Football Outsiders prospects - metrics and other goodies

[BradyManny]

Anyone pick this up? I've perused it for some relevant Pats information...

A$ante v. Hobbs:
Note: Originally I thought Hobbs had the better success rate, but when I read over the legend, football outsiders does success rate for DBs based on how often the DB was successful, normally the stat is just completion % against.

Hobbs had a 47% success rate, 8.6 yd/attempt, 25% success rate against the run & 8 yd/run in his direction (can we attribute that to the injuries??). He also had 14 pass deflections.
Asante had a 51% success rate, 6.3 yd/attempt, 44% rate against the run, but 12 yd/run in his direction (ouch). He had 17 pass deflections.

So the difference between the two isn't major. Hobbs clearly played with a little bit bigger cushion as his yards per completion was higher, but that's about it. Neither were good against the run.

Fernando Bryant stacks up well:
Obviously, coming from a diff system, the numbers can't be compared directly with Hobbs/Asante, but worth looking at:
45% success rate, 6.5 yd/attempt, 11 pass deflections, 50% success rate stopping the run.

So, allowed a higher completion %, but for less yards, had plenty of deflections and was stronger against the run than the incumbents.


For those that think Ty Law is over the hill, his metrics measure up quite well:

55% success rate, 6.9 yd/attempt, 12 deflections, 2 ints, 5.7 RuYd


One thing that stood out quite a bit, Ty Warren may be even more underrated than we thought and Big Sey did not match the effectiveness of his line-mates...

Warren and Wilfork were 18th and 26th in the league, respectively, with the percentage of stops for plays in their direction. Seymour was down at 47. Specifically against the run, Warren had an 80%, Wilfork was 81% and Sey was 75%. Imagine how dominant this line will be with a healthy Seymour.

The numbers say that Bruschi was [according to these potentially flawed statistics] the least effective backer in the group. He was effective on 56% of the runs in his direction. Specifically take the stats with a grain of salt, the guys on the outside did better than the guys on the inside, maybe thats just part of the system. AD was 64%, both Vrabel and Colvin were 71% and Seau was 61%. Bruschi also had a 33% success rate against the pass, but AD and Seau - both who also played inside - had lower rates there, too, 40% and 53%.


More to come later if anyone is interested...

note - edited after reading stat legend better...

 

[xmarkd400x]

I thing Yds/Cmp is stupid (no attack on you, OP). Yds/Att would be better. Imagine if vs. Hobbs was 1/10 with the 1 completion going for 10 yards. Yds/Cmp = 10 (looks pitifut); Yds/Att = 1 (actually pretty damn good). At any rate, I'll try to break it down a bit.

==============================
Hobbs: 53% completed against, 8.6 ypc = 4.55 ypa
Samuel: 49% completed against, 6.3 ypc = 3.087 ypa

That tells a pretty big difference in the stat lines in my opinion.

Others:

Law: 45% completed against, 6.9 ypc = 3.105 ypa
Bryant: 55% completed against, 6.5 ypc = 3.575 ypa
===============================

Methodology: I figured the following:
[(%completed_against * ypc) + (%not_completed * ypnc)] / 100% = YPA

Since ypnc (yards per not complete) = 0, the %not_complete * ypc = 0, and can be reduced to:
(%completed_against * ypc)/100 = ypa

 

[unoriginal]

You can attribute long runs to the corners due to the Pats penchant for playing their corners in deep thirds.

 

[BradyManny]

I thing Yds/Cmp is stupid (no attack on you, OP). Yds/Att would be better. Imagine if vs. Hobbs was 1/10 with the 1 completion going for 10 yards. Yds/Cmp = 10 (looks pitifut); Yds/Att = 1 (actually pretty damn good). At any rate, I'll try to break it down a bit.

==============================
Hobbs: 53% completed against, 8.6 ypc = 4.55 ypa
Samuel: 49% completed against, 6.3 ypc = 3.087 ypa

That tells a pretty big difference in the stat lines in my opinion.

Others:

Law: 45% completed against, 6.9 ypc = 3.105 ypa
Bryant: 55% completed against, 6.5 ypc = 3.575 ypa
===============================

Methodology: I figured the following:
[(%completed_against * ypc) + (%not_completed * ypnc)] / 100% = YPA

Since ypnc (yards per not complete) = 0, the %not_complete * ypc = 0, and can be reduced to:
(%completed_against * ypc)/100 = ypa

Actually, I think I've once again read one of their stats wrong, I'm just reading through the "how to read these stats" section again, and they again do their stats a little differently than I'm used to. It's not YPC they were measuring, it was in fact YPA, which as you say is more meaningful. They also have an ADJUSTED yards per attempt (adjusted to quality of the receiver), which frankly rarely differs much from unadjusted.

