Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2018 Free Agency Cost-Benefit Analysis

Is Kirk Cousins the best free-agent quarterback in recent memory? Should Trumaine Johnson or Malcolm Butler have gotten the larger contract? And what makes a free-agent contract good or bad, anyway?

07 Aug 2007

Adjusted Penalty Rate, Part One

by Doug Farrar

On the afternoon of November 14, 2004, the Seattle Seahawks were driving downfield against the St. Louis Rams in an important divisional contest. Down 20-12 with 13:49 remaining in the game, Seattle quarterback Matt Hasselbeck was ready to take the snap from center Robbie Tobeck on the Seattle 40-yard line when referee Bill Leavy blew the play dead. Left guard Steve Hutchinson jumped early and was flagged for a false start, which pushed the Seahawks back five yards into a first-and-15 situation. Seattle couldn't convert this drive into points -- Shaun Alexander fumbled four plays later, and the Seahawks lost the game, 23-12 -- but that's not why Hutchinson's penalty was significant.

It was significant because Hutchinson has started every game for the Seahawks (2004-05) and Vikings (2006) since that Week 10 loss in 2004, and he hasn't been called for a regular-season penalty since. That's zero flags in 39 games for a player who's been in on just about every offensive play. His one false start in the 2005 NFC Championship Game aside, it's still a remarkable achievement.

Penalties are part of life for every (other) NFL offensive lineman. When analyzing their effectiveness, high numbers of holding and false start penalties seem to indicate two things: An inability to consistently block the men in front of them, and/or a lack of situational discipline. I recently read an ESPN.com article in which an “Overrated/Underrated list� of offensive linemen was based primarily on two factors: Holding penalties and sacks allowed. For this article, I'm going to include false starts, because linemen are debited as much, if not more, for false starts than for holds, especially in the public eye.

Ask yourself: What ELSE is Luke Petitgout famous for?

While a "sack allowed" might be subjective and based on the eye of the beholder, penalties are supposedly objective, in that their value is based on no external factors. A false start is a false start, and so on, and even the occasional blown call shouldn't skew the data. The problem with that concept is that the number of penalties called from crew to crew, and from season to season, fluctuates wildly.

In Pro Football Prospectus 2007, I wrote about the decisions made by the NFL Competition Committee and how they affected last year's play on the field. A primary point of focus for the Committee in the 2006 preseason was the frequency of the two most called penalties every year (with at least a 2-1 margin over any others): holding and false starts. In both cases, the concern was that officials were making too many calls. As a result, both decreased in 2006 -- holding calls were down 33 percent (868 to 570) and false starts down 15 percent (849 to 723).

There was an enormous variance from crew to crew, especially in holding penalties. In 2006, the 44 holds called by Walt Anderson's bunch were four times more than Bill Carollo's 11. The swing wasn't quite as drastic in 2005, but from Larry Nemmers' 75 down to Tony Corrente's 34, if we're rating players by their penalties, we need to take a more comprehensive look at the conditions under which those penalties are called.

So, as any writer with FO's methods drilled into his head would do, I established a league average for the 2006 season in holds and false starts, which allowed me to place a value on the calls from each crew:


Average Calls per Game: 2.22
Average Calls per Crew per Season: 33.39

Crew Penalties Yards Adjusted Rate
Bill Carollo 11 89 3.04
Walt Coleman 17 144 1.96
Peter Morelli 21 179 1.59
Bill Leavy 25 198 1.34
Ron Winter 28 220 1.19
Scott Green 30 248 1.11
Mike Carey 33 278 1.01
Jerome Boger 36 293 0.93
Terry McAulay 39 340 0.91
Gerry Austin 39 319 0.86
Tony Corrente 39 248 0.86
Bill Vinovich 41 322 0.81
Jeff Triplette 41 367 0.81
Ed Hochuli 42 332 0.80
Gene Steratore 42 309 0.80
Larry Nemmers 42 347 0.80
Walt Anderson 44 371 0.76


