Writers of Pro Football Prospectus 2008

22 Jun 2005

Ratings Game

My god, we are not alone! One of the frustrations we've had since starting Football Outsiders comes from the huge holes in the play-by-play data, all the information that isn't available: length on incomplete passes, yards after catch on every play, which defensive backs are covering which receivers, and so forth. We've tossed around the idea of a project to log games off videotape to track these statistics ourselves.

Well, guess what, I found out about a month ago that somebody is already doing this. Today, Dr. Z has a glowing review of a new self-published book called Scientific Football 2005 by a fan named K.C. Joyner, who has logged nearly every game from last season and developed all kinds of new metrics that measure how often defensive backs get burned, how often quarterbacks throw long (not just complete passes long), and so forth.

Yesterday I had a nice phone call with Joyner, chatting about how we each got into this and how our methods compare. His work and our work are in much the same spirit of objective analysis combined with pure love of football, but we come at things from very different (and complementary) angles. The main thrust of my analysis is "How can we filter out the biases inherent in conventional football statistics?" The main thrust of his work is "How can we measure the actions on the field that aren't measured by conventional football statistics?" We come to many of the same conclusions: That Eli Manning is overhyped, that Kurt Warner is not the answer in Arizona, that Jeff Garcia might be the answer in Detroit.

Whether we ever work together or simply pursue analysis on two separate tracks, the more the merrier. The popularity and quality of NFL analysis can only be helped by more people who do interesting work. But the Workman folks want to remind you to make sure that if you order his book, you order Pro Football Prospectus 2005 too!

Posted by: Aaron Schatz on 22 Jun 2005

64 comments, Last at 26 Jun 2005, 8:47pm by Ryan Mc

Comments

1
by Pat on the Back (not verified) :: Wed, 06/22/2005 - 5:35pm

Cool! Very exciting, now I have even more corroborating evidence to prove I am not the only looney who follows this stuff.

2
by harry (not verified) :: Wed, 06/22/2005 - 5:37pm

I'll just get it over with. What's Joyner's answer - Brady or Manning?

3
by ABW (not verified) :: Wed, 06/22/2005 - 6:00pm

The players that get mentioned specifically in the article:

Ashley Lelie - He like this guy a lot, calling him "possibly the most dangerous deep threat in the league", which is maybe going a little far, but he did have 25.4 DPAR, good for 16th in the league, and a DVOA of 22.7%

Jermain Wiggins - another player that both statistical systems like, he was 7th in DPAR with 15.4 and 14.5% DVOA.

He then rips on Manning and Warner, neither of whom did particularly well according to the stats on this site(although Warner wasn't that bad).

It would be really interesting to see if this trend of agreement between the two systems holds - it would be very convincing if two completely separate statistical analyses liked(or didn't like) the same player. It would probably be pretty illuminating which players they disagreed on as well, perhaps revealing strengths and weaknesses of the two approaches. Too bad my budget for books about statistical analysis of football is already used up for the year...

4
by Richie (not verified) :: Wed, 06/22/2005 - 7:29pm

I wouldn't mind quitting my job to analyze football games. But, I would need it to pay. I am for hire on this subject, and I work relatively cheap.

5
by Pat (not verified) :: Wed, 06/22/2005 - 7:57pm

“How can we filter out the biases inherent in conventional football statistics?�

One of the statistics which would be exceptionally useful to actually come up with is "corrected completion percentage". It's been noted elsewhere (on this site, in fact) that completion percentage is a horribly biased statistic, because short passes have a high completion percentage, and long passes have a low completion percentage, so what passes you attempt drive the final percentage.

It'd be great to have a way to fix that bias, much like "Running Back Success Rate" fixes the issue that short yardage can sometimes be much harder than long yardage on runs, but of course, you need yard of completion/incompletion for that.

6
by Vince (not verified) :: Wed, 06/22/2005 - 8:28pm

I just think it's cool that Aaron is plugging and promoting a product that is creating competition for him. I for one will be sure to buy both.

7
by Will Allen (not verified) :: Wed, 06/22/2005 - 10:03pm

Unless he gets injured (a distinct possibility), Garcia will likely have a good year. I'm assuming, of course, that he will be starting no later than the third game.

8
by Sean (not verified) :: Wed, 06/22/2005 - 10:04pm

I'll probably pony up the fifty bucks as well, as it sounds like a good read. (Needless to say, I've already got the PFP on pre-order.)

Of course, I stand by my belief that Eli Manning is going to be very good, but it would be tough to do any sort of analysis that would come up with anything but a failing grade for the way he played last year.

9
by VarlosZ (not verified) :: Wed, 06/22/2005 - 11:19pm

This seems like it's right up my alley, and if I were a little less poor, I'd probably buy it. However, I'm skeptical of the ability to derive meaningful, objective statistics for football players at most positions.

