Writers of Pro Football Prospectus 2008

29 Sep 2005

Dr. Z: Giant Mistake

Dr. Z is sick of Jeremy Shockey whining to the media. Who isn't? The good doc also points out the similarities between last week's Patriots-Steelers game and the one that broke the New England winning streak in 2004, and he's a big fan of the New England defensive line. Also feel free to discuss his latest power rankings. Doing these myself this year, I realize how silly it is to base things on win-loss records this early in the year. I mean, really, does Dr. Z think Washington is a better team than San Diego? No, of course not. Now I don't think that Cincinnati is a better team than New England, either, but at least I can blame a heartless statistical formula that doesn't have enough evidence to work correctly yet. Dr. Z's ratings are purely subjective.

Posted by: Aaron Schatz on 29 Sep 2005

36 comments, Last at 03 Oct 2005, 9:23am by Flux

Comments

1
by pawnking (not verified) :: Thu, 09/29/2005 - 6:37pm

Since everyone asked (ok, since everyone else seems to have one), here's...
PAWNKING'S TOP TEN NFL TEAMS!!!

1) Philadelphia - Looked bad for about a quarter against a good Atlanta team. Otherwise, all of their poor play is due to an injured kicker and QB. If McNabb gets healthy, no one beats them, period. I believe he's going to be healthy, so they're on top.

2) Patriots - Only second because they really looked bad for most of the Carolina game. Other than that, they've beaten Pittsburgh on the road and handled Oakland at home.

3) Pittsburgh - A team that runs the ball and stops the run is one of the best teams in the NFL, who'd a thunk it? Yes, they have a home loss, but all things considered, I like them a whole lot.

4) Indy - Struggling against two crappy teams and beating a good team at home doesn't get you number one.

5) Cincinnatti - Neither does clobbering three crappy teams. Although I admit that the fact they won two of those on the road weighs pretty heavily in their favor. I guess I just couldn't see them giving points at home against the Colts, so there you go.

6) Atlanta - They beat Philly at home, and won on the road against Buffalo, with their only loss being a letdown game on the road against a team which usually wins at home, and it was still a close game. Yes, they're better than I thought they were. A lot better.

7)Tampa Bay - A team which is the worst 3-0 team in the league is still 3-0.

8) Jacksonville - A team which is 1-1 on the road, playing really good defense. A dangerous team.

9) Kansas City - Over Denver because Denver did, in fact, play at home. Don't discount the teams which have not won on the road (Denver, Seattle) and think they're a lot better than 2-1 teams with a road win (Atlanta, Jacks, KC).

10) Denver - Yes, they lost to Miami at Miami, and no, they haven't won on the road. By the time you get to #10, it's a lot tougher to pick who's good and who's bad.

There it is! Worth every penny, I'm sure you will be talking around the cooler tomorrow about these daring and amazing rankings! At least, as much as you'll be talking about Dr. Z's rankings, anyway.

2
by andrew (not verified) :: Thu, 09/29/2005 - 6:53pm

Did you just make excuses for each of your top ten?

3
by Pat (not verified) :: Thu, 09/29/2005 - 6:56pm

One t in Cincinnati!

4
by David (not verified) :: Thu, 09/29/2005 - 7:29pm

If McNabb gets healthy, no one beats them, period. I believe he’s going to be healthy, so they’re on top.

Not going to happen. McNabb's injury is a sports hernia, which does not heal without surgery. Surgery would sidline McNabb for two months. So either he plays hurt or he doesn't play at all, and neither situation looks good for the Iggles.

5
by JasonK (not verified) :: Thu, 09/29/2005 - 7:49pm

"The Giants always have been a team of whiners, going back to the Jim Fassel days when Michael Strahan & Co. were popping off every 20 minutes about some indignity or another."

All the way to the "Jim Fassel days"? For a guy who regularly goes on tirades about the injustice of some OG from the '50s you've never heard of not being in the Hall of Fame, he has a pretty limited concept of "always."

Anyway, as Aaron intimated, "Shockey Makes Stupid Remark" isn't exactly column-leading material these days.

6
by JasonK (not verified) :: Thu, 09/29/2005 - 7:50pm

"The Giants always have been a team of whiners, going back to the Jim Fassel days when Michael Strahan & Co. were popping off every 20 minutes about some indignity or another."