Average Calls per Game: 2.82
Average Calls per Crew per Season: 42.36

Crew Penalties Yards Adjusted Rate
Peter Morelli 31 155 1.37
Scott Green 35 172 1.21
Mike Carey 35 171 1.21
Jeff Triplette 35 174 1.21
Bill Vinovich 36 178 1.18
Terry McAulay 40 200 1.13
Gerry Austin 40 190 1.06
Walt Anderson 41 205 1.03
Bill Leavy 42 209 1.01
Bill Carollo 43 212 0.99
Walt Coleman 45 211 0.94
Larry Nemmers 45 218 0.94
Ed Hochuli 46 221 0.92
Jerome Boger 47 234 0.90
Gene Steratore 51 252 0.83
Tony Corrente 53 264 0.80
Ron Winter 58 288 0.73

Once I had a baseline, the next step was to integrate these Adjusted Penalty Rates into the FO penalty database (which is based on official play-by-play data) and re-run the numbers for the prime offenders in both holding and false starts among offensive linemen. Teams listed are the players' 2006 teams.

Player Team Actual
Penalty Rate
Chris Snee NYG 4 6.71 1.68 1
Andrew Whitworth CIN 4 4.59 1.15 2
Bob Whitfield NYG 5 5.65 1.13 3
Robert Gallery OAK 4 4.48 1.12 4
Richie Incognito STL 5 5.36 1.07 5
Langston Walker OAK 4 4.26 1.07 6
Casey Rabach WAS 4 4.14 1.04 7
Marco Rivera DAL 4 3.98 1.00 8
Bryant McKinnie MIN 5 4.96 0.99 9
Jake Grove OAK 4 3.92 0.98 10
Kwame Harris SF 5 4.89 0.98 11
Jeff Backus DET 4 3.80 0.95 12
Marcus Johnson MIN 4 3.60 0.90 13
Walter Jones SEA 5 4.49 0.90 14
Kevin Shaffer CLE 7 6.18 0.88 15
Khalif Barnes JAC 5 4.31 0.86 16
Wayne Gandy ATL 4 3.43 0.86 17
Jon Jansen WAS 5 4.10 0.82 18
Jonas Jennings SF 4 3.20 0.80 19
Reggie Wells ARI 6 4.77 0.80 20

A few notes about this list: Snee's APR is so high because he's the only offensive lineman on our list called for holding by Bill Carollo's crew in 2006. Of Carollo's 11 holds, four were called on special teams plays, three on receivers, and one on a tight end. When you get flagged by a crew that called three holds on offensive linemen all year, it's safe to say that either the NFL spoke to the officials before your game, or you must have been pretty active in the extracurricular department. Snee also had a hold called by Walt Coleman, who called the second fewest holds in the league. And does anyone else find it amusing that he's Tom "Mr. Discipline" Coughlin's son-in-law? Snee probably doesn't have a lot to laugh about at family dinners. Browns tackle Kevin Shaffer led the league in holding penalties with seven, but he was flagged by six different crews, and only one (Mike Carey's) had an APR above 1.0.

While 2006 was Andrew Whitworth's rookie year, Bob Whitfield retired after the 2006 season and a 15-year career. Robert Gallery continues his journey to legend as one of the most disappointing high draft picks in recent memory. We'll see what a new, supposed emphasis on offensive position coaching, and a zone blocking scheme, does for that Oakland line.

No, the addition of Walter Jones is not a typo. The league's best lineman in 2005 (and possibly in some seasons before) struggled with injuries and line inconsistency and had his worst season to date in 2006. And isn't it a shame that Langston Walker signed with the Bills? This means that the Raiders could lose their tie for the league lead in holds (25) with Minnesota. Speaking of Minnesota ... between Bryant McKinnie and Marcus Johnson, no wonder Hutchinson wasn't called for any penalties last year. Who had time to look at the guards? Reggie Wells' Adjusted Hold Rate was a product of the fact that four of his penalties were called by three of the bottom four crews in APR, including two by low man Walt Anderson.

