Writers of Pro Football Prospectus 2008

14 Nov 2005

2005 Quick Reads: Week 10

Here are this week's Quick Reads, the top players according to DPAR and all the quarterbacks. Some surprising names on top of the list this week, in part because the best quarterback performances of Week 10 came against the league's worst pass defenses and thus were less impressive. Plus, find out which familiar name from the past matched Jake Plummer's magical trick of making interceptions disappear, and find out which running back projection from PFP 2005 was shockingly accurate.

Posted by: Aaron Schatz on 14 Nov 2005

31 comments, Last at 16 Nov 2005, 10:58pm by Matt Weiner


by Dan (not verified) :: Mon, 11/14/2005 - 9:21pm

Wow, we've had a Bob Lobel reference. However, I can't let that go by without mentioning the popular local game of "How sauced is Bob Lobel tonight."

by Al (not verified) :: Mon, 11/14/2005 - 9:25pm

That looks like a bizarro QB top 10. Dilfer! Batch! Carr! Oh my!

by MCS (not verified) :: Mon, 11/14/2005 - 9:28pm

I'm curious. How did Samkon Gado rate?

by MCS (not verified) :: Mon, 11/14/2005 - 9:31pm

Am I missing soemthing or are you guys a little behind on the stat updates?

See Link.

by admin :: Mon, 11/14/2005 - 9:40pm

Been busy. Promise to update after tonight's game.

by rk (not verified) :: Mon, 11/14/2005 - 9:41pm

Given that Charlie Batch only played the first half, his DPAR is pretty amazing for a guy who started the year as the 3rd stringer.

by MCS (not verified) :: Mon, 11/14/2005 - 9:54pm

Thanks, guys. As always, great work!

by Matt Weiner (not verified) :: Mon, 11/14/2005 - 10:24pm

I can't possibly be the first person to be thinking of a Cadillac and Edsel joke....

by Erik (not verified) :: Mon, 11/14/2005 - 10:53pm

What's going on? 170 rushing yards, 1 for 1 on receptions for 9 yards, THREE touchdowns, and Shaun Alexander isn't one of the top five running backs? Huh?

by luz (not verified) :: Mon, 11/14/2005 - 11:33pm

i think your QB ratings are screwed up. Randle El for President! No, seriously, he'd be an improvement as President!

by Devin McCullen (not verified) :: Mon, 11/14/2005 - 11:44pm

I would assume that Shaun Alexander's a victim of opponent adjustment (Rams are ranked 27th in Rush Defense). I do think he wins Best Fantasy Player, though - although Tomlinson having the week off never hurts.

by JonL (not verified) :: Tue, 11/15/2005 - 12:20am

A somewhat tangential question, but what's the overall DPAR for rushing plays by wide receivers? It has to be negative, right? What about the percentage of those plays that actually work? Are there any particular types of WR running plays that work better than others?

by Israel (not verified) :: Tue, 11/15/2005 - 2:56am

Yesterday, this boy became a man. I'm sending him some Israel Savings Bonds.

and a fountain pen.

by Kibbles (not verified) :: Tue, 11/15/2005 - 4:18am

Re #9: You're placing too much emphasis on the rushing TDs. DVOA values touchdowns far less than conventional stats.

Here's a hypothetical for you. If a QB gets the ball at the 10 yard line, goes 6 of 6 for 89 yards, and the RB runs it in for the 1 yard score, was that an amazing drive for the QB, or for the RB? A rushing TD is valued about the same as a rushing first down, since they're essentially just as difficult to get.

That's why Alexander's stats are misleading. You mention he was 1 for 1 on receptions for 9 yards. What if that 1 reception came on 3rd and 10? You mention he rushed for 170 yards? What if they all came in 9 yard chunks on 3rd-and-10? How valuable would his day have been, then, if that was the case? In fantasy terms, very. In reality terms, not very. In addition, there's the adjustment for opponent strength.

I actually checked the play-by-play to see why DVOA was rating Alexander lower than conventional stats. According to DVOA, a play is considered successful if it nets 40% of the necessary yards on 1st down, 60% on second down, or 100% on 3rd/4th down. By that simple measure, Alexander had 13 successful plays in 34 touches. Included in that list is a long litany of plays that look like the following:

2nd and 10- gain of 2.
2nd and 10- loss of 3.
2nd and 5- no gain.
2nd and 10- gain of 2.
2nd and 9- gain of 3.
2nd and 5- no gain.
2nd and 7- loss of 1.

That's a lot of really bad 3rd down situations to put your QB in. If it weren't for a slew of timely Rams offsides penalties, Seattle's offense would have scored a lot less.

Shaun Alexander definitely had a GOOD day, but it was too inconsistant to be considered great.