All the way to the "Jim Fassel days"? For a guy who regularly goes on tirades about the injustice of some OG from the '50s you've never heard of not being in the Hall of Fame, he has a pretty limited concept of "always."

Anyway, as Aaron intimated, "Shockey Makes Stupid Remark" isn't exactly column-leading material these days.

7
by jbf (not verified) :: Thu, 09/29/2005 - 8:10pm

While I follow Dr. Z's logic in detailing how the Patriots beat the Steelers, the "fatigue factor" is not the reason, but a part of the reason.

After all, he surely cannot apply fatigue to those -- what, two? -- drives earlier in the game when the Patriots turned over the ball as they drove toward the Steelers' endzone.

8
by BryanS (not verified) :: Thu, 09/29/2005 - 8:35pm

We all know that New England is the best team in the NFL. I don't care how they played against Carolina, they went into Pittsburgh and beat a good steelers team without matt light, rodney harrison, and a depleted secondary.

9
by BryanS (not verified) :: Thu, 09/29/2005 - 8:36pm

We all know that New England is the best team in the NFL. I don't care how they played against Carolina, they went into Pittsburgh and beat a good steelers team without matt light, rodney harrison, and a depleted secondary.

10
by B (not verified) :: Thu, 09/29/2005 - 8:42pm

If New England is better than Pittsburgh, and Pittsburgh's only loss is to New England, shouldn't Pittsburgh be rated above New England, cause both teams have one loss, and Pittsburgh lost to a much stronger opponent?

11
by Matthew Furtek (not verified) :: Thu, 09/29/2005 - 9:29pm

Re: Redskins
I think Redskins' fans are notorious for e-mailing all the football writers and bashing them for ranking the 'Skins so low.... hence Dr. Z giving into the pressure. He mentions in his article that he didn't even watch the Redskins play... and I think some fans are upset because these writers talk poorly about our team, yet haven't observed them. For instance, Peter King didn't even attend Redskin's training camp but based all of his observations on a scrimmage the first week of August against the Ravens.

I think our fans are very media sensitive especially since Snyder has become owner, almost turning them into wild Internet fan boys. Part of this is because it seems like Snyder gets bashed a lot in the media for being hands on, when Jerry Jones can't do anything wrong, yet their teams performances over the past 6 years has been equally mediocre if not bad. Part of this is because of resentment over Art Monk not being in the Hall of Fame, especially since Dr. Z and Peter King both are vocal objectors. Part of it is resentment because the Redskins as a dynasty is overshaddowed by the 49ers and Cowboys, even though they went to 4 Super Bowls in 10 years, from 1982-1992, and won 3 of them. I think that's a similar track record to the 49ers over the same period of time? Part of it is frustration because our once great franchise only gone to the playoffs once since Gibbs left.

Anyways, now that FO has a column in FoxSports I'm sure that they will begin to see this phenomenon first hand. To be honest, I'll take the "dull" 2-0 Redskins over the "exciting" 0-3 Raiders at any time. I hope we can continue our march to making all the other fans fall asleep this weekend vs. the SeaHawks.

12
by N (not verified) :: Thu, 09/29/2005 - 9:31pm

Wait, so you're saying that the worse team won?

The transitive property should remain in the domain of mathematical functions and not ranking better football teams.

Is it better to be objective and wrong or subjective and wrong?

13
by CaffeineMan (not verified) :: Thu, 09/29/2005 - 11:19pm

Re: Skins and Media - a couple comments from someone not close to the situation but kindly disposed towards the Redskins (from being a kid in DC in '72) and with no particular love for the Cowboys but who watched them for 3 years when I lived in Texas.

First, I thought Dr. Z got treated badly somehow by the Snyder regime. I think he's alluded to it a couple of times. So there may be that.

