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).
13 Jan 2012
by Aaron Schatz (DEN-NE) and Rivers McCown (HOU-BAL)
The AFC playoffs give us two games that are expected to go very differently. Denver-New England is expected to be a shootout. Baltimore-Houston is expected to be a defensive struggle. Both games are rematches. Our stats see one game as a lot closer than the other.
For those who may be unfamiliar with the Football Outsiders stats, they are explained at the bottom of the page. Scroll down or click this link. Please remember that all stats represent regular season only, except for weighted DVOA and anything else specifically noted. Game charting data is still incomplete, but represents most of Weeks 1-16.
All readers can click here for in-game discussion on our message boards. If you have FO Premium, you can click here to see all the matchup of DVOA splits for this game.
If Denver wins this weekend, it will be the third-biggest playoff upset in DVOA history. The two games that were bigger upsets by DVOA standards:
There are three games with a bigger DVOA gap that wouldn't qualify as bigger upsets because the better team was on the road. The first is last week's Denver win over Pittsburgh. The other two are last year's New Orleans-Seattle game, and Arizona's win over Philadelphia in the 2008 NFC Championship game.
And yet... even with this potentially being an upset that historic, there's a reasonable chance of Denver winning. Fans always underestimate the possibility of upsets in the NFL. They want to talk about "this team will definitely beat that team." There are no definites. Even with New England at home, even with New England being so much better than Denver, there's probably a one in six chance of Denver winning this game. That sounds like a big chance, but compared to most playoff matchups, it really is not.
Denver's offensive advantage over the Patriots pretty much comes just on first downs. Denver is 15th in DVOA on first down, New England is 30th. The two teams are much closer on second down (Denver 23rd, New England 16th) and third down (Denver 29th, New England 30th). That first-down advantage isn't just about running the ball; it's also about passing. Denver is 18th in DVOA passing on first down, but 30th in passing on second down and third down. Of course, Denver rarely passes on first down. Since Tim Tebow became the starting quarterback, Denver has run the ball on 65.3 percent of first downs.
The Broncos ran the ball all over the Patriots in the first quarter of their Week 15 game. It's very, very unlikely that the Broncos can run for 167 yards on 15 plays again. Things slowed down in the second and third quarter, and the Broncos ran 12 times for 48 yards. In the fourth quarter, with the Pats playing prevent up by 18, the Broncos gained another 47 yards on four carries, mostly on a 30-yard scramble by Tebow. It also helps the run defense that safety Patrick Chung and especially linebacker Brandon Spikes are back; both missed the first Denver-New England game.
One of the interesting outcomes in the advanced stats this year is that Tebow's running wasn't all that efficient. Tebow actually had a below-average -16.8% DVOA rushing. He had serious butterfingers, with six fumbles on running plays. He also gained a lot of yards on failed third-down scrambles. Tebow had 18 gains of five yards or more on third down. Only nine of those runs actually moved the chains.
Tebow is scrambling on third down because he has to drop back to pass on third down. But he may not be scrambling quite so much in this game, because he should be able to get passing yards against the porous Patriots secondary. However, the Pats did improve their defense after halftime by going to a primarily man scheme against Tebow. That's what most defenses have played against Tebow this year. Like most of those defenses -- and unlike Pittsburgh last week -- the Patriots are likely to keep a safety back deep just in case Tebow does hit one of his frequent deep passes. As Vince Verhei pointed out in Any Given Sunday this week, Tebow threw 17.3 percent of his passes this year at "bomb" length (over 25 yards). That was by far the most of any regular starting quarterback. And the Patriots allowed 21 successes on such passes, either completions or DPI calls. No other team allowed more than 16. In the first game between these teams, to try to prevent these big plays, the Pats always had a safety back -- and Tebow didn't try a pass that went more than 20 yards downfield until there was 5:26 left in the third quarter.
