Advanced metrics

[primetime]

I like to read Advanced Football Analytics because it uses statistical models to derive advanced statistics, unlike PFF which is just using qualitative analysis and assigning it numbers so it looks like a metric.

Especially interesting are their defensive statistics, which I like because they seem to match up with a lot of what the general observations on the field say (unlike PFF) though they tend to reveal some things that we may not notice, especially in comparison to other teams. Here's the link to the Patriots individual defensive statistics

(Glossary is here since it's not entirely clear at first what EPA and WPA are)

Some interesting tidbits

• Ninkovich has been a top-15 defensive end in the NFL this season by their win probability metrics, which is interesting given the tendency for fans to rag on him. I've always thought Ninkovich's contributions were underappreciated. Chandler Jones doesn't make an appearance in the top-50 defensive ends.
• Chris Jones and Wilfork have both been effective, but Vellano and Siliga were not. Siliga's ineffectiveness was likely injury related, while Vellano was pretty clearly a stiff. Interestingly, Buchanan's 1 game was really good, so it's too bad he got hurt.
• Mayo has been a top-10 linebacker this season, and Hightower has been highly effective as well. Collins has been the least effective starting linebacker. Not a surprise.
• Dennard doesn't qualify for the leader boards since he only played 2 games, but his per game average would place him as the 7th best corner in the league. Revis falls in the top-25.
• McCourty has been okay. Patrick Chung sucks.

Offensive stats are here

• The offensive line has been awful, but not as bad as the Bills or Jets!
• The stats don't like Brady much. That's not a surprise, though he was the third best quarterback this past week, so small sample.
• The stats like Vereen and don't like Ridley. I think they tend to overrate receiving backs a little, though.
• Edelman is a top-20 receiver. Everyone else is not great. Amendola really sucks.
• Tim Wright, in limited action, has been a top-10 tight end on a per play basis.

Important to remember it's a small sample size, so one bad week can really skew things, but I thought you might all find this more interesting than PFF or whatever.

 

[Ring 6]

Chandler Jones only played 1 game according to these numbers.
If I understand the descriptions properly, the only plays taken into consideration are positive plays for the defense, which seems to be an incomplete analysis.

 

[primetime]

Chandler Jones only played 1 game according to these numbers.
If I understand the descriptions properly, the only plays taken into consideration are positive plays for the defense, which seems to be an incomplete analysis.

I think it screwed up and has Jones listed separately. Chdt Jones-95 and Ch. Jones-95. Likely because the play-by-play listed it as the latter when Chris Jones was out and the former when Chris Jones was healthy. I noticed that last week because I was trying to figure out where the "t" in Chdt was coming from.

I understand the rationale behind using only positive plays, since the reasoning makes sense. An open field tackle 30 yards down the field is a positive play for a player but a negative one for the defense. Limiting the analysis in such a way prevents a lot of subjectivity from seeping in (who's "to blame" for a given play), but you're right insofar as it is incomplete.

 

[Rob0729]

All these advanced metrics are overrated no matter who does them. Many of them fly in the face of what you can plainly see on the field.

Football is not baseball. Stats can be misleading in football. Doing things with the stats to make them have more added value does not change that.

 

[pencilneckgeek]

Your link points to the wrong site for some reason:
It's http://www.advancedfootballanalytics.com/

 

[JackBauer]

Ninkovich has been a top-15 defensive end in the NFL this season by their win probability metrics, which is interesting given the tendency for fans to rag on him. I've always thought Ninkovich's contributions were underappreciated.

Not a Ninkovich detractor, really, but that seems a bit difficult to believe.

 

[mayoclinic]

All these advanced metrics are overrated no matter who does them. Many of them fly in the face of what you can plainly see on the field.

3-2 is the only metric that counts right now.

 

[pencilneckgeek]

I'm a big fan of AFA and have been for years. Their game strategy articles are excellent, beginning with the 4th down study, three seasons before the big 4th and 2 debate, which to them was a no-brainer. I put more stock in the advanced team stats than the advanced individual stats, since the fundamental measure of the stats is the increased probability of team success and a player's involvement in those plays is much more loosely connected. The team stats, however, translate directly, and I regularly consult them when evaluating teams. The stats don't tell a full story, of course, especially early in the year, when the strength of opposition for each team is more variable. However, I think EPA and WPA, whether broken-out by pass/run or integrated, are much better indicators of the strength of a team than yardage allowed or defensive QB rating or even points allowed, as that can be a function of special teams play, turnovers, situational play, etc.

 

[primetime]

Your link points to the wrong site for some reason:
It's http://www.advancedfootballanalytics.com/

Fixed, sorry.

 

[Danger Zone]

Sorry, nothing with "metrics" in the title can ever be fully applicable to American football which is all about the yards and feet and not the meters and .33 meters..

 

[Saint Pat]

Don't need advanced metrics to know Patrick Chung sucks...