Next, Matthew, I think you're focussing on the wrong thing. It's not the fact that Snyder is hands on, it's his relationship with the media. Snyder has the same problem with the media that Belichick had in Cleveland: They both think the media is useless and blow them off. Belichick had (maybe still has, that's a separate debate) the arrogance of the successful football coach and Snyder has the arrogance of the successful internet businessman. Both are accustomed to being accountable to only a handful of people (Belichick to Parcells and Snyder to his board). The thing is, Belichick learned that the media are part of the job and they will FRY you if you try to just blow them off (and sometimes even when you don't). Snyder's never learned that. Jerry Jones gets a pass because he just LOVES the media attention to death so he gives them access. It doesn't matter what you do to your team, if you treat the press like your buddy and confidant, they'll eat it up. Access is what gets the press on your side.

14
by Countertorque (not verified) :: Fri, 09/30/2005 - 12:32am

RE: 7

I would think that he's attributing the loss of those - yes 2 - turnovers deep in Steeler territory to an unfatigued Steeler defense making some plays (forced fumble and tipped pass plus INT) and keeping the Pats off the scoreboard.

15
by Countertorque (not verified) :: Fri, 09/30/2005 - 12:33am

RE: 7

I would think that he's attributing the loss of those - yes 2 - turnovers deep in Steeler territory to an unfatigued Steeler defense making some plays (forced fumble and tipped pass plus INT) and keeping the Pats off the scoreboard.

16
by Matthew Furtek (not verified) :: Fri, 09/30/2005 - 1:07am

Re: Dr. Z and Snyder
I think Dr. Z just doesn't like him, in fact a lot of people don't like him because of the way he runs his organization.

www.basketballdraft.com/football/nfl/news/1999/12/09/power_rankings/
"I must have seen 100 closeups of Jerry Richardson in the owner's box during a game and I've yet to see his expression change. That's what comes from being an ex-NFL player. Are you listening, Dan Snyder?"

http://sportsillustrated.cnn.com/inside_game/dr_z/news/2000/12/07/drz_in...
"But Terry Robiskie was the passing-game coordinator, Dan Snyder's man, and he might be obeying instructions from the boss. Which, of course, would be to have George flinging the ball like crazy, just like Snyder saw in the highlight films. This is all conjecture, you understand, which all us experts removed from the action love to do."

http://sportsillustrated.cnn.com/inside_game/news/2001/09/28/drz_mailbag...
"Yes, I agree with you. Schottenheimer and George were a bad mix from the start, philosophically, strategically, every way. Don't forget, though, that George was Dan Snyder's man, not Marty's. Marty wouldn't touch a guy like that, given his druthers."

"in the past I've been accused many times of hating the Skins because I've repeatedly taken shots at their owner."

17
by BK (not verified) :: Fri, 09/30/2005 - 2:55am

RE: Boston College coaching

I'm glad someone has written about the lack of game planning and overall strategy regarding BC coach Tom O'Brien. Game in, game out, the worst majoe college coach I have ever seen - although I hear he does recruit well.

18
by Bruce Dickinson (not verified) :: Fri, 09/30/2005 - 8:54am

Re the college comments.

Somebody mentioned that USC looked so good that they might be able to score on a bad NFL team. If they played the Texans, the texans would probably win 38-10. so yeah, they might score, but i never understand how people think the college game is so great. each NFL team is made of the best 2% of all college players. and at an average age (25-27) which is much more their physical prime.
sometimes i think i'm the only hardcore football fan that doesn't care at all about the college game.

19
by jbf (not verified) :: Fri, 09/30/2005 - 10:07am

RE: 14 (and 15)

I'd like to think so; it's just that, as I read, I got the impression that he was citing fatigue as the lone factor in the outcome.

20
by PatsFan (not verified) :: Fri, 09/30/2005 - 10:32am

Re: #18

You're not the only one. I've never been able to get into college football, either.

Re: #13

That indirectly touches on something that's been going on lately up here in Boston. The Globe has started whining about alleged lack of access to the players. The beat writer (Jerome Sullivan) manages to work in a complaint about it in every other column, it seems. The sports page editor has whined about it. Then there's Ron Borges. And the Globe (owned by the NYT which owns a piece of the Red Sox, BTW) has had an atrociously small amount of Pats coverage so far this year. Yet the other Boston paper, plus other regional papers (Providence, Hartford, elsewhere in MA, etc.) have better coverage of the Pats and no whining. Even more bizarrely, the Globe's best Pats coverage, Mike Reiss's blog never sees print (and in his mailbag he apparently has enough "access" to ask players some of the questions readers send it).