It will be interesting to see how the Denver formations change with Eric Decker out, because Tebow and Eddie Royal really weren't on the same page this year. Royal had a terrible year, dead last in DVOA among WR with at least 50 passes at -44.2%. But he was even worse once Tebow became the quarterback:
|Denver WR with Tim Tebow at QB|
You don't often see the guy who runs the shorter routes as the guy with the really low catch rate. In the first Patriots-Broncos game, Tebow only threw to Royal twice, both incomplete. Oddly, he also went the entire game against New England without throwing to the guys who are probably his favorite receivers: his tight ends. After Tebow took over in Week 7, Daniel Fells and Dante Rosario combined for a 19.2% DVOA, 65 percent catch rate, and 10.2 yards per pass. The Pats were 29th in DVOA against tight ends, although they really were no worse against tight ends than they were against wide receivers. (They were a surpringly strong fourth in DVOA against running backs as receivers.)
Cornerback Devin McCourty, who generally lines up on the offensive right, had an absolutely dismal sophomore campaign. Our charting has him with 10.7 yards per play and a 45 percent Success Rate. The yards per play figure is one of the ten worst in the league for starting corners, and the Success Rate is pretty bad too. Kyle Arrington, on the offensive left, was a little better: 8.1 yards per play and a 50 percent Success Rate. Frankly, whichever one is covering Demaryius Thomas on a given play is in trouble. Tight ends are generally played by a variety of safeties and linebackers, none of whom (except Chung) is particularly strong in coverage.
One more surprising stat from game charting: During the regular season, at least, play-action fakes didn't seem to open up the passing game for Tim Tebow the way you might expect. Denver averaged 5.1 yards per play with play action, 5.0 yards per play otherwise -- virtually the same, where the league average is for teams to average about 1.2 more yards per play with play action.
There are two secrets about the Denver defense. First, the Denver defense is very consistent. The Denver defense actually led the league with the lowest defensive variance. That's not necessarily a good thing because of the other secret: Denver's defense is also really mediocre.
When Tim Tebow was leading the team on that six-game winning streak, there was a lot of talk about how an improved defense was a big part of the wins. Well, according to DVOA, the defense really didn't improve much. Denver's defensive DVOA in Weeks 1-9 was 7.1%, 17th in the NFL. Denver's defensive DVOA in Weeks 10-17 was 5.7%, 18th in the NFL. Neither the pass defense nor the run defense improved much in the second half of the season.
(Yes, Denver had five games where they gave up 13 or fewer points. They also had three games where they gave up more than 30 points. It seems inconsistent, but it really isn't. In the games where they gave up fewer points, they generally gave up more yards than you would otherwise expect. And in two of those games where they gave up lots of points, defensive DVOA isn't much higher than 0%. Buffalo scored two defensive touchdowns and a special teams touchdown, while the New England game has big opponent adjustments.)
New England is an even worse matchup for the Broncos defense because the Patriots' biggest strength matches Denver's weakness. The Broncos' run defense (13th) was better than their pass defense (24th), but the Patriots aren't likely worried about that because they don't run that much (and when they do run, they run from spread formations that suggest pass). The strength of the Broncos' pass defense was the pass rush, which ranked fifth in Adjusted Sack Rate. They'll take Tom Brady down a couple times -- they took him down twice in the first matchup -- but that probably won't slow him down much.
But the weakness of the Denver defense is coverage. Yes, Champ Bailey is still back there, but he's not the Champ Bailey of old. Don't be fooled by the idea that Bailey shut down Mike Wallace last week; Ben Roethlisberger's foot did a lot of the work for him. Bailey has 55 percent Success Rate and 7.9 yards per pass in our game charting compiled so far. Those numbers are about average. Andre' Goodman is worse in Success Rate (50 percent) but better in yards per pass (6.4). The real problem isn't Bailey or Goodman but the young safeties, especially with veteran Brian Dawkins injured. Denver ranked 31st in DVOA against passes to the "deep middle" section of the field, and 29th in DVOA against passes to the "short middle" section of the field. Where do you normally find the New England tight ends? Yep, that's right. The Patriots threw 32 percent of passes to the middle of the field, third in the league. In Week 15, Brady was 8-of-9 for 153 yards on passes to the middle of the field.
