In this week's Varsity Numbers, Bill Connelly revisits some measures and concepts: Adjusted Scores, Covariance, and momentum (or whatever you choose to call it).
26 Apr 2004
by Aaron Schatz
Two weeks ago, I analyzed the best and worst defenses against opposing number one and number two receivers according to our DVOA statistics for 2003. You can read the beginning of that article for a description of how I put all these numbers together. This week, a look at how teams do against third and fourth receivers, as well as tight ends and running backs (receiving, that is, not rushing). Remember that since DVOA means more scoring, defenses are better when DVOA is more negative.
Before we get to those DVOA numbers, let's answer a question posed in the discussion thread from the first article.
|Opponents Passed to
Top WR Least Often
Opponents Passes to
|Team||% Passes||DVOA||Team||% Passes||DVOA|
As Vince Verhei pointed out in the comments section for the original article, rating defenses based on DVOA against certain receivers has one obvious blind spot. Often times, teams will not even throw towards the top cornerbacks, instead concentrating on second and third receivers and staying away from the best defenders. Since DVOA is a rate stat, there's no allowance for when a team doesn't even face many passes to the opposition's top wideouts.
So, out of curiosity, here is a table of which defenses faced the fewest passes to the opposing team's top wideouts. The list is based on percentage of all passes, not total passes, because otherwise the Ty Law would be listed as one of the most picked-on corners in the league... thanks to New England's opponents often playing from behind. Instead, the Pats are ninth, with 23.9% of opposition passes thrown to the top receivers. The league average is 26%.
Why on earth is Denver on top? The whole point of the Champ Bailey trade was that they needed a shutdown corner, shouldn't teams be taking advantage of this? It's a statistical fluke, something unavoidable when we're playing a little game that involves designating one receiver on each team as numero uno. The teams that weren't throwing a lot to their top receivers against Denver were the teams where the title of number one receiver wasn't that clear: Cleveland, Green Bay, New England, Oakland. Denver faced passes to "number two receivers" more often than any other team. This fluke isn't the explanation for Oakland, however, which at least is known as a good secondary even though they weren't so great in 2003. The other teams on the left had pretty good pass defenses. On the right, you see the effect of Tim "The Human Torch" Wansley for Tampa Bay as well as a lack of respect for rookie Terrance Newman and the Dallas secondary.
And now back to our regular scheduled program. When we left off last week, you might remember that I pointed out that Tennessee had a terrible DVOA against second receivers but a great DVOA against third receivers, while Minnesota was the opposite. Does this indicate some kind of counter-correlation in general between how teams did against second receivers and third receivers? Nope. Just a strange coincidence of these two teams.
Some of Tennessee's rating against third receivers is a bit questionable because it isn't really clear who should be noted as Houston's "second receiver," Corey Bradford or Jabar Gaffney. Gaffney started more often, but Bradford was involved in the offense more. Right now, I have Bradford listed as the "two" and Gaffney as the "three," but they could easily switch. Since Tennessee falters against Bradford (8-for-14, 163 yards) but eats Gaffney for breakfast (1-for-5, 14 yards), this explains part of why they rate very low against number twos and best in the league against number threes.
Explains part, but not all. In one game against the Colts, Tennessee kept Troy Walters to a single eight-yard reception; in the other, they kept Brandon Stokely to 10 yards on six passes. They also kept Carolina's Ricky Proehl to 12 yards on seven passes, and Jacksonville's Cortez Hankton to 11 yards on six passes (over two games).
The Colts appearance high on this list is a bit of a shock, and since they are one of three AFC South teams in the top eight you might just think that this division was filled with poor third receivers. Except, of course, the numbers are already adjusted for strength of opponent. The Colts not only kept opposing third wideouts away from two-thirds of intended passes, best in the league by far -- they also seemed to force a lot of teams to just ignore their third wideouts altogether. James McKnight of Miami and Karl Williams of Tampa each were thrown only one pass, incomplete. Ricky Proehl was thrown only two passes, both incomplete. Most impressive: Drew Bennett, the #4 overall wide receiver by DVOA, was thrown five passes in two games against the Titans, all incomplete. If anyone knows a reason why third wideouts were used so sparingly against Indianapolis, and performed so pathetically, please share with the class.