21
by Rick "32_Footsteps" Healey (not verified) :: Fri, 09/30/2005 - 10:35am

Does Jerry Jones get a free pass? I've seen plenty of press who mock him.

Of course, I'm an Eagles fan, so perhaps my preferred sources are more than a little biased. But you'd think that DC fans would be able to find similarly biased ones.

22
by Johnnyel (not verified) :: Fri, 09/30/2005 - 11:11am

Click my name for an article about Snyder and the media. In short, soon, Dan Snyder will be the Redskins media.

23
by Matthew Furtek (not verified) :: Fri, 09/30/2005 - 11:18am

My biggest complaint as far as media and Redskins goes is that Art Monk isn't in the Hall of Fame. The teams during the 80's era up to the last Super Bowl had a lot of underappreciated talent. When Art Monk retired he held all the major WR records. That should be merit enough to put him in.

As a fan, I don't care much what the media says, I'm sure Snyder doesn't either. Snyder has been doing an excellent job of making them irrelevant. We can watch press conferences on the Redskins websites and interviews with players. Snyder has "partnered" with the largest fan messageboard in order to co-ordinate chats with players and coaches. The founders of that messageboard have been given press passes to use during games and watch from the press box. The post game comments from Gibbs and locker room interviews are all online on the Redskins website. I think of the changes as similar to more people getting their news from blogs and the Internet. I know that probably explains some of the animosity from the media. Snyder was upset at the Washington Post because their tickets were scalped allowing away team fans to enter our Stadium. He has taken steps to ensure the Stadium will be full of Redskins fans from now on. Yes he makes money and does well from the business side, but he also runs his franchise like one of the fans. He's not afraid to spend money, and wants fans to enjoy themselves at the Stadium.

Post #13 gives some background information as far as why our fans are media sensitive and #16 was to show that I couldn't find any backstory on if Snyder did anything to Dr. Z. It appears Dr. Z doesn't respect the way he runs the franchise... bringing in his own men and being too hands on.

24
by CaffeineMan (not verified) :: Fri, 09/30/2005 - 11:37am

RE: Globe access to Pats

Gee, you don't think it would have anything to do with the fact that their feature football columnist has an agenda against the head coach, do you? I'm in favor of tough, accurate, fair reporting, but Wrong Borges is so clearly grinding an axe that it's ridiculous.

25
by MJK (not verified) :: Fri, 09/30/2005 - 11:54am

"If New England is better than Pittsburgh, and Pittsburgh’s only loss is to New England, shouldn’t Pittsburgh be rated above New England, cause both teams have one loss, and Pittsburgh lost to a much stronger opponent? "

You can really go in circles trying to figure out these things just like this. For example, one could counter that if Team A beats Team B, shouldn't Team A be rated higher regardless? But what if A beats B, B beats C, and C beats A? Then what?

For answering questions like this, I'm a big fan of using a technique called "Maximum Liklihood Estimation"--math and probability people have probably heard of it. Basically, you make some assumption about the probability of Team A beating Team B if Team A is X good and Team B is Y good. Then, based on the outcomes of all games to date, you figure out what X and Y values apply to each team that maximizes the liklihood of what actually happened. It builds in strength to schedule automatically. I'm actually working on a system to rank NFL teams using MLE, just to see what it says. (I'm working on getting a description of the method up, outlining all the underlying assumptions, as well as the current ratings. If anyone is interested, I'll let you all know when its done...) Of course, there haven't really been enough games yet to get reliable ratings (for example, based on current W/L, Miami looks like the current strongest team in the AFC East!), but I'm curious to see how this method will work compared to DVOA or subjective ratings. Incidentally, a very similar system is (presumably) used to rank NCAA football teams, although the exact algorithms they employ are not made public.

And for the record, according to my current calculations NE is supposedly better than Pittsburgh but not by much, and both teams are rated only slightly above average. Hmm....