Mostly, this is due to sample size. From the link, for example, Sheldon Brown did very well against 117 pass attempts last year, which is a small sample size. He did especially well on 24 deep passes, allowing only 2 completions, but that's a tiny sample size. Each year in baseball, how many great hitters get only a couple of hits in the first five games, and how many utility infielders are batting .450 after their first 24 ABs? In baseball, such results are considered meaningless, but in football it's easy to overestimate their significance because there's no greater sample size on the way.

And, of course, there's always the problem of football's interdependency. There are so many variables on each play that it must frequently be impossible to know what the statistics really mean. In Brown's case, it could be that the peculiararities of Philly's secondary scheme, the collection of QBs and WRs faced by Brown, the weather, and so forth conspired to stack the deck in his favor. Or, he could have had a statistically "lucky" year, having what would be, in baseball terms, an outstanding month. Or maybe he's just really good. Most likely, it's some inscrutable combination of all three.

10
by Reinhard (not verified) :: Wed, 06/22/2005 - 11:53pm

Which of these quarterbacks threw the most touchdown passes?
- Carson Palmer
- Joey Harrington
- Byron Leftwich
- Michael Vick
- Chad Pennington
- Ben Roethlisberger

Whcih quarterback of the group threw the least interceptions?
- Trent Green
- Kerry Collins
- Tom Brady
- Joey Harrington
- Brett Favre
- Marc Bulger

Joey Harrington! Not that I've ever seen him play, but it doesnt sound unreasonable to wait for the season before saying his 2006 campaign is doomed...

the Q's from Kirwan's quiz, click my name

11
by Jim A (not verified) :: Thu, 06/23/2005 - 12:52am

This sounds a lot like what Ron Antinoja is doing for baseball with his Tendu company. Tendu records every MLB game off the satellite feed and its reviewers attempt to quantify events from the video that are often subjective, such as pitch types and locations and batted ball trajectories. Tendu sells a software product to display and analyze the data they collect. They also employ a team of former pro and college players and utilize various techniques for checking to promote consistency and reduce biases among individual reviewers.

I hope that if this football book is successful, the author will continue to strive toward improving quality and minimizing the subjectivity and capriciousness inherent in judging and collecting this type of data. I can't even begin to fathom how bleary-eyed and sloppy I'd be after reviewing 48 hours worth of football tape each week.

But Aaron is right, this type of research could be very complementary to analysis of more conventional and objective football stats.

12
by Glenn (not verified) :: Thu, 06/23/2005 - 12:57am

I agree that it's great for Aaron to be alerting FO readers to other product, but in marketing terms, it's not really "creating competition" for him. One product feeds interest in the other, and there's certainly room for sampling both. Rather than taking Aaron's "piece of the pie", the Joyner project is helping to create a bigger overall pie, expanding the universe of people who can potentially get hooked on using objective football analysis. More interest in such analysis by more fans creates more interest in reading Football Outsiders.

13
by JasonK (not verified) :: Thu, 06/23/2005 - 9:44am

I'm wondering how he has overcome the difficulties of the truncated view that the TV broadcast gives you. I gather from the Z story that he's watching the same network broadcasts as you or I (i.e., he doesn't have the coaches' tape), and, as Z has commented before, it can be tough to analyze the passing game if you can't see how the safeties are lining up, etc.

14
by Sophandros (not verified) :: Thu, 06/23/2005 - 9:55am

Re: #3

The Lelie comment might not be going to far, because DVOA and DPAR take into account EVERY play, not just the deep passes. I think that Joyner was looking ONLY at the long ball in his analysis of Lelie.

15
by MDS (not verified) :: Thu, 06/23/2005 - 11:24am

"Sheldon Brown did very well against 117 pass attempts last year, which is a small sample size."

I don't think that's too small a sample size to get a good understanding of how well Brown played. I mean, wouldn't you say that five games into the season, you have a pretty good idea of which quarterbacks are playing well? That's about how many games it takes a QB to throw 117 passes.

16
by Erik (not verified) :: Thu, 06/23/2005 - 12:26pm

Is there any chance of getting Joyner to write an article for FO describing his approach in more detail?

17
by ABW (not verified) :: Thu, 06/23/2005 - 1:23pm

Re: 10

OK, this is a little off-topic, but that's a pretty misleading list. First Kirwan compares Harrington to a list of guys who are either in run-first offenses(Roeth.., Vick, Pennington to an extent) or guys who didn't play a full 16 game schedule(Leftwich, Pennnigton, Palmer) and all of them only had a couple fewer TDs. Then he compares INTs to a group of passers who threw a lot of deep balls this year - except for Collins, who just sucked, all those guys had YPA a yard or more larger than Harrington.

It's true Harrington didn't kill his team last year(6.5 DPAR) like he did the year before(-24.6 DPAR...ouch). There is some reason to think that he might do better this year. But I wouldn't base that hope on Kirwan's list of cherry-picked examples.

