First.

Interesting article, but nothing all that surprising. Stuff like yards per completion instead of yards per attempt are obvious to readers on this site, but talking heads always talk about yards per attempt.

The concluding sentence in the article has to rank among the all-time brainless ones ever written.

I loved Dungy's quote about opponents running for field goals while they pass for touchdowns. Billick's slam on the NFL stats is pretty awesome. Granted, much of this article's content is already well-known to readers here, but compared to much of what passes for football analysis, this is pretty awesome stuff.

Stats like total yards allowed have been used historically because they were easy (possible) to compute not because they were the best.
Now we have far better tools and ways to share data so better stats are being developed.
It isn't genius it's inevitable , except DVOA - that's genius ;)

The anecdote about Holmgren removing the play from Montana's repetoire is interesting. Also, the article lends insight as to why Childress may be smart in having Adrian Peterson return kicks, while limiting his rushes to around 20 per game. What's more wearing and more valuable; having Peterson return a kickoff to the 45, or having Peterson helping to move the ball from the 22 to the 45?

"Which only proves that a Tom Brady, like Montana and the other greats, is much more comforting to a coach than a glitzy stat sheet."

Nah, Bill Belichick would rather have lots and lots of tape of the opponent's coaches.

BTW, a ESPN Hashmark blog entry from a week or two ago had Mosely mentioning that a number of NFL defensive coordinators told him they love DVOA.

Good article.

Man, the 2001 Patriots remain one of the most confounding teams of all time. By any statistical measure, there is simply no way that team should have competed for a Super Bowl, let alone won one.

And no, I am not saying they cheated.

1 - I think your comment makes more sense if you replace "yards per completion" in your post with "yards per attempt," and "yards per attempt" in your post with "QB rating." Is that what you mean? That would fit the article more.

#3 and 5 Ditto and ditto. Will, I love anecdotes like that. They usually end up buried in biographies 30 years after they were relevant (though I am sure Mr. H shared it with prospective employers when interviewing). A great teaching tool for players today, kids especially! Interesting that a QB with a historic low INT ratio is described as having a gambling streak.

#7 That is WAY cool.

#8 - There must be some statistical measure that explains how the Patriots won games. Its apparently just not any of the major offensive stats.

It was too long ago for me to remember, and to be honest, in 2001 I don't recall seeing any Patriot's games until the Super Bowl.

Well, you can bet they're not all that concerned about some of the dizzying array of information that has been thrown at the fans already this season ... Devin Hester returned his 10th kick for a touchdown in 25 career games

Um, I'll betcha just about every coach is highly concerned about that one, to the point of adjusting special teams strategies and coverages.

Good article, nevertheless.

The importance of starting field position after kickoffs makes me think of the kicker ratings in PFP - how kickoff distince is extremely important, and far more reliable year-to-year than FG%.

FO is ahead of the game again...

8: One thing about the Patriots is, defensively, their biggest weakness was against the run, and in three playoff games, nobody tried to run on them, for various reasons: (Snow, Bettis coming back from Injury, Martz)

Also, by the end of the year, Seymour had become improved quite a bit, and become a dominant player.

Re: #7, #10

Here's the exact quote and a link:

I've had a couple of defensive coordinators tell me they're big fans of the DVOA because it throws out a lot of the garbage stats that opposing teams pile up when they're trailing by large margins.

And the post is here

#6 - "Nah, Bill Belichick would rather have lots and lots of tape of the opponent’s coaches." Indianapolis or Pittsburg?

The list of stats coaches won't pay attention to includes:

Devin Hester returned his 10th kick for a touchdown in 25 career games ...

Anyone else think that maybe they *should* pay attention to that one?

11:
I remember watching the Patriots' conventional stats as the 2001 season progressed. They were an average team, significantly above average in one part of the game only: punts. Every exchange of punts netted them 5-10 yards (If I recall).

It's hard to believe that was their secret weapon, but it helped keep them in games.

#16: I like the comments on Mosley's article. The only thing they missed was that the Cardinals had more swagger.

For me the best stat on whether a team will win a game is Points Scored minus Points Allowed.

There's a 100% correlation that when this is a positive figure the team will win the game, and lose if it's minus. If it's zero they will tie ...

BBS :)

#21, That only determines who wins that day. Doesn't do much good at predicting which team is better overall (see Titans/Colts last year).