26
by MJK (not verified) :: Fri, 09/30/2005 - 11:54am

"If New England is better than Pittsburgh, and Pittsburgh’s only loss is to New England, shouldn’t Pittsburgh be rated above New England, cause both teams have one loss, and Pittsburgh lost to a much stronger opponent? "

You can really go in circles trying to figure out these things just like this. For example, one could counter that if Team A beats Team B, shouldn't Team A be rated higher regardless? But what if A beats B, B beats C, and C beats A? Then what?

For answering questions like this, I'm a big fan of using a technique called "Maximum Liklihood Estimation"--math and probability people have probably heard of it. Basically, you make some assumption about the probability of Team A beating Team B if Team A is X good and Team B is Y good. Then, based on the outcomes of all games to date, you figure out what X and Y values apply to each team that maximizes the liklihood of what actually happened. It builds in strength to schedule automatically. I'm actually working on a system to rank NFL teams using MLE, just to see what it says. (I'm working on getting a description of the method up, outlining all the underlying assumptions, as well as the current ratings. If anyone is interested, I'll let you all know when its done...) Of course, there haven't really been enough games yet to get reliable ratings (for example, based on current W/L, Miami looks like the current strongest team in the AFC East!), but I'm curious to see how this method will work compared to DVOA or subjective ratings. Incidentally, a very similar system is (presumably) used to rank NCAA football teams, although the exact algorithms they employ are not made public.

And for the record, according to my current calculations NE is supposedly better than Pittsburgh but not by much, and both teams are rated only slightly above average. Hmm....

27
by PatsFan (not verified) :: Fri, 09/30/2005 - 12:00pm

Re: #23

I wonder if something similar is behind the Globe's vendetta. Kraft has been working hard to bypass the media by talking directly to the fans by doing stuff like coming up with in-house productions like Patriots All Access, Totally Patriots (both TV shows), Patriots Football Weekly, etc. And Belichick, Brady, Seymour, Harrison, and Brown (and maybe a couple of others) have (presumably paid) weekly appearances on the big sports talk station in town.

Re: #24

Yeah -- does Borges think there's actually anyone out there who is too dumb to see through his totally obvious act? Cafardo's another one.

28
by Pat (not verified) :: Fri, 09/30/2005 - 1:10pm

MJK:

There are actually better ways of doing this than maximum likelihood, but for some reason, advanced statistics never seem to get outside of scientific fields.

One of the problems you have with maximum likelihood is that if you put in a huge outlier, the result will skew. It won't skew much, depending on how many other data points are there, but it will skew. So New England last year would drop because of the loss to Miami. (In statistics, that ranking would be called a "biased estimator").

You can deal with this problem using an unbiased estimator method, although it takes a fair amount more effort to do it.

29
by Alan P (not verified) :: Fri, 09/30/2005 - 1:16pm

#13 - I think you have the cause and effect wrong. The media is brutal to Snyder but not to Belichick not because of how they treat the media or how hands on they are, but because Snyder has made awful decisions and Belichick has made great decisions.

When the Raiders won Al Davis was colorful and brilliant, now that they are losing he is looked at as old and pathetic.

Winning can make anyone into a media darling, losing can make the nicest guy into a pathetic jerk.

30
by MJK (not verified) :: Fri, 09/30/2005 - 1:23pm

Thanks Pat. I'll try reading up on those...

31
by MJK (not verified) :: Fri, 09/30/2005 - 1:43pm

Pat,

After thinking about it, I have this question. Don't we WANT the results to skew if there's an outlier? To use your example, isn't the goal to figure out how well New England played over the entire course of last year, not to figure out how well they played except for the one time where they lost really badly? It seems to me that if a team has a whole bunch of impressive victories and one embarassing defeat, it should be ranked lower than a team that just had a whole bunch of impressive victories. Similarly, a bad team that lost horribly all the time but managed to beat one really good opponent ought to get a lot of credit for playing really really well for 1/16 of the year.

Or is your point that MLE OVERESTIMATES the effects of one outlier? The page you pointed me to didn't have a whole lot to say about what an unbiased estimator method would be; it just defined "biased estimator".

32
by Pat (not verified) :: Fri, 09/30/2005 - 3:27pm

Yah: the entire point of an unbiased estimator method is that you need to find an unbiased estimator (or a minimum-bias estimator). How you do that depends on the situation at hand, and I couldn't find a good book on that. It's in statistics texts, though.