18
by MDS (not verified) :: Thu, 06/23/2005 - 2:00pm

Obviously, ABW, you're right. A fair list would be:
Carson Palmer
Vinny Testaverde
David Carr
A.J. Feeley

Those were the four quarterbacks who had BOTH more interceptions AND fewer touchdowns than Harrington. I will now bring up a subject I've brought up before and gotten very little feedback: Why does everyone say Carr is better than Harrington? I really don't understand. I don't think either one of them has lived up to the draft position, but it seems like everyone in the world thinks Carr is superior and I can't figure out why.

19
by MCS (not verified) :: Thu, 06/23/2005 - 2:17pm

Gotta be the hair!

20
by VarlosZ (not verified) :: Thu, 06/23/2005 - 2:38pm

"I don’t think that’s too small a sample size to get a good understanding of how well Brown played. I mean, wouldn’t you say that five games into the season, you have a pretty good idea of which quarterbacks are playing well? That’s about how many games it takes a QB to throw 117 passes."

Well, yes and no. I implied that Brown's results on the 24 deep passes might be meaningless, not that the stats for the 117 total passes were useless.

That said, I don't put too much stock the 117 number. You made the analogy to a QB's first five games. Let's say that Brown's performance last year is rougly equivalent to, hypothetically, Trent Green having a QB Rating of 120 a third of the way through the season. Setting aside the fact that a QB's statistics are largely a reflection of the performance of the passing offense in general, wouldn't you expect Green to regress to the mean over the next 11 games? In other words, no matter how impressive the stats, you'd discount the performance somewhat because of the sample size.

The same goes for Brown. I think the safest assumption is that he's a good player who had a phenomenal year, rather than a phenomenal player who had (for him) an ordinary year.

21
by ABW (not verified) :: Thu, 06/23/2005 - 3:10pm

Well, for what it's worth, Carr's been worth about 20 points a year more than Harrington for the past 2 years according to DPAR. That's something. Also, he's played behind a worse O-line, with a team that hasn't made a huge deal out of the receivers it's drafted for him to throw to, like the Lions have for Harrington. He also had a somewhat better year than Harrington I think even throwing for 2 more INTs and 3 less TDs by completing a higher percentage of passes for 500 more yards and a YPA that was a yard larger than Harrington's, so he's probably a little more exciting to watch, which isn't going to hurt.

Looking at the stats on this site, Houston's offense had a 10.1% offensive unadjusted passing DVOA - adjusted it was only 1.4%. So Carr got to pass against worse defenses, which probably helped the perception of him. Detroit's adjusted passing DVOA was -7.7%, so in addition to passing against pretty bad defenses, Carr's offense was in fact better than Harrington's, and regardless of how correct it is, QBs get judged by the offense they lead.

So while Carr probably isn't worthy of the adoration that people heap on him because he hasn't really done that well, I think he has done at least a little better than Joey Harrington, who I think is headed for Bustville. I agree that neither of them have lived up to their draft expectations, but it does seem legitimate to think that Carr is in fact a (slightly) better QB than Harrington.

So in conclusion, MDS, it probably is the hair that makes people think he's some kind of star.

22
by ABW (not verified) :: Thu, 06/23/2005 - 3:33pm

Hmm, that last line didn't come off quite as I wanted it to. I put it in because reading over what I'd read it didn't seem like the argument for Carr actually is that good, not to try and be demeaning.

23
by Reinhard (not verified) :: Thu, 06/23/2005 - 3:53pm

Oh yeah, its a cherrypicked list. Maybe a gerrymandered list. But although I wouldn't argue that Harrington is better, it is a valid point that he threw more TDs less picks that Carr and Palmer, the young QB studs. All three of them have a young go-to reciever, which helps equalize things too.

24
by Pat (not verified) :: Thu, 06/23/2005 - 4:24pm

The same goes for Brown. I think the safest assumption is that he’s a good player who had a phenomenal year, rather than a phenomenal player who had (for him) an ordinary year.

That being said, however, last year was Brown's only starting year. So the only thing you can say about him from actual data is that he's a phenomenal player, based on one year's data. It may be that he is a good player who had a phenomenal year, but he needs to have an ordinary year for you to know that.

More succinctly put, the best thing you can say right now is that Brown is a phenomenal player, with big error bars on that "phenomenal."

25
by Richie (not verified) :: Thu, 06/23/2005 - 4:32pm

I'm not sure the inherently small sample-size in NFL statistics is relevant.

While evaluating a CB, if you think the number of plays to evaluate him is small, yet CB's always have a small number of plays each season, don't we want to evaluate said CB by how well he performs within that limited sample?

If we have one CB who has a horrible year based on a small sample, then a great year, then a horrible year, can't we conclude that over the course of a whole season, this CB is going to be unpredictable?