Now, the false starts:

Player Team Actual
False Starts
False Starts
Penalty Rate
Chad Slaughter OAK 5 5.85 1.17 1
Matt Light NE 5 5.78 1.16 2
Chris Spencer SEA 5 5.47 1.09 3
Leonard Davis ARI 8 8.30 1.04 4
Shane Olivea SD 6 6.22 1.04 5
Alex Barron STL 13 13.40 1.03 6
Jake Scott IND 5 5.11 1.02 7
Jason Brown BAL 7 7.15 1.02 8
Todd Steussie STL 6 6.06 1.01 9
Derrick Dockery WAS 6 5.95 0.99 10
Jeff Backus DET 8 7.93 0.99 11
Brandon Moore NYJ 6 5.90 0.98 12
Jeremy Trueblood TB 6 5.72 0.95 13
Tarik Glenn IND 7 6.62 0.95 14
Fred Miller CHI 5 4.72 0.94 15
Mike Schneck BUF 5 4.56 0.91 16
Reggie Wells ARI 5 4.51 0.90 17
Jason Peters BUF 5 4.44 0.89 18
Jordan Gross CAR 6 5.16 0.86 19
Roberto Garza CHI 5 4.23 0.85 20

Note: We're talking about offensive linemen specifically, which is why Jason Witten's six (APR: 6.61), and Frank Gore's five (APR: 5.72), false starts are not included. And no, there's no Luke Petitgout here. His 10 false starts in 2005 (which didn't even lead the league -- thank you, Leonard Davis) fell to three in 2006.

An Oakland player at the top of this list? Gosh. Spencer started at center for the first time in the last half of the season on Seattle's depleted line. He could be a good one in time, but those numbers are no surprise. Steussie begins his second season as a member of the Rams' line in 2007, which begs the very obvious question: Steussie and Barron? On the same line? Huh? Given the stellar reputation of the Indianapolis offensive line, some might be surprised to see Scott and the now-retired Glenn on this list, but it's easy to assume that Peyton Manning's detailed and time-consuming audibles have something to do with that.

Schneck's story is interesting for two reasons: he's the only long snapper on our list, and he made the Pro Bowl for the first time after the 2005 season. The follow-up season? Not so much. Both Leonard Davis and Derrick Dockery became members of the $49 Million Club in the off-season, with Davis going to Dallas and Dockery joining Langston Walker in Buffalo. If you're looking for a penalty upswing in 2007, Buffalo's a good place to start.

Lions left tackle Jeff Backus is one of two players on both lists. That's one of many reasons Detroit's line ranked dead last in Adjusted Line Yards and 29th in Adjusted Sack Rate last season. That Reggie Wells is the other player ... well, all I should have to say is that he's a member of Arizona's offensive line. That line actually had a decent upturn in the second half of the season, but three of Wells' five false starts also came in the season's second half. Two of those were called by Ron “I call more false starts before 9:00 AM than other crews do all day� Winter, which is the primary reason Wells' APR is so low. Not that he's a future Hall-of-Famer by any means, but based on the data in these two tables, Wells might have a point if he ever wants to complain that he has a bull's-eye on his uniform.

The data that goes into Adjusted Penalty Rate isn't 100 percent comprehensive; I'm considering how best to integrate factors like red zone and third-down penalties for both players and crews. A home/away adjustment for false starts would make sense. Different data could be more or less important for different penalties, and not every crew calls games for every team in every season. (In fact, with a small sample like penalties, it is probably best to judge players over multiple seasons.) However this is done, teams should start scouting the tendencies of officiating crews as they would any other “opponent.� Some do this already. Until penalties are called more consistently, statistical adjustment to an officiating crew might be the only way to get a real sense of which penalties matter the most, and who's really most responsible.