The Broncos defense was reasonably good on third down this year (eighth in DVOA) but first and second down are where Tom Brady can kill them. The Patriots had the best offensive DVOA in the league on both first and second downs. The Denver defense was 21st on first down and 27th on second down -- including dead last on passes on second down.
The Broncos passing game of course is based off a lot of play-action because they run so much, but play-action could really benefit the Patriots in this matchup as well. Based on our game charting so far, New England gains 3.5 yards more per play when they use play-action, the third-highest figure in the league. And the Broncos defense allows 3.4 yards more per play against play-action, the second-highest figure in the league.
The Broncos big-blitzed (six or more) on 20 percent of pass plays, more often than any team in the league except for New Orleans. However, big-blitzing Tom Brady means asking for a big play. Based on this year's charting numbers compiled so far, blitzing Brady does give the defense a better Success Rate: 43 percent with a normal rush, 49 percent with five men or more rushing. And Brady actually is gaining slightly fewer yards per play against five pass rushers compared to four. But against six or more, he's gaining 11.1 yards per play, second only to Eli Manning.
If the Patriots do want to run, ALY numbers suggest they want to run outside. However, Denver's strong run defense on runs up the middle comes mainly from standard running plays, not the spread shotgun-type plays that the Patriots often run. The Broncos allow 3.4 ALY per carry on runs up the middle from standard formations, but 4.4 ALY per carry on runs up the middle from shotgun formations.
New England has the special teams advantage, but it is mainly in one area: kickoffs. You'll often hear announcers talk about how Matt Prater has a strong leg, but he really doesn't. He just happens to kick half his games in altitude. Adjusted for altitude, he was average on kickoffs and below average on field goals. Stephen Gostkowski does have a strong leg to boom kickoffs, and the Patriots had excellent kickoff coverage. The Patriots also got excellent punting from Zoltan Mesko. The Broncos' punt return numbers are interesting: main punt returner Quan Cosby had negative value, but the Broncos' total punt return figure is high because of good punt returns from Eddie Royal and Eric Decker. Cosby was waived after the first New England game, so Royal will be returning punts on Saturday night.
The miracle in this game won't be if Tim Tebow can move the ball on the Patriots defense. The miracle will be if Tim Tebow can come out and cover Rob Gronkowski or Aaron Hernandez. It's unlikely that Tebow will throw the ball as well as he did last week, but against the Patriots defense, he should be able to move the ball on the ground and in the air. It will be enough to make Denver look like a competent NFL offense. It will probably not be enough to outscore Tom Brady.
All readers can click here for in-game discussion on our message boards. If you have FO Premium, you can click here to see all the matchup of DVOA splits for this game.
These teams clashed for the first time in Week 6, where the Ravens let the Texans hang around before finally putting them away, 29-14, by scoring 13 unanswered points in the fourth quarter. That Houston squad had Matt Schaub, but did not suit up Andre Johnson or James Casey. Baltimore was missing a key player as well in offensive lineman Ben Grubbs -- the Ravens noticeably struggled to protect Joe Flacco when he was out. They'll also reclaim some of their secondary depth, since Jimmy Smith and Tom Zbikowski are expected to suit up this time around. On paper, these two teams have incredibly similar styles -- how will that translate on the field? Gentlemen, prepare your play-action passes.
This one is pretty simple: the Texans are a run-first football team, and the Ravens have the best pass defense in the NFL by our metrics this year. As long as Houston doesn't dig themselves a big hole, you're sure to see a lot of Arian Foster and Ben Tate. Foster was worth -19 DYAR on the ground against the Ravens in Week 6, but he was still coming back from a hamstring injury at that point, and subjectively appeared to leave a lot of yards on the field. From his season debut in Week 2 to Week 6, he compiled just - 7 DYAR. Since then? Foster has accumulated 148 DYAR in 9 games, a pace that would have placed him as the second-best runner in the league this year, behind only LeSean McCoy. Oh, and he acquitted himself quite nicely against Cincinnati last week, in case you missed that game.