If we have a CB who does very well in 3 straight seasons, I think we can assume that he gets his game in order early, and plays at a high level the whole year.

We want to evaluate how well a player plays over a short, 16-game season. So, wouldn't we care less about how well a guy would play over a 200-game season?

26
by MDS (not verified) :: Thu, 06/23/2005 - 4:33pm

I'll grant you that the Lions have had a better o-line than the Texans, but for the most part Carr gets sacked more because he holds the ball more. The offensive line gets too much credit/blame and the quarterback gets not enough on sacks.

As for receivers, it's not even close. Is anyone going to suggest that a trio of Johnson, Gaffney, and Bradford isn't vastly superior to what the Lions have? Especially when you consider that those three guys have missed exactly one game combined in Carr's three years. Harrington can never get comfortable with a set of receivers because they're always getting hurt.

I'm no Harrington apologist. As a Lions fan, I was very glad they signed Garcia. But it just seems like Carr gets a free pass even though he hasn't produced.

27
by mistamaxwell (not verified) :: Thu, 06/23/2005 - 4:49pm

This is a bit off subject, but how about Running Back Slugging Percentage? Maybe count only yardage from successful runs and divide it by number of successful runs. It might help differentiate which runners are picking up short-yardage runs and which runners are setting up second and shorts. Maybe someone is all ready charting this?

28
by Trogdor (not verified) :: Thu, 06/23/2005 - 5:21pm

"I’ll grant you that the Lions have had a better o-line than the Texans, but for the most part Carr gets sacked more because he holds the ball more."

How much is that saying really? Doesn't Harrington make Georgia Tech's QB on 4th down look like Rob "No Limit Hold 'Em" Johnson?

For those who don't follow college, in a game last year GT's QB, playing for renowned guru Chan Gailey no less, dropped back on a 4th down pass against Georgia, quickly scanned the field, and threw the ball out of bounds before even finishing his drop. Harrington has been known to throw it away faster than some QBs could spike it. I think that's a pretty decent nickname for Johnson as well. Sorry about overexplaining.

One thing about the CB ratings - it looks like he has ratings for plays when the ball is thrown their direction, but what about when they're on the field and the ball doesn't go their way? It should be pretty hard to tell what kind of coverage they had based on TV angles. I'm sure he has numbers for % of plays where that CB is targeted (and if not, from what his book apparently has it should be easy to compute), but that doesn't necessarily tell the story. A low % of targets could mean you have Deion '94 over there, or it could mean he's a mediocre CB but the guy on the other side is Mike Rumph as a rookie, or is also mediocre but covering Owens. Anyway, just throwing some stuff out, and wishing we could all get coaches' film for this kind of stuff.

29
by MDS (not verified) :: Thu, 06/23/2005 - 5:42pm

That Georgia Tech QB was Reggie Ball. I found this with a quick search:

"He thought the clock said third down," Yellow Jackets coach Chan Gailey said. "He got discombobulated. I know he would not have thrown it away on fourth."

Really? He wouldn't have intentionally thrown the ball away if he knew doing that would end the game? Sounds like you've taught him a lot, Chan.

30
by B (not verified) :: Thu, 06/23/2005 - 5:53pm

I've always thought any extra slack Carr receives over Harrington comes from the fact that Carr is playing for an expansion team. Of course, compared to the Lions, that may be an advantage.
In Reggie Ball's defense, maybe he thought GT deserved to have 5 downs, too.

31
by Björn (not verified) :: Thu, 06/23/2005 - 6:42pm

Maybe he was following the CFL and was discombobulated because he was confused over the extra down the team had received.

32
by Dennis (not verified) :: Thu, 06/23/2005 - 6:56pm

Re #25: The small sample is totally relevant to the question of how meaningful the stats are as an indicator of actual ability. If they are a meaningful indicator of ability, then the generally the same players should be rated good or bad from year to year. Obviously players can have years that are much better or worse than their norm, but there should be some consistency to the numbers.

For example, if you look at baseball, a lot of the same players are at the top of the hitting stats from year to year. Sometimes you'll get a not-so-good hitter playing over his head like Jack Wilson last year or a good hitter having an off year like Todd Helton this year, but guys like Manny Ramirez are at the top of the league every year.

So if 117 attempts is meaningful for a cornerback, the you should see a lot of the same players doing well consistently or doing poorly consistently. I have no idea how these stats fare in that regard.

IMO, statistics serve two purposes: to measure past performance and predict future performance. The sample size is irrelevant when you are measuring past performance, you just want to know what the player actually did. But the sample size is extremely important to try to predict future performance.

33
by Paul (not verified) :: Thu, 06/23/2005 - 9:35pm

You want small sample size? Try calculating how likely the Shuttle is to fail on launch. Then to estimate the likelihoods for when and how the failure occurs. AND THEN to figure out how likely the debris will hit a person on the ground.
Now repeat the exercise for a brand new, never been launched rocket.