In part two of our series on Adjusted Penalty Rate, the focus will shift to pass interference and illegal contact. If you thought the variance on false starts and holds was wacky, you ain't seen nothin' yet.

Posted by: Doug Farrar on 07 Aug 2007

32 comments, Last at 17 Aug 2007, 9:53am by stan


by JasonK (not verified) :: Tue, 08/07/2007 - 6:17pm

The ref tendencies are interesting, but I have trouble believing that a sample size of a half-dozen plays is enough to make any meaningful conclusions about the quality of the players.

by asg (not verified) :: Tue, 08/07/2007 - 6:20pm

Maybe I'm being dense, but what adjustment exactly is being made to get APR from penalties and yards?

by Dan (not verified) :: Tue, 08/07/2007 - 6:52pm

Shouldn't the adjustment be made based on which crews officiated the player's games, rather than on which crews called the penalties on that player? Shouldn't a player be rewarded for getting no holds when faced with a crew that calls lots of holds (just as Adjusted Sack Rate rewards the pass protection of teams that give up no sacks against a defense that gets lots of sacks)?

Also, I think you're overadjusting for crew. By forcing the number of "adjusted penalties" called by each crew to be constant, you're essentially assuming that all of the variation in penalties called by the crews is due to the crews' biases. But surely some of the variation is caused by which crews happened to officiate the games where players committed more penalties and which happened to be assigned to the cleaner games.

by perrin (not verified) :: Tue, 08/07/2007 - 7:48pm

Love the article, Doug, and I'm looking forward to the rest of them.

For me, the biggest problem in professional sports (including college teams) is inconsistent (and frequently imprecise) refereeing. I feel yards better when Hochuli is in command of a game, and considering all the money and time invested by so many people, that's a huge problem. Any analysis of the tendencies of officials is welcome. Thanks, FO.

by Pippin (not verified) :: Tue, 08/07/2007 - 8:06pm

This just makes me even more stressed out that the Cowboys picked up Leonard Davis. Why? Why?!

by Richie (not verified) :: Tue, 08/07/2007 - 8:21pm

That's a great stat about Hutchinson. Do we have any idea how that compares to recent league history?

Why can't a stat like that be more mainstream?

by Brian (not verified) :: Tue, 08/07/2007 - 8:27pm

I have to agree with both #3 and #4. I think it would have been just as interesting if you just stopped the article at the analysis of which crews throw the most flags and how much they vary.

by Erik Smith (not verified) :: Tue, 08/07/2007 - 8:42pm

"Once I had a baseline, the next step was to integrate these Adjusted Penalty Rates into the FO penalty database (which is based on official play-by-play data)."

Are there any FO metrics for penalties by team? Or is this an area whose measurement for ranking purposes is as murky as I expect it to be?

by Larry (not verified) :: Tue, 08/07/2007 - 11:06pm

Why are the players ranked by APR and not by Adjusted Penalties? It would seem to me that Todd Steussie's 13 false starts make him the worst in that category, whatever the adjustment is. Also, getting close to 1/game, with your formula we'd expect his APR to be close to one (the rating of the average crew), assuming he drew a pretty representative sample of crews over the season.

Also, what Dan said in #3, I agree completely and couldn't say it any better.

by kibbles (not verified) :: Wed, 08/08/2007 - 12:12am

Steussie begins his second season as a member of the Rams’ line in 2007, which begs the very obvious question: Steussie and Barron?

It raises the question, it doesn't beg it. Begging the question is a logical fallacy in which the conclusion of an argument is implicitly or explicitly assumed in one of the premises.

by Pat (not verified) :: Wed, 08/08/2007 - 12:15am

Yeah, I don't understand what "adjusted penalty rate" is supposed to mean, either. It looks like it's just the average rate that the officials called penalties at for each lineman. I don't see why this would have anything to do with the quality of each player.

Adjusted holds/false starts seems more indicative.

By forcing the number of “adjusted penalties� called by each crew to be constant, you’re essentially assuming that all of the variation in penalties called by the crews is due to the crews’ biases.

