Bill Connelly takes a look at the early-season S&P+ outliers and whether his eyeballs agree or disagree with the stats. Spoiler: Alabama is underrated. Alabama! Underrated!
10 Sep 2003
by Aaron Schatz
This is the seventh of a series of eight articles taking a closer look at every team in the NFL, division by division, using our new statistics such as line yards and DVOA (explained here). Don't be scared away by all the numbers -- the goal of these IN FOCUS articles is to go through all the numbers and translate the most interesting trends into actual English paragraphs for those allergic to endless tables of stats. You enjoy the stats separated by team, or just enjoy the insights of each team's commentary.
The format of these articles is explained at the beginning of the one on the AFC East. Remember, offensive numbers are better the more POSITIVE, defensive numbers are better the more NEGATIVE. Overall total is offense minus defense, so the more positive the better. Schedule strength is harder the more negative, with the hardest schedule ranked #1 and the easiest ranked #32. New players on a team are colored blue and players who have left the team are colored red.
Yes, like the AFC South, the DVOA formula thinks that this division turned out very wrong, with Cincinnati ranking above Cleveland. What's amazing is that Cleveland actually should have been 10-6, if their players knew to keep their helmets on at the end of games. This time, I have a pretty good idea what's wrong here: this is a good indication that I need to figure out special teams DVOA values, which is a project for this upcoming season. The Browns were seriously powered by Mr. Northcutt last year.
OUT: NFC South, AFC South
IN: NFC West, AFC West
HURTS: Both passing and rushing defense
HELPS: Passing offense, to a very large extent, and rushing offense to a small extent
|Stokley, Brandon (IND 03)||-70.2%||6||33||-68.7%||-4.8|
|Stokley, Brandon (IND 03)||-7.1%||65||52||386||-9.0%||70||-4.8||62|
|Sanders, Frank (ARI 02)||9.0%||27||58||429||7.7%||29||6.8||35|
|Robinson, Marcus (CHI 02)||-45.1%||97||53||244||-46.1%||97||-31.3||98|
Other important additions: QB Kyle Boller (R1), DE Terrell Suggs (R1), RB Musa Smith (R3), CB Corey Fuller (CLE), OT Orlando Brown (FA)
Other important losses: LB Shannon Taylor (HOU)
So, here is the problem I am faced with, presenting new statistics like DVOA and line