I don't see how "tendencies" are really a statistic, but it's an interesting article.

I agree with Mike Nolan about how sacks are overrated. To me the best thing about a sack is that you have a chance of causing a fumble by the QB - otherwise it is just a super sized tackle for a loss. Sure, it gives the D a good chance of ruining a drive, but geez it's not a mega-uber super terrific play. And think about the guys being 'crowned' for having say 16 sacks in a season - 1 per game, 1 per 60 odd snaps. It's a pretty rare event that doesn't change games much at all.

Re: 24

Think about it like this:

Each team gets about 10 drives in a game. Whenever a QB is sacked, the chances of that drive continuing beyond that set of 4 downs is quite low (unless you have Paytom Bramming). Therefore, a sack usually results in the end of 10% of the other team's offense for the entire game.

I agree that it is just a tackle for a big loss, but how often do you see a RB get tackled for a loss of 10?

Add in the greatly increased chance for a fumble, possibly pushing a team out of FG range, and striking fear into the heart of the QB, and it all adds up to a super play.

Just an opinion.

Well, I don't think I'd be going out on too much of a limb to say the most important factor for the Patriots' Superbowl in 2001 was luck. They were worse than the teams they played, especially against that Rams team, but they came away with wins anyway. Sometimes there isn't a great explanation, because sometimes the best team doesn't win.

I thought this article was rife with banalities, at least for people familiar with FO. I did really like the Holmgren anecdote, because it makes me wonder about how many similar situations occur for other teams. Are there simple problems of stubbornness with other QBs and coaches that should get fixed?

Also, I don't understand the animosity towards QB rating. I mean, compare the top 10 QBs in passer rating:
1. Tom Brady
2. Jake Delhomme
3. Peyton Manning
4. David Garrard
5. Jeff Garcia
6. Kurt Warner
7. Ben Roethlisberger
8. Tony Romo
9. Jon Kitna
10. Matt Schaub

to the top 10 in VOA (the fair comparison, since QB rating isn't opponent adjusted):
1. Tom Brady
2. Peyton Manning
3. David Garrard
4. Jeff Garcia
5. Kurt Warner
6. Tony Romo
7. Jake Delhomme
8. Derek Anderson
9. Ben Roethlisberger
10. Matt Hasselbeck

So the only names that are different are Anderson/Hasselbeck for Kitna/Schaub, although Schaub is #11 in VOA and Hasselbeck is #11 in rating. I think the only real flaw in the passer rating formula is that touchdowns, while sometimes a good indicator, are often a better indicator of coaching tendencies in the red zone than actual QB performance. Hence Jon Kitna's place of honor in the QB rating and his poor place in VOA.

26: I think the major problem with QB rating is it doesn't count sacks as negative, so a QB who throws the ball away instead of taking a sack gets punished by QB rating. Also it doesn't consider down and distance, but including that would make it a lot harder to compute.

I was glad to see that they used points per red zone possession instead of percentage. It always irks me when they combine field goals and touchdowns and show scoring percentage in red zone. It seems to make more sense to me to show points per possession.

27: Well, down and distance is certainly a big step, which would almost make it into VOA. Sacks are a good point; at the very least they should be counted as negative pass plays, rather than subtracting yards from "team" passing yards.

Still, all these coaches claiming it's meaningless, or saying that it doesn't show who good QBs are, don't seem like they know what they're talking about.

This article and the comments here make me wonder: What would be the most misleading drive/play/game for each of the primary stats and separately for DVOA/DPAR.

Consider: 99 yard pass play, and fumble at the 1 behind by 3 with 1:59 remaining. Would be very bad but look good for YPC, YAC, Comp %, but be very bad.

Perhaps first and goal following a turnover - intercepted pass in the endzone and a Troy Brown strip for a TD. Good result but very bad for many stats.

The follow up might be to consider plays that are relatively common that wreak statistical havoc.

Any takers?

Re:2001 Pats. While not ignoring that they probably had good game film, the Giants beat the Bills in a SB with Belichick as the DC. It was a similar game-the Giants pounded the WRs, and kept the ball away from the Bills spread offense. Something like 45 minutes of possession for the Giants. And it was another Superbowl that came down to a FG.

30: Tecmo Super Bowl took away any stats you got on a play that ended in a fumble. Maybe it was prescient rather than bad programming.