scores.
- Jacksonville: longest play was 53 yards (ouch!). 2 plays over 20 yards, though both were fairly long (36 and 53 yard receptions). Neither was for a score.
- San Francisco: longest play was 38 yards. 6 plays over 20 yards, 3 over 30. The killer was 4 TD plays over 20 yards - passes of 24, 34, 27 and 38 yards. Obviously the worst game of the group.
- Houston: longest play was 30 yards. 5 plays over 20 yards, none of them a scoring play.
- Miami: longest play was 28 yards. 4 plays over 28 yards, none of them a scoring play.
- New York: longest play was 39 yards. 5 plays over 20 yards, none for a score. All came after the game was a rout.
- Indianapolis: longest play was 43 yards. 6 plays over 23 yards, including a 43 yard TD pass to TY Hilton, which came well after the game was decided.

So over 8 games the Pats have given up 36 plays of 20 yards or greater, for an average of 4.5/game, including 2 plays of 40 yards or more. 6 of those 36 plays were long TDs, with 4 of them coming in the San Francisco game. But overall, the defense is doing a much better job of cutting down on giving up the big play.

So the defense seems to be doing a much better job of getting off the field and not giving up the big play. That works for me.

 

[Tunescribe]

NOBODY in the national media is talking about the Patriots defense except for Wilfork and occasional mentions of Talib. It's all Ray Lewis, Suggs, Reed, Ngata and the Ravens defense. I hope this ticks the Pats off. They need to get a pass rush on Dorky Joe.

 

[Nunchucks]

So, it looks like the defense is getting the offense more possessions. That is really what the problem I have had with the defense for the past few years. They were not getting off the field, which resulted in less possessions for the offense and thus less scoring opportunities.

 

[Nunchucks]

NOBODY in the national media is talking about the Patriots defense except for Wilfork and occasional mentions of Talib. It's all Ray Lewis, Suggs, Reed, Ngata and the Ravens defense. I hope this ticks the Pats off. They need to get a pass rush on Dorky Joe.

I am enthused by this years version of the Patriots defense, they have really shaped up and are playing very well. Garbage time excluded of course.

Best defense I've seen the Patriots have in a long time. Hope they resign Talib, will be even better next year.