The worst defense against third receivers, as noted above and in the last article, was Minnesota at +49.6% DVOA. Close behind were the New York Giants and Philadelphia, and then after a gap came Buffalo and Atlanta.
The problem with these stats, of course, is that we're measuring how defenses performed against number three receivers but not how they performed with three receivers on the field. For all we know, Tennessee was having all those problems with number two receivers when the number three receivers were getting the attention. Someday we'll have data on what formations each team is using on each play and we'll be able to measure more accurately, but for now, this is what we have.
Passes to third receivers are a good indicator of which teams like to spread the field with multiple receiver packages, though, and the leader is St. Louis. The Rams threw to wideouts other than Holt and Bruce 85 times (mostly Dane Looker). Other teams which liked to use their third wideout included Oakland (Jerry Porter and Alvin Whitted), Cleveland (Quincy Morgan before Kevin Johnson was released, Andre' Davis afterwards), Buffalo (Bobby Shaw), Chicago (David Terrell), and Seattle (Bobby Engram).
The teams that were least likely to use a third wideout were the teams that used their tight ends the most, with the notable exception of Kansas City. The Giants only threw 23 passes to third wideouts (David Tyree when Hilliard was healthy, Willie Ponder when Hilliard was out and Tyree was the second receiver). Other offenses that didn't use many third wideouts were Baltimore, Atlanta, Denver, and Cincinnati. That last one is somewhat interesting given that the Bengals kept the fantasy football waiver wires jumping with their constant receiver switching in 2002.
What about fourth receivers, and even five wideout sets? Well, as you get further down the wide receiver food chain, the numbers become less and less reliable, and I'm not sure there's much to learn from fourth receivers. The average team threw to fourth receivers about once a game, so there just aren't a lot of numbers. Tampa faced a grand total of four passes to fourth receivers all year. For the curious, the best defenses against fourth wideouts were Baltimore and Kansas City, both at -99.5% DVOA. The worst defense, by far, was Indianapolis. Despite the fact that they did pretty well against third receivers, the Colts had an odd tendency to give up huge receptions to complete unknowns: a 62-yard bomb to Jonathan (Who?) Carter of the Jets, a 16-yard first down pass to Jimmy Redmond of Jacksonville, and two catches, a 19-yard first down and a 31-yard first down, by newly-signed Patriot Dedric Ward.
The top three teams are teams you would expect, teams with overall good pass defense. The next few teams are a bit more unexpected.
Dallas didn't intercept a lot of the passes intended for tight ends, but they did by far the best job at making sure those balls fell incomplete. Two of the three interceptions came in the same game, Week 2 against the Giants. The Giants threw eight passes to Jeremy Shockey, and the Cowboys caught as many of them as Shockey did himself. Shockey had eight yards and a touchdown, but the Cowboys matched that with a touchdown on one of their interceptions. This game gives an example of how touchdowns for individual players can be an overrated statistic. Twice the Giants threw to Shockey on 1st-and-goal on the 1-yard line. He caught one and missed one. If you remember from our article on short-yardage conversions, passes on 1st-and-goal from the one are successful a little over half the time, so Shockey didn't really do anything very special. But to most fans, Shockey's feeble Week 2 performance looks reasonable thanks to six fantasy points.
The Patriots are interesting, because they actually rank third in DVOA against tight ends despite allowing more receiving yards by tight ends than any other team in the NFL. Of course, they also faced more passes to tight ends than any other team in the NFL. Part of their success was allowing only eight first downs on 22 third down passes to tight ends.