And being able to perform these calculations does NOT help me to beat my wife in picking football games during the fall.

34
by senser81 (not verified) :: Thu, 06/23/2005 - 11:54pm

re: post #32

Reminds me of the time when someone tried to downgrade Emmitt Smith's 1996 performance because he averaged 0.2 less yards per carry than Sherman Williams on 258 more carries than Williams.

35
by Basilicus (not verified) :: Fri, 06/24/2005 - 12:01am

One of the things I love about this site is that not only do the posters here debate the topics of the article, but we also discuss Aaron's business motives for posting a link to someone who's experimenting with football statistics (my opinion is that Aaron is truly excited someone else is pioneering new statistical paths and is simply geeking out about it and wants to share it since he knows we generally like to geek out about that sort of thing as well.)

On Carr vs. Harrington, I'll just put the stats for the QBs of the first three years of every new franchise since 1970.

David Carr: 685 of 1,205 (56.8%) for 8,136 yards (6.8 ypa), 34 TDs, 42 INTs; 159 carries for 732 yards and 5 TDs

Mark Brunell: 818 of 1,338 (61.1%) for 9,816 yards (7.3 ypa), 52 TDs, 34 INTs; 195 carries for 1,133 yards and 9 TDs

Kerry Collins: 618 of 1,177 (52.5%) for 7,295 yards (6.2 ypa), 39 TDs, 49 INTs; 100 carries for 177 yards and 4 TDs

Steve Spurrier (76 Bucs): 156 of 311 (50.2%) for 1,628 yards (5.2 ypa), 7 TDs, 12 INTs; 12 carries for 48 yards and 0 TDs

Gary Huff (77 Bucs): 67 of 138 (48.6%) for 889 yards (6.4 ypa), 3 TDs, 13 INTs; 8 carries for 10 yards and 0 TDs

Doug Williams (78 Bucs): 73 of 194 (37.6%) for 1,170 yards (6.0 ypa), 7 TDs, 8 INTs; 27 carries for 23 yards and 1 TD

Jim Zorn (76-78 Seahawks): 560 of 1,133 (49.4%) for 7,541 yards (6.7 ypa), 43 TDs, 66 INTs; 136 carries for 677 yards and 11 TDs

Statistics differ over the ages, so this only offers a limited scope, and parity has made it easier for new franchises to be successful earlier on, but even so it's obvious that Carr has been the most successful starting QB for a recent new franchise outside of Mark Brunell. I'm tired of writing down so many stats, so I'll just provide a few more new franchise morsels for thought: 66-68 Atlanta Falcons QB Randy Johnson 24 to 52 TD:INT ratio, 68 Atlanta's Bob Berry 7 to 13, 67 New Orleans Gary Cuozzo 7 to 12, 67-69 New Orleans Billy Kilmer 41 to 45, 68 Bengals John Stofa 5 to 5 (5.1 ypa), 69 Bengals Greg Cook 15 to 11 (and 9.4 ypa) but he only threw 3 passes outside of this year and that wasn't until 1973, 70 Bengals Virgil Carter 9 to 9 (5.9 ypa). Point is (sorry if I've gone nuts with the statistics), it's incredibly hard to QB a fledgling franchise and Carr's doing a decent job of it with a terrible offensive line and thus far erratic RB performance.

36
by senser81 (not verified) :: Fri, 06/24/2005 - 1:19am

I've kinda thought that Fran Tarkenton was the best expansion QB ever, and held Jim Zorn and Greg Cook in high regard. Even Eddie LeBaron was superior to David Carr. Anyways, Carr better start producing or he will be a footnote in history.

37
by Erasmus (not verified) :: Fri, 06/24/2005 - 2:29am

Re #28.

Thats why Joey's leap from getting sacked 11 times in 2003 to 36 times this past season was seen as a good thing by some Lions fans. He was actually taking sacks instead of throwing the ball away quickly.

and really the starting WRs for the last game of the 2004 season were Az-Zahir Hakim and Scotty Vines. Or you drive your team to a potential game-tying TD, he finally had that defining moment in your career-I mean he had the flu!!..and the long snapper botches the snap. I never get the extreme Joey bashing either. He needs work on mechanics and in-game decisions and really the WCO might not be the offense for him either-too bad he is not real fast or he will get a pass like Vick does. Rumor has it the Lions will open it up some this season-I will believe it when I see it.

Is Joey as bad as fans think he is? I do not think so.
Was Joey worth the 3rd pick of the NFL draft? I do not think so.
Will Joey get it right this season with a real TE (Pollard), and hopefully a running game and 3 should-be good WRs all season? As a Lions fan I hope so, but I have faith in Garcia back in the WCO as well.