Don’t we WANT the results to skew if there’s an outlier?

No - because you're in the small-numbers limit. If NE/MIA play that game a hundred times, Miami would probably win once. Suppose I've got a distribution with long tails, but a well-defined central peak. Suppose I then sample 16 numbers from that distribution. Chances are I'm going to get elements only inside that peak, but I could get one from the tails. If you don't get one from the tails, then the maximum-likelihood analysis will tell you "the most likely value is here". If you get one from the tails, that value will move, even though it's the same distribution, with the same central value. See here, down at the "bias" section, for a good real-world example.

The problem is that with a small amount of information, you can't infer what the probability distribution actually is via maximum likelihood. You're far more likely to be biased by a small piece of information (like the Dolphins beating New England). A better method in this case would be to attempt to construct an a priori probability distribution for a team's performance, and then fit that to the data via chi-squared minimization. In this case, something with long tails, which would tell you "that loss is probably in the tail of the distribution", wouldn't bias the result.

Note that the bias is still there - it's now moved into the estimate of the variance of the team - but now it doesn't modify the median value. To explain in football terms, if the Patriots hadn't lost to the Dolphins, maybe you'd rate them at 90%, with a spread of 10% (out of 100%). Now, using an unbiased estimator method, if the Patriots had lost to the Dolphins, you'd still say they're at 90% (which is good - that loss doesn't bias the result) but the spread might be 20%. Now, that 1 game has biased something, but it's biased your estimate of the error, and not the estimate of the central value.

33
by CaffeineMan (not verified) :: Fri, 09/30/2005 - 4:52pm

Alan P., actually, I think I have cause and effect just exactly right. The only thing that makes it harder to see with the Pats as an example is that the national press (for the most part) cuts Belichick slack (because of his success) and it's the local press that have the agenda. As far as I can tell, Snyder's managed to alienate both local and national media so that the only ones that cut him slack are Skins fans. Al Davis, good or bad, has always been a good story, which the press likes. Belichick would love to avoid being a story, as would Snyder, I think. You can argue about his methods, but Snyder wants to win and would rather not have to explain anything to anybody on his way to doing it.

34
by doktarr (not verified) :: Fri, 09/30/2005 - 5:15pm

I think Dr. Z gets his advice wrong on the last section, about the FSU-BC game.

Situation - down 11 late, first down inside the 5.

What really happened: they ran a bunch of times, got stuffed, turned it over on downs, game over.

What Dr. Z suggests: First, he says that they can't run on FSU in short yardage. Then he says they should kick the FG on 4th down, onside kick, and try to score again to tie it.

My suggestion: Maybe he's right about the short yardage, but in that case there's many other options that don't involve a power run. You're inside the 5, so there's no reason to think your next opportunity to score (after recovering the onside) is going to be easier than this one.

But more importantly, kicking the FG is a strategic error. Aside from what I just mentioned (this chance is the best you will get), there's the fact that you need to go for 2 on your TD to force OT. If you score the TD and then miss the 2-point try, you're down 5, and you know you need a TD to win after the onside kick. If you make the 2-point conversion, then you know you can kick and try your chances in OT.

On the other hand, if you kick the FG first, it may end up being meaningless if you score later but miss the 2-point conversion. So while the BC coach may have made the wrong PLAY call, he made the right call to try to get the TD first.

35
by Kaveman (not verified) :: Sat, 10/01/2005 - 2:13am

#33:

> You can argue about his methods, but Snyder wants
> to win and would rather not have to explain
> anything to anybody on his way to doing it.

And if he was winning, this would probably be just fine. But... uh...

36
by Flux (not verified) :: Mon, 10/03/2005 - 9:23am

The article #22 linked to was probably the saddest thing I have ever read. If someone emailed it to me, I would assume it came from The Onion.

Redskins' radio broadcasts can't really be that bad, can they? And if they are, aren't their ratings plumetting? Why does anyone listen to football on the radio anyway, what with the whole "seeing it with your own eyes on TV for free" option?

Then again, I've pretty well given up on live football in any form (I tape them and watch them later) since I've got too much else to do and the amount of dead time during sports telecasts is just painful. So perhaps I'm not the best control group for this question.