Right - what you want to do is an iterative convergence. Take, say, Ed Hochuli's crew: for each team they faced, sum up the false starts/holds for the teams they reffed in games where they weren't the refs, divide it by the number of games (to get an average), then divide the number of false starts/holds by the average rate that that team experiences. Average that number for the year, and there's your "adjusted average penalty calling rate" for each crew.

Repeat for the teams involved to get an "adjusted average penalty rate," then reinsert that to recalculate the calling rate, etc., and let it converge.

by Doug Farrar :: Wed, 08/08/2007 - 1:25am

I appreciate the comments, everyone. As I told Aaron, I think I'm on to something here, but I'm not 100% sure what just yet. The calculations were my primary concern in that I wasn't completely sure what they would tell me, and sample size is an obvious issue. I have some ways in which I think this can be improved, there are some great ideas here, and I'll be running the concept by a few more people before the interference/contact data is assembled (we do have multiple years of penalty data).

by NY expat (not verified) :: Wed, 08/08/2007 - 3:28am

Perhaps you've already considered it, but you might try to consider how many plays were run and how many pass vs. rush plays were called.

by Moses (not verified) :: Wed, 08/08/2007 - 7:37am

I think the analysis is bogus because it's clearly not adjusting for referee error. I've seen some of those holding penalties because I've watched the games in which the occurred.

Since you're not factoring out the obviously bad calls that even I know about, your analysis, exceptionally sensitive on the player side has little credibility and should be omitted in that area. For example, I know for certain at least two of the linemen in your Top-20 were called for holding when the d-lineman actually won the battle of the LOS and bowled the o-lineman over. After winning the battle, the d-lineman tripped over the o-lineman and fell on him. Holding was called, even though replays clearly indicated the beaten lineman wasn't holding, because the referee, behind the o-lineman ASSUMED he was and threw the flag.

You know, one of Paul Brown's HOF accomplishments was introduction of statistical film study into the grading process. Without using film, you can't correct for errors and actually get it right. Especially in something like this analysis where one bad call in 500 plays could represent 50% of the "failure" assigned to the player.

Another possible failure in this analysis is that, according to STATS, Inc., your numbers are all wrong. Chris Snee had 2 holding calls. Kwame Harris had 3 holding calls. Robert Gallery had 3 holding calls. Walter Jones had 2 holding calls. And, having seen you go 100% wrong with my spot checking, I guessing that even without going over all of them, I'm thinking they're ALL wrong.

So, highly sensitive analysis not cleaned up for garbage and numbers that don't jive with a very reputable source... I'll be blunt, if this were your dissertation, your advisor would send you back to the lab to try again.

by Doug Farrar :: Wed, 08/08/2007 - 8:24am

Re: 14 - I can't speak to how the Stats, INC information was put together. Frankly, I haven't seen it. All I know is that the information used in this article comes from line-by-line data in official play-by-play, including offsetting and declined (which might explain the discrepancy; I included offsetting and declined because the penalties were called, regardless of what happened after).

Let's take Snee, for example. He had a holding penalty in Week 2 against the Eagles (accepted, Carollo); another in Week 6 against the Falcons (accepted, Coleman), another in Week 14 against the Panthers (declined, McAulay) and the fourth in Week 15 against the Eagles (offsetting, Austin).

Harris had five total: Week 1 against the Cardinals (accepted, Hochuli); Week 2 against the Rams (declined, Steratore); two in Week 6 against the Chargers (both accepted, Corrente) and Week 15 against the Seahawks (accepted, Morelli).

In Snee's case, it's quite possible that the offsetting and declined were not counted. Not sure on Harris, unless maybe the declined penalty wasn't counted and that still leaves one unexplained acording to your numbers.