38
by Bockman (not verified) :: Fri, 06/24/2005 - 9:07am

Aaron -

Can we please let Eli play a whole season before we label him overrated? I don't think anyone overrates him. The kid was a rookie, and he sucked. All rookie QBs suck. Big Ben had a great defense behind him last year, and still stunk up the place late in the season and in the playoffs. Peyton was 3-13 as a rookie. Just as everyone is in love with Clinton Portis for no reason, let's please be fair, as supposedly more knowledgeable football fans, to give Eli a little more than 7 games (all against top defenses mind you) before we label him anything.

39
by CatholicSamurai (not verified) :: Fri, 06/24/2005 - 10:31am

Can someone please explain to me the loathing when it comes to Warner? They act like he was horrible in NY when they had a WR corps that caught 1 touchdown last year, their offensive line was the biggest bunch of rejects this side of a NAIA school, and he had the #1 overall pick breathing down his neck.

Now he goes to a team that has a speed RB, a WR corps that could be one of the best in years, and an O-Line that can actually block. Hell, Emmmit Smith, who I believe was a back up to Jim Thorpe when he came into the league, ran for almost 1000 years and 9 touchdowns last year.

40
by Aaron (not verified) :: Fri, 06/24/2005 - 11:47am

Eli Manning certainly won't play at this level his entire career but there simply is no historical precedent for a rookie who played that bad to turn into the quarterback superstar the Giants were expecting. It could happen, but I doubt it.

Warner has been poor with two different teams for three seasons. Could he turn it back around in Arizona? It could happen, but I doubt it.

KC emailed me, as you can imagine he's deluged after getting press like this -- unlike PFP, his book isn't available on Amazon or in bookstores, so he's the writer, salesman, and shipping clerk -- but he plans on dropping in next week to answer some of the questions posed about his work.

And hey, it's reasonable to ask about my business motives. As long as KC and I can each pay our mortgages, it's all good.

41
by Richie (not verified) :: Fri, 06/24/2005 - 2:49pm

Count me amongst those who have high hopes for Warner.

42
by Glenn (not verified) :: Fri, 06/24/2005 - 3:28pm

I'm not trying to start more flame wars, but Joyner does have some thoughts on Peyton Manning: "What most people didn't realize about the Colts is that they were building up most of their big numbers against really, really bad secondaries."

And: "When the Colts faced a Patriots team that wasn't nearly as banged up in the secondary as everyone said they were (they were really only missing one starter, as Asante Samuel would very possibly have started at CB by the end of the year anyway), and the Colts didn't light it up, everyone was surprised it turned out like it did. Had they known the types of matchups Indy was facing and beating, if they'd kept perspective on it, the outcome wouldn't have been so surprising."

This linked excerpt doesn't give his thoughts on Brady, so technically, this is not a Brady-Manning thread diversion.

Technically. Honest.

43
by MDS (not verified) :: Fri, 06/24/2005 - 3:46pm

If he doesn't think highly of Manning, that's obviously one area where he has a major disagreement with Aaron's stats.

Joyner points out that Manning piled up the numbers on bad players like Roc Alexander, but then again, Alexander played in every game last year, and I don't recall anyone singling him out as a terrible player until he played against the Colts.

Anyway, my copy of Joyner's book should be arriving today, and I plan to devour it this weekend. I can't wait.

44
by B (not verified) :: Fri, 06/24/2005 - 4:06pm

The Colts may have faced inferior competition, but who else put up 6 TDs on Detriot or 200 yards on Roc Alexander?

45
by Kibbles (not verified) :: Fri, 06/24/2005 - 4:11pm

Re #38: I think the fact that Portis had the highest ypc, yards per game, and yards per start in the history of the NFL after his first two seasons is sufficient reason to be in love with him. Yes, he spent those 2 years in Denver, but he spent his last year playing with one of the worst passing games in the league. I have a feeling he'll be closer to the former greatness than the latter mediocrity over the rest of his career.

As for Ashley Lelie... I don't question that he's one of the scariest deep threats in the league. After all, he was 16th in DPAR despite the fact that he was abysmal on short routes, and pretty bad on intermediates, too.

I also think that there have to be a lot of asterisks on CB stats, based on who they faced and how much help they had. After all, Champ Bailey was facing guys like Marvin Harrison with no help, while guys like Brown were facing the opposing #2. I'd also bet he had a safety cheated over to help him more often, too.

46
by Kibbles (not verified) :: Fri, 06/24/2005 - 4:14pm

MDS, I hate to disagree, but Alexander definitely did NOT play in every game last year. He was the #5 guy on the depth chart, behind Bailey, Walls, Herndon, and Middlebrooks. He got work in the dime in all the games Walls missed, and work in the nickle after Middlebrooks was lost, too (which was the final two games of the season, iirc), but the only game he saw extensive action was the playoff game against the Colts, where Denver was in nickle all game.