Re #12: I would imagine that the overall DPAR for rushing plays by receivers would actually be POSITIVE, much like the overall DPAR for rushing plays by QBs would in all likelihood be positive. Since they're called so rarely, the opposition usually isn't ready for them, and they often result in a big play. They have a high percentage of failures, but they have a high percentage of huge plays. For instance, in week 1, Miami gained as many yards rushing on a single end-around to Chambers as they did on every other play of the game combined.

Theoretically, if football coaches were perfect and teams played an infinite number of snaps, the overall DPAR for all plays would be 0. If the overall DPAR for rushes by receivers was greater than 0, then the coaches would call more rushes by WRs. That would result in defenses focusing more on stopping them, which would result in a drop in the DPAR of rushes by WRs until it was negative, which would result in offenses not calling them, which would result in defenses growing lax in defending them, which would result in the rushing DPAR for WRs to rise, and on and on in a big cycle until the league achieved equilibrium. That's the same reason why coaches throw on 3rd-and-2 sometimes. They know that rushing plays have a better chance of converting, but they ALSO know that if they run every time, the defenses will sell out against the run, which will reduce the chances of rushing plays converting (and increase the chances of passing plays converting). The reason coaches pass on 3rd-and-2 isn't because they think they're more likely to convert that specific third down... it's because they're then more likely to convert every single rushing third down for the rest of the season.

by tubbs (not verified) :: Tue, 11/15/2005 - 5:12am

Does quarterback DPAR take into account accurately thrown passes that were dropped by receivers and other types of incompletions that were more the fault of the receiver, like an obviously blown route (not sure if you can accurately measure something like that)?

PS That "Billy Joe" joke is really lame.

by JonL (not verified) :: Tue, 11/15/2005 - 11:38am

RE: #14

That makes sense, but I guess my reasoning relied more on all those failures. I guess it all depends on in which situations those plays are called.

by Todd S. (not verified) :: Tue, 11/15/2005 - 11:46am

Aaron, you should link these to an Atlanta fan board. "See: Our rankings this week have Vick only one spot behind Peyton Manning. AND, Vick is higher than the average of PeyTom Branning. What's not to like?"

#15 I dunno...I kind of like the "Billy Joe" thing.

by Scott de B. (not verified) :: Tue, 11/15/2005 - 11:54am

Theoretically, if football coaches were perfect and teams played an infinite number of snaps, the overall DPAR for all plays would be 0.

I think you're confusing DPAR with DVOA. DPAR has a replacement player baseline. If the overall DPAR for all player were zero, then a team could achieve the same results by dumping all its players and playing NFL Europe scrubs the minimum salary. Total value for all plays should have a positive DPAR.

by Parker (not verified) :: Tue, 11/15/2005 - 11:56am

It's nice to see Kurt Warner's name mentioned in a positive way again. I had some admittedly insane fantasies that he would rise from the ashes in Arizona. While that seems pretty unlikely, it's still nice to see him have a decent game here and there.

Go Kurt Warner.

by MJK (not verified) :: Tue, 11/15/2005 - 1:53pm

I would expect DPAR to be positive for runs by WR's, but DVOA to be negative. Since DVOA is a measure of percentage of success per play, and since so many more of them fail than succeed, wouldn't that give a negative DVOA? On the other hand, because DPAR measures total value, not average value, I would expect the few that succeed big to outweigh the many that fail. In other words, a 20 yard run on a 1st and 10 reverse is just as good as a 10 yard run from DVOA's point of view, and if four other reverses all lost a yard, then the play would have succeeded only 20% of the time. But I would think the 20 yard gain would have more value towards total points gained than the net 4 yards lost from the other attempts. Am I understanding correctly?

And Kibbles, you're talking about a Nash Equilibrium, right?

by Shalimar (not verified) :: Tue, 11/15/2005 - 2:00pm

Steve Mariucci told the sideline reporter, "It's nice to get our receivers back, Roy and Charles."

He should have said "It's nice to get our receivers back, Roy and the other one." Anything to motivate.

by Ashley Tate (not verified) :: Tue, 11/15/2005 - 2:10pm

Gruden Goes For Two: Foolish Folly or Genius Gamble?

(A probabilistic "analysis")

Washington's painful one-point loss to Tampa Bay yesterday was about as gut-wrenching a seesaw battle as I've seen in a while, with huge momentum-changing plays every few minutes, four critical instant-replay reviews, and one unreviewable blown call.*


by DavidH (not verified) :: Tue, 11/15/2005 - 3:29pm

VOA, DVOA, and DPAR are explained on the page linked in my name. To summarize, no you don't have it quite right. Yes, you get zero points for an unsuccessful play, but successful plays are not limited to one point, so it's not as simple as your example. Also, DVOA and DPAR are both based on the same success points system. The only differences are that DVOA is a rate stat (per play) while DPAR is a counting stat, and that DVOA is compared to average while DPAR is compared to replacement level.

A successful play is worth one point, an unsuccessful play zero points. Extra points are awarded for big plays, gradually increasing to three points for 10 yards, four points for 20 yards, and five points for 40 yards or more. There are fractional points in between. (For example, eight yards on 3rd-and-10 is worth 0.54 "success points.")