Here's some fun with stereotypes: The Texans offensive line has been exceptional this year, but as a unit they rely more on technique and finesse. It would seem like they would struggle to get much push against physically imposing players like Haloti Ngata and Terrence Cody -- except the Ravens' defensive line was in the bottom eleven in both power success percentage and stuffed percentage. Ray Lewis is still one of the best in the business going downhill, but it seems like he can sometimes struggle to get off blocks -- except that the Ravens finished second in both open-field and second-level yards allowed.
Baltimore is markedly worse on their right side by Adjusted Line Yards (23rd in the league on runs to right end, 14th to right tackle), and the Texans run at that side more frequently (33 percent compared to 26 percent) and more effectively (ninth in ALY to the right end as compared to 23rd on runs to left end) than they do to the left. Or, to sum up those statistics in five words: Eric Winston is pretty good. As Ben Muth noted in yesterday's column, Andre Johnson's return also means a lot to the Texans in the blocking game. They'll have more scheme diversity because of his versatility. One final note: Ed Reed made a couple of brilliant run reads the first time these teams met, so it'll be interesting to see if Houston can disguise themselves to keep him from reading the play at the snap.
So then, that leaves T.J. Yates to be thrown to the
lions Ravens pass rush, right? Well, not exactly. It's probably safe to rule out any sort of dominant air performance from Yates, but the Texans ran play-action on 33 percent of all passes, gaining 4.7 yards more per pass than they did on normal passes. Both of those figures led the NFL. Baltimore was below-average at defending the play-action pass, giving up 1.7 more yards per attempt against it than they did against normal passes (league average was about 1.2 yards, as noted above).
However, of the 41 dropbacks Houston had in Week 6 against the Ravens, Matt Schaub was sacked on four occasions and hurried on nine others. The Ravens zone-blitzed the Texans five times according to our charting stats, and on those plays, Schaub was 2-of-3 for 8 yards, taking two sacks. Houston's offensive line traditionally does not do very well against this look, and in a hostile environment, with a rookie quarterback who has shown some troubles accounting for blitzers at the line of scrimmage, it'll probably be dialed up a little more often than usual.
The Ravens were better than average against every type of receiver by DVOA, but they were closest to average against the No. 1 (-4.0% DVOA, ninth) and No. 2 (-4.1% DVOA, 13th) receivers, which jives with the notion that they don't really have an ace cornerback at this point. It'll be mighty interesting to see how much respect the Ravens pay Johnson as they try to prevent the big pass plays that could change this game.
Many a night I've spent this week trying to find the Rosetta Stone that tells us why Joe Flacco is so inconsistent. He makes a few killer throws every week, but his deep ball comes and goes. Out of his sixteen games this season, he has had a DVOA above 20% eight times, and he's had a negative DVOA rating seven times. I had two working theories on Flacco's diverse range of effectiveness: one was that perhaps, like some quarterbacks, he didn't respond well under pressure. The other was that due to the importance of the deep play-action pass in Cam Cameron's system, maybe his effectiveness depended solely on how often he hit those deep balls,
Turns out, neither of those things really matter. Flacco had four games where he was knocked down (via quarterback hit or sack) seven or more times. In three of them, including Week 6 against Houston, he posted a DVOA above 30%. How about the deep passes? While that correlated a little better, he had five passes of 20 or more yards against Arizona in Week 8 and wound up with a -12.2% DVOA for that game. Against the Texans in Week 6, he completed just two passes of that length, and wound up with a 37.6% DVOA, and against the 49ers in Week 12, he completed just one, yet had a 84.1% DVOA. Some players just won't fit into a neat little box. Such is this year's vintage of Flacco -- we're not sure if we're getting a great quarterback or a below-average quarterback, but it's definitely one of them.