The method needs work (and I've never denied that), but I'm comfortable with the data.

by Jimmy (not verified) :: Wed, 08/08/2007 - 9:26am

Is there any way of finding out who gets held the most? It would seem likely that good defensive linemen get held a lot, but have no statistical information to back that up. Drawing a hoding penalty will stop most drives just like a sack or a tackle for loss, does that kind of thing come up in DVOA or game charting?

by Fan in Exile (not verified) :: Wed, 08/08/2007 - 9:28am

I would be curious to see this data integrated with the data for D-lines. Are there some d-lines that get held more? It might help to get a handle on a D-linemen who is better than commonly thought because he gets held more often. Or even be a way of supplementing the sacks, hits, hurries, data that was in the book. Or it could be nothing, but that would be nice to know as well.

by Sergio (not verified) :: Wed, 08/08/2007 - 10:35am

Amazingly, Miami's OLmen aren't in the top offenders. I say amazingly, because I got the distinct impression of them being prime candidates for this sort of study. Apparently it was unfounded...

...well, it was wrongly founded. They seemed to have the "bad place, bad time" syndrome (they had a three-play consecutive false start problem at least twice in the season, IIRC).

by Theo, Holland (not verified) :: Wed, 08/08/2007 - 10:43am

Are you going to find out if Tackles commit more penalties then interior linemen?
Are you adjusting it per play??
I would guess that players on a bad teams (or running teams) play more snaps and pass more. And commit more penalties. But maybe that's a chicken and the egg story.
Is a hold always bad? I guess Alex Smith wished some Kwame Harris plays were holds instead of sacks.
Where's Pittsburgh?!

by Phill Skelton (not verified) :: Wed, 08/08/2007 - 11:35am

How much wieght should be assigned to crews vs to individual referees? I'd assume that all O-line holding calls would be more or less calls by the same person on that crew. Whereas a special teams holding call, or a receiver downfield holding is unlikely to be the same ref. And I can't see any compelling reason that the willingness to call holding is going to be the same for all refs in a crew.

So maybe pull out only holding calls by O-line, which are all likely to be the same ref on a crew. Of course the same ref would also call holding on TEs and RBs with pass blocking assignments too I imagine, but trying to judge *which* ref was responsible for calling holding on a particular play is probably going to create more problems than it could ever solve.

To establish some validity for this though given the small numbers, you really need to do the same things for previous years and show whether the year to year variation for crews is the same size as or much smaller than the inter-crew variation (for crews that have kept the same set of people, or at least the same guy calling O-line penalties).

by NewsToTom (not verified) :: Wed, 08/08/2007 - 11:44am

Re #16
To the best of my knowledge, that data is not kept by any publicly-available source. Specifically, it doesn't show up in the official PBP from which Doug got the stats in this article, it's not tracked in the Game Charting Program; in fact, it's not even officially announced by the ref, so we the public can't be sure what they called the hold for. Most of the time you can see from the broadcast who the hold was against, but not always. I agree it would be interesting data to have, and I'll probably keep track of it for the Titans this year, but I'm pretty sure it's not out there anywhere.

by Marko (not verified) :: Wed, 08/08/2007 - 12:19pm

"All I know is that the information used in this article comes from line-by-line data in official play-by-play"

Does the official play-by-play correct errors by officials in identifying players? It's not unheard of for an official to call (for example) "holding, offense, number 75," when in reality, (1) the team doesn't even have a number 75 (or it does have a number 75, but he is a defensive lineman and obviously was not on the field), (2) number 75 is an offensive lineman, but he wasn't even on the field at the time, or (3) number 75 is an offensive lineman and was on the field, but the hold actually was committed by someone other than number 75.

by NewsToTom (not verified) :: Wed, 08/08/2007 - 1:46pm

Re #22
When the game charters saw an error in the PBP, they tended to note it. I believe in PFP it says like 175+ changes were made to the PBP based on game charter comments, though I can't be sure if any of those were penalties or not. I know that when I thought the penalty was called on the wrong person, I noted it as an error.