If anyone out there is better at figuring out these stats than me, please chime in...

 

[BradyManny]

You can attribute long runs to the corners due to the Pats penchant for playing their corners in deep thirds.

Good point.

 

[trench]

It's not YPC they were measuring, it was in fact YPA, which as you say is more meaningful. They also have an ADJUSTED yards per attempt (adjusted to quality of the receiver), which frankly rarely differs much from unadjusted

The number you listed was for Adjusted Yards per Pass (APaYd) which is the average number of yards gained per pass play where the particular defender was targeted adjusted for the quality of the receiver defended. The non adjusted values (PaYd) for the same receivers are 8.5 yds for Hobbs and 6.0 for Samuel.

 

[Metaphors]

I don't trust these stats without knowing how they determine when a DB is thrown at. That is a subjective judgement which can lead to skewed stats. Remember the Steelers game last year and the deep TD to a RB (Davenport?). It was a blown coverage and Hobbs peeled off his responsibility to help, getting close to the play but unable to stop the reception. He was the closest defender to the play, so does that count against him? Like all stats, these should probably be taken with a grain of salt and not used to draw any definitive conclusions without context.

 

[BradyManny]

I don't trust these stats without knowing how they determine when a DB is thrown at. That is a subjective judgement which can lead to skewed stats. Remember the Steelers game last year and the deep TD to a RB (Davenport?). It was a blown coverage and Hobbs peeled off his responsibility to help, getting close to the play but unable to stop the reception. He was the closest defender to the play, so does that count against him? Like all stats, these should probably be taken with a grain of salt and not used to draw any definitive conclusions without context.

No doubt. Football Outsiders does a pretty good job of trying to be as objective about stats as you can be, though. That's why their stats often go over my head as they try to take into account variables that otherwise would make stats useless. For instance, they do take into account zone defense - if a pass hits the seam of a zone and there's no defender seemingly defending, its listed as "hole in the zone" and they make a judgment about which defender was responsible.

 

[BradyManny]

The number you listed was for Adjusted Yards per Pass (APaYd) which is the average number of yards gained per pass play where the particular defender was targeted adjusted for the quality of the receiver defended. The non adjusted values (PaYd) for the same receivers are 8.5 yds for Hobbs and 6.0 for Samuel.

Awesome - thanks, as I said, I'm still trying to figure these things out so any insight is greatly appreciated

 

[Box_O_Rocks]

I don't trust these stats without knowing how they determine when a DB is thrown at. That is a subjective judgement which can lead to skewed stats. Remember the Steelers game last year and the deep TD to a RB (Davenport?). It was a blown coverage and Hobbs peeled off his responsibility to help, getting close to the play but unable to stop the reception. He was the closest defender to the play, so does that count against him? Like all stats, these should probably be taken with a grain of salt and not used to draw any definitive conclusions without context.

It would also be worth while to know if the Safeties were rolled to one side of the field or another.

 

[trench]

Awesome - thanks, as I said, I'm still trying to figure these things out so any insight is greatly appreciated

It doesn't help that there is a typo involved when they explain APaYd.

 

[wdkantro2]

Sorry, but I have a lot of difficulty buying into these type of stats for the Pats defense. Hobbs and Asante both played CB, but they are much different positions. Asante consistently covered the flat while Hobbs would often cover the deep third in the cover 3. So obviously, Asante's average yards alllowed per completion will be lower just based on the system. Too much goes into it here to just use stats to explain something.

 

[trench]

Sorry, but I have a lot of difficulty buying into these type of stats for the Pats defense. Hobbs and Asante both played CB, but they are much different positions. Asante consistently covered the flat while Hobbs would often cover the deep third in the cover 3. So obviously, Asante's average yards alllowed per completion will be lower just based on the system. Too much goes into it here to just use stats to explain something.

Actually, this is more or less covered in those stats. There is a portion of the same stat line that refers to the average distance the ball traveled in the air beyond the line of scrimmage before reaching the targeted defender. That number is higher for Hobbs (13.6 yds) than it is for Samuel (11.4 yds). So you can pretty much figure out from the stats that one likely plays deeper than the other as well as being able to relate that value with the respective yards allowed.