We covered the Houston pass rush last week, and even with Ben Grubbs back in the lineup, they are likely to get some pressure on Flacco anyway. However, part of the reason that Houston's pass rush finished third in adjusted sack rate is that between J.J. Watt and Antonio Smith, they can get pressure inside just as easily as they can outside. Baltimore has one of the best interior line trios in the league with Matt Birk, Grubbs, and Pro Bowler Marshal Yanda -- that could very well be the battleground that determines the winner of this game. Rookie deep threat Torrey Smith is also an important player, especially given that Anquan Boldin will likely draw Johnathan Joseph's attention in coverage. Given their respective strengths in a potential Smith matchup, it would be prudent for the Texans to take Jason Allen's deep speed over Kareem Jackson's underneath coverage. Of course, nothing about Jackson continuing to get playing time actually makes any sense, so it's probably a safe bet that Flacco will have a couple of open deep balls to Smith at some point in the game. In the first game, Flacco hit Smith for 51 yards with Jackson in coverage.
Baltimore has a very good run game in its own right, with Ray Rice being guided to the hole by long-time Texans fullback Vonta Leach. They don't have quite the versatility in the backfield that Houston has, but they have a lot of leverage to exert between Leach and those three maulers in the middle. Despite his 101 yards, Rice actually had -7 DYAR against Houston in Week 6, mainly because 58 of those yards came on three carries.
While Rice will sometimes bounce things outside, look for the overwhelming majority of Baltimore runs in this game to go straight up the gut. Nose tackle Shaun Cody has been serviceable this season, but that hasn't kept teams from running up the middle on Houston a league-leading 67 percent of the time. Houston also allows 4.18 ALY in the middle, their worst in any direction. Baltimore hasn't been especially inclined to pound the ball between the tackles this season, but 15 of Rice's 23 carries in Week 6 went there.
Here's the one part of this game where a team has a distinct advantage: the Ravens finished 30th in our special teams rankings, and were one of two teams (along with Detroit) to not be above-average in any individual special teams category. They've had to change return men a few times this season, Billy Cundiff has dealt with a nagging calf issue since December, and the kick and punt coverage teams have combined to cost them around 15 points this year. Of course, the Texans aren't particularly likely to be the beneficiary of a random event like David Reed's two-fumble game, but with Jacoby Jones having less to do on offense and Danieal Manning returning kickoffs, they should get some edge in field position.
Given that this looks like it will be a reasonably cold game, neither team should have much faith in their specialists. Cundiff was 8-of-15 from beyond 40 yards, and Neil Rackers was 8-of-13 despite playing half his games indoors. It could create some interesting fourth-and-medium decisions for the coaches, especially since both defenses have been so stellar this year.
Think of this game as a mathematical equation that you can balance. The running games (slight edge to Houston) and the defenses (slight edge to Baltimore) cancel each other out. You're left with the Texans special teams and Yates against Flacco.
The Ravens deserve to be favored in this game, both because they are at home and because Flacco has shown that he can dominate a game. Ultimately though, the fate of this game probably rests entirely on which version of Flacco shows up -- not something that's entirely easy to predict, as you might have gathered. Given how much better Yates has looked with Johnson on the field, my gut instinct is to say that Houston is probably going to cover that big spread. The game itself is too close to call.
DVOA (Defense-adjusted Value Over Average) breaks down each play of the season and compares it to the NFL average based on situation and opponent. You'll find it explained further here. Since DVOA measures ability to score, a negative DVOA indicates a better defense and worse offense, and a positive DVOA indicates a better offense and worse defense.
SPECIAL TEAMS numbers are different; they represent value in points of extra field position gained compared to NFL average. Field goal rating represents points scored compared to average kicker at same distances. All special teams numbers are adjusted by weather and altitude; the total is then translated into DVOA so it can be compared to offense and defense. Those numbers are explained here.
Each team is listed with DVOA for offense and defense, total along with rush and pass, and rank among the 32 teams in parentheses. (If the DVOA values are difficult to understand, it is easy to just look at the ranks.) We also list red zone DVOA and WEIGHTED DVOA (WEI DVOA), which is based on a formula which drops the value of games early in the season to get a better idea of how teams are playing now (explained here).
Each team also gets a chart showing their performance this year, game-by-game, according to total DVOA. In addition to a line showing each game, another line shows the team's trend for the season, using a rolling average of the last five games. Note that even though the chart appears in the section for when each team has the ball, it represents total performance, not just offense.
37 comments, Last at 15 Jan 2012, 9:42pm by t.d.