As a general rule, the official scorer will tend to correct for (1) and (2) (I don't remember any penalties in the PBP this year called on people not on the field), and (3) is where you have to review the games and can't be sure even if you do. Given the small number of offensive line penalties over the course of a year, I agree that this can be a distorting factor, but the bigger distorting factor is what officials miss or decide not to call (see MDS's post-SB XL EPC, Sean Locklear and the 22 Uncalled Holds).

by Pat (not verified) :: Wed, 08/08/2007 - 3:50pm

Does the official play-by-play correct errors by officials in identifying players?

Yep. That's why the play-by-play changes over the season. They do try to make them as correct as possible.

I don't think #3 happens very often, and a lot of times, it's due to the fact that the TV announcers think it was a different person, when it wasn't.

by Marko (not verified) :: Wed, 08/08/2007 - 4:22pm

Thanks, Pat. As for (3), yes, I am quite aware that TV announcers are often wrong. In my example, I was talking about when the penalty really was committed by a different player, not when the TV announcer mistakenly says that the penalty was committed by a different player.

by Dan (not verified) :: Wed, 08/08/2007 - 6:16pm

Re: 14

I could be wrong, but I think you’re missing a large part of the purpose of the article. You say that it should be corrected for “bad calls�. That’s fine. But as far as I can tell, that’s partly what Doug is doing by adjusting for each crew. By adjusting for crews, you not only adjust for the leniency of officials when it comes to certain penalties, you also adjust for the frequency of bad calls each crew makes on those penalties. Though as others have mentioned, this sample size is probably not sufficient to do that very well.

by Rich Conley (not verified) :: Wed, 08/08/2007 - 7:03pm

"I don’t think #3 happens very often, and a lot of times, it’s due to the fact that the TV announcers think it was a different person, when it wasn’t."

Pat, I've definitely seen the referees call out the wrong number more than once....

but if the nfl is correcting that..sweet.

by bengt (not verified) :: Thu, 08/09/2007 - 7:09am

There are 17 crews, and they officiate 256 games per (regular) season. That makes 15.06 games per crew, which is obviously impossible. I take it you have used a normalization, but it is not mentioned in the text?!

by A for Anonymous (not verified) :: Thu, 08/09/2007 - 12:34pm

#5, #19
I'm really interested if tackles commit more/less penalties than interior lineman, because Davis is supposedly changing from T to G. He won't be worth $49 Million, but he probably won't have 8.3 AFS. DT's try to go through him more than DE's that tried to go around him (pretty successfully, he's not going to be good because he has quicker feet than DE's)

by Fire Millen (not verified) :: Thu, 08/09/2007 - 2:40pm

Overall an interesting article. I have one large question, why does an APR have significance/interest? It is the ratio of adjusted to actual penalties, why do I care what this ratio is? A more signifcant rate measure would be adjusted penalties per play, or the inverse the expected number of plays before a player incurs a penalty.

by refchat (not verified) :: Fri, 08/10/2007 - 8:09am

Great article! Good to see you recognize the limitations also. It is very interesting, though.

In my statistical analysis of referees, I've been trying to figure out what characteristics actually vary among referee crews. The yardstick I have been trying out is to see whether there is a correlation between the rates in the first half of the season and the second half of the season. Do you think that's a good yardstick and how would crunching the numbers for each type of penalty turn out? I'm not sure it's the best measure, but my rough analysis suggested some referee crews are better than others for favorites. But that there does not seem to be a consistent characteristic of whether one referee crew is better for home teams than another crew.

Is there a better yardstick for seeing which characteristics of a referee crew stay consistent over the course of the season?

by stan (not verified) :: Fri, 08/17/2007 - 9:53am

Tarik Glenn always had a bad tendency to jump when he was struggling against a good DE. I always thought that his making the Pro Bowl was a joke. It wasn't based on observations of his play, but rather faulty "logic" (e.g. left tackles are critical to pass pro, Manning rarely gets sacked, therefore ... Tarik Glenn must be really good).