Cincinnati's presence on this list is a bit of a puzzle, but those six interceptions have a lot to do with it. There's a lot of random chance there, considering that the Bengal defenders picked off almost as many passes intended for tight ends as they did intended for all other receivers combined (they had only 14 interceptions all season). Two of those interceptions came on passes meant for Shannon Sharpe, and two came on passes meant for Todd Heap (and both in the same game, Week 14).
San Francisco actually allowed fewer receiving yards to tight ends than any other team (395) and would have the fourth-best VOA if we didn't consider the strength of opponents. But San Francisco's schedule was packed with the players at the bottom of our DVOA rankings: Cleveland's Steve Heiden, Tampa's Ken Dilger, Arizona's Freddie Jones.
The worst defense against tight ends was Detroit (+42.5%) followed by Washington, Kansas City, the Giants, and Oakland. Kansas City is the interesting story here. Their terrible rush defense masked the fact that the pass defense was better than league average -- except against tight ends. The Chiefs had a DVOA of -1.9% against all wide receivers and -4.5% against passes to running backs, but they allowed tight ends a DVOA of +30.3%. And, except for Shannon Sharpe, we're not talking great tight ends here. We're talking about Desmond Clark getting two first downs and 34 yards on two passes and 64 yards from an over-the-hill Wesley Walls. The strangest of all were two great games by two different San Diego tight ends. San Diego tight ends??? In Week 1 Josh Norman had 64 yards and a touchdown; in Week 13, Antonio Gates had 49 yards and a touchdown. Total Charger TE touchdowns against Kansas City: two. Total Charger TE touchdowns against the other 12 teams on the schedule: two.
You'll notice that passes to running backs -- and this will shock very few -- are caught more often, intercepted less often, and run for fewer yards compared to passes to receivers and tight ends.
Tampa Bay may have had injury problems in the secondary in 2003 but I'm guessing that this high rating against passes to running backs indicates that the linebackers were just fine. Tampa was particularly stunning if you tried to get a third down conversion with a receiver out of the backfield. 26 times in 2003, teams passes to receivers on third down against Tampa, and only four times did teams get a first down. Incredibly, half the third down passes were complete but the Bucs tackled the back before he reached the yardage needed for a first down..
Buffalo's defense wasn't talked about much in 2003, but they were very good, and one area of strength was this one. Ricky Williams, in two games against Buffalo, was thrown eight passes and totaled two yards. He had three receptions for a loss, two of which were fumbled, and one of those was turned over. He also went 0-for-4 in converting passes to him on third down. The Bills also did a good job of stopping pass-heavy fullbacks like Richie Anderson (only 15 yards) and Jerald Sowell (only 20 yards in two games).
Oddly, the team that did the best job preventing running back catches -- New Orleans -- did a terrible job preventing yards on those catches. The Saints were the only defense to allow complete passes to running backs less than two-thirds of the time, but the DVOA was +14.7% thanks to nine yards per reception.
The worst defense against running back receptions were Houston and Cincinnati (both +36.9%) followed by Washington, Arizona, and San Diego. That's pretty much a list of the worst pass defenses in general, right? Well, the sixth-worst team defending passes to running backs was New England. Yes, we've finally found the Achilles heel of the New England pass defense. There was no one running back who had an amazing receiving day against the Pats, but it all added up. If you need that third down conversion against the stingy New England defense, try swinging it out to the running back.
|Team||DVOA vs. WR||DVOA vs. TE||DVOA vs. RB|
Finally, an interesting note about the NFC North last year. Chicago, Minnesota, and Detroit were all above average against passes to wide receivers, and all below average against passes to both tight ends and running backs. Was it something about the schedule, not only against the rest of the division, but the AFC West and NFC West? Perhaps, except that Green Bay's defense was the opposite -- better against tight ends and running backs than against wide receivers. Now, the other three NFC North teams had to face the Packers, who were very good at using backs in the passing game. But the Packers didn't have particularly high-rating tight ends in 2003, so that explanation can only clarify part of the mystery.