There is also a red zone adjustment, and you can get negative points for fumbles.

This is just the raw "success value" part. The next step is to compare a team's success points on the play to the average success points for all other previous plays that had the same (or similar) down, distance, field location, score, time remaining. The difference between the actual success points and the predicted success points is then divided by the predicted success points, and you get VOA.

DVOA is VOA that gets adjusted for opponents, average fumble recovery rates, and I don't know what else.

DPAR starts with the same success point calculation, but then instead of comparing the play to an average play, it compares it to a "replacement level" play, which they estimate as 13.3% below average. Then, instead of dividing the difference by the average to get a percentage, they just add up the differences from all the plays. This results in "total success points above replacement" which gets converted to points by multiplying it by a conversion factor (.48). Where they get the factor is explained on the linked page as well.

by DavidH (not verified) :: Tue, 11/15/2005 - 3:53pm

Having said all that, I think it means that you can't have a very negative VOA and yet have a positive PAR. I believe a player with a VOA of -13.3% should have a 0 PAR. If you want to say that the WR runs have a 0 VOA, then yes, they'd have a positive PAR, since you're comparing PAR to a lower-than-average baseline.

Once you add in the defensive adjustment, I have no clue what would happen.

by JonL (not verified) :: Tue, 11/15/2005 - 4:10pm

Perhaps my question is better stated as, how often do wide receivers run the ball above replacement level? Would a team be better off putting in a replacement level RB? Probably not strictly speaking, since WR running plays involve all sorts of shenanigans and tomfoolery, but I suppose a team could put in a RL RB at WR and run the play that way. It's possible that the answer is, in fact, fairly often.

My original thinking on why the DPAR would be negative is that WRs aren't really taught to tuck the football, make their bodies small, etc., but instead tend to run more upright, leaving them vulnerable to tackling. Comma.

by Mercury815 (not verified) :: Tue, 11/15/2005 - 4:15pm

My favorite line in this article is the pat on the back Aaron gives to PFP 2005 for predicting Tiki Barber's production. As a guy who relied upon PFP 2005 to take Kevin Jones as a very early draft pick, I strenuously object to these types of comments. I'll never trust PFP again.

by Vlad (not verified) :: Tue, 11/15/2005 - 4:53pm

"Gado is closer in age to Favre's daughter, Brittany, than he is to Favre."

Anybody else take a moment to size them up as a potential couple when they read that? Just me?

by Kibbles (not verified) :: Tue, 11/15/2005 - 6:16pm

Re #20: And Kibbles, you’re talking about a Nash Equilibrium, right?

I don't know. I'm talking common sense. Is that what a Nash Equilibrium is, though?

I said the same thing about Tennessee's run defense in... 2003, I think. Conventional stats had them as the #1 run defense in the entire NFL. DVOA had them as smack in the middle of the NFL, and said they gave up so few yards because they faced so few runs. I always argued that head coaches weren't stupid, even if they aren't always on top of progressive statistics. They watch film, and if Tennessee really was only average against the run, they'd run more often against Tennessee. Then, since teams ran less frequently against Tenn, Tenn played the run less frequently, and dropped the safeties a little bit further back to help out their pass coverage. As a result, on the rarer times that opponents DID run, they got more yards, because they were catching Tennessee off-guard. This caused DVOA to think they weren't as good on a per-play basis, when really they WERE that good, and that's why they faced so few plays.

I have always found that the process of the NFL trying to find an equilibrium is absolutely fascinating. Offensive rule-changes make it harder for CBs to cover receivers. As a result, defenses blitz more, which causes teams to run more max protection schemes, which leaves fewer WRs running routes, which leaves more WRs double-covered, which leaves passing numbers pretty much right where they were before the rule change.

Re #27: Anybody else take a moment to size them up as a potential couple when they read that? Just me?

I suspect if Gado starts dating Favre's daughter, he stops scoring TDs. Just a hunch.

by Parker (not verified) :: Tue, 11/15/2005 - 8:53pm

But that might not mean he stops scoring!! Nudge, nudge. WInk Wink.


by JonL (not verified) :: Wed, 11/16/2005 - 1:38am

I realized that I'm an idiot, and looked at the WR stat page for rushing plays by receivers. It turns out that only nine WRs have a positive DPAR, and only five (or at least only five of the top ten) have positive DVOAs.

by Matt Weiner (not verified) :: Wed, 11/16/2005 - 10:58pm

28, 20: Nash equilibrium linked in my name.

Basically, if you have two teams picking a strategy, and neither side can do better by switching their strategy (unless the other side also switches its strategy), it's a Nash equilibrium. I think it is in line with what Kibbles said.

It was discovered by John Nash from A Beautiful Mind, though my friends who work on this tell me that the movie got the math wrong.