47
by pm (not verified) :: Fri, 06/24/2005 - 4:17pm

RE: 43 "If he doesn’t think highly of Manning, that’s obviously one area where he has a major disagreement with Aaron’s stats.

Joyner points out that Manning piled up the numbers on bad players like Roc Alexander, but then again, Alexander played in every game last year, and I don’t recall anyone singling him out as a terrible player until he played against the Colts."

MDS, will you have a review on this book.

Anyway, my copy of Joyner’s book should be arriving today, and I plan to devour it this weekend. I can’t wait.

48
by pm (not verified) :: Fri, 06/24/2005 - 4:23pm

RE:47. MDS, I meant to say will you review the book?

49
by MDS (not verified) :: Fri, 06/24/2005 - 4:28pm

Alexander did play every game, but I don't know how many games he played only special teams.

I don't know if I'll review the book. I'm looking forward to seeing what Joyner has to say when he drops in here next week.

50
by Vern (not verified) :: Fri, 06/24/2005 - 4:28pm

Re: Peyton Manning (42,43)
I think we have to be careful how we read such analysis. It is not always a knock to say someone is "over-rated". You can be very good, yet still be expected to perform higher in certain games than your prior performance really suggests. It's the difference between your average performance vs your median or even better, your standard deviation in performances. I think Manning's standard dev is fairly high, (meaning more variation in game by game performances). Like a blue-chip stock, it will deliver over the course of a year, but that doesn't tell you whether it will go up today or even this month.

Side question, does FO take a position ala Bill James on whether "clutch" performance exists?

51
by MDS (not verified) :: Fri, 06/24/2005 - 4:32pm

Quick follow-up on Roc Alexander, click my initials for the play-by-play of the Broncos' opener. He made a tackle on a T.Green pass to E.Kennison in the fourth quarter of a close game, so appparently the Broncos' coaches had confidence in him from Day 1.

52
by Aaron (not verified) :: Fri, 06/24/2005 - 4:45pm

Regarding "does clutch play exist":

1) Dunno, haven't studied it yet.
2) I'm guessing no, except
3) There's an element of football that simply does not exist in baseball: the clock. Managing the clock is absolutely a skill that exists, and it can be the difference in close games, and it often comes off looking like "clutchness."

53
by Kibbles (not verified) :: Fri, 06/24/2005 - 5:02pm

Yeah, Roc won the coaches' confidence in training camp, and Walls was injured in the season opener, so he was the dime CB. Dime CBs will see the field on a handful of plays, especially since Champ was matched against Gonzo that game, meaning Roc was lined up against the #3 WR in sets with 3 WRs and a TE.

Still, it isn't exactly possible for a team to "expose" a guy who is only on the field maybe 10 plays a game and who is facing a receiver who, in general, really sucks.

He may have made an APPEARANCE in every game, but against the Colts was the first time he was regularly on the field for the majority of the game, which means Manning was pretty much the only guy able to expose him as a "terrible player".

For the record, I don't think he's terrible. I think he was a rooking facing a 1000 yard receiver and the best QB in the league in a playoff game after only about 2 games of experience in nickle packages.

54
by MDS (not verified) :: Fri, 06/24/2005 - 5:15pm

By the way, the latest I've heard is that Bill James doesn't take a position on whether clutch play exists. I think all he says is that he hasn't seen any evidence that it exists, but that doesn't mean it doesn't. I haven't read it, but I've heard that in Baseball Research Journal No. 33 he says he thinks clutch play might exist.

55
by MDS (not verified) :: Fri, 06/24/2005 - 5:17pm

"For the record, I don’t think he’s terrible. I think he was a rooking facing a 1000 yard receiver and the best QB in the league."

Agreed. I thought it was a little unfair how everyone piled on him after that game. I mean, Dre Bly has been to the last two Pro Bowls, and Manning made him look terrible, too.

56
by Kibbles (not verified) :: Fri, 06/24/2005 - 7:52pm

If anyone deserves to get piled on after that game, it's Coyer or Shanahan. Normally I'm a big fan of Shanahan, and I make no secret of the fact that I consider him one of the top 5 coaches in the league (and on the same level as Bill Bellichick, who is *not* the #1 coach in the league). Still, the definition of insanity is doing the same thing over and over again and expecting different results. It was obvious that Roc Alexander couldn't defend Reggie Wayne. They needed to change something, get him some help, put him on Brandon Stokely... ANYTHING other than just sit there and watch him get burned time and time again.

57
by Richie (not verified) :: Fri, 06/24/2005 - 8:12pm

Bill Bellichick, who is *not* the #1 coach in the league

Them's fighting words!

58
by Pat (not verified) :: Fri, 06/24/2005 - 8:38pm

1) Dunno, haven’t studied it yet.

Aaron:

From what I've understood, you have. From your own descriptions of the stats, Estimated Wins comes from the Forest Index which includes "offense and defense in the second half of close games" - which is the definition of clutch performance, right?

If Estimated Wins (projected to final record) at midseason predicts the final record of a team better than their actual standings, clutch performance exists.

The argument in baseball was that past clutch performance didn't correlate to future clutch performance, but if the previous statements are true, that's not true in football.

I would doubt that clutch performance wouldn't exist in football, for exactly the same reason you mentioned - the clock.

59
by Basilicus (not verified) :: Fri, 06/24/2005 - 8:48pm

My opinion on Shanahan has been slipping lately - I used to consider him a top 5 coach, and I don't think he's gotten any worse. I do think there are less bad coaches in the league than there were a few years ago and some really good assistants have ascended to head coaching positions - John Fox, Jim Mora, Jr. And I anticipate a major step back for the Broncos this year - last spring Aaron was saying Shanahan needed to make major changes or be dismissed because he kept doing the same thing year after year. I think he's made some terrible decisions this offseason that make me think he should be dismissed if they miss the playoffs even by a little this year. And, not to start a big discussion on this, but I do consider Belichick the best coach in the league right now (Reid being the second and Billick the third.)

60
by Björn (not verified) :: Sat, 06/25/2005 - 9:56am

Well, Shanny sure has made plenty of changes this offseason.

61
by Jeff F (not verified) :: Sat, 06/25/2005 - 11:02pm

As far as comparing Palmer and Harrington goes, look at the defenses they faced last year. When Palmer was starting off, the Bengals faced teams of far superior passing defense compared to what the Lions had to deal with.

Carson Palmer, a rookie, was introduced to the NFL by the following defenses:

at Jets, Miami, Baltimore, at Pittsburgh, at cleveland (first bad defense), and later in the season had to play against Baltimore and Pittsburg again, and also had to play New England, Buffalo, and Philly.

Ten of the games they played were against the best of the best defensive teams in the league (Pittsburg x 2, Baltimore x 2, New England, Buffalo, Jets, Washington, Denver, and Philly). And the first four games of the season were a gauntlet of daunting defenses, particularly for a rookie. Ten games against tough defenses.

By contrast, Harrington and the Lions had to deal with only 4 good defenses over 5 games: Chicago x 2, Washington, Philly, and Atlanta. Most of Detroit's games were against teams that didn't have very good defenses, like Green Bay and Minnesota.

Running a gauntlet of teams as a rookie versus playing a normal variety of teams as a third year player should count as a major difference. Palmer started slow but got hot at the end of the year, missed some significant game time too, whereas Harrington was his usual over-careful, underperforming self.

In short, Carson Palmer already has shown a tremendous amount of potential, and he has demonstrated this on the field with excellent performances later in the season, whereas Harrington is hitting the end of his trial period, and not impressing.

As far as Manning goes, maybe his ability to pick on players *that aren't doing their job properly* is better than anyone else, and that could be some of the reason for his exceptional numbers. Put him up against a team with few defensive flaws, and he isn't going to put out the same kind of performance as he does against most teams, which have definite weaknesses. So, in essence, Manning takes advantage of people that lack discipline.

62
by MDS (not verified) :: Sat, 06/25/2005 - 11:23pm

Palmer wasn't a rookie last year; his rookie year was 2003 when he didn't play.

Palmer played against better defenses than Harrington and also played with better teammates than Harrington. Playing with the same set of receivers, Kitna was better in 2003 than Palmer was in 2004.

"Put (Manning) up against a team with few defensive flaws, and he isn’t going to put out the same kind of performance as he does against most teams, which have definite weaknesses."

Yes, Manning looks better against bad opponents than he does against good opponents. Great insight. It's also true of every other athlete in every sport.

63
by Jeff F (not verified) :: Sun, 06/26/2005 - 12:55pm

Okay, it was Palmer's first year playing. I knew that, but forgot it. I don't know the Bengals 2003 schedule, but I know that the Steelers weren't nearly as formidable, and that their schedule was likely to be a lot easier in out of conference games, too.

And I was saying that Manning is better than others at picking up on mismatches/undisciplined players, of course people will perform better against worse opponents, but I think Manning takes better advantage of this than others. Or, maybe it's just because he has three great WRs to throw to, or a combination of both.

64
by Ryan Mc (not verified) :: Sun, 06/26/2005 - 8:47pm

Just a quick word of support for collecting these kinds of statistics.
I remember a Steelers-Bills game back in the 1990's when the Bills called a screen pass on 3rd and long. Kevin Greene was on the pass rush, but recognized the screen early and dropped into coverage on Thurman Thomas. Jim Kelly looked Thomas's way, but had to hold the ball and was sacked by Chad Brown. The whole play was made possible by Greene, and yet he was not credited with a single stat on the play. Maybe keeping track of great coverage which forces QBs to plan B, C or D would be revealing (although, admittedly, very difficult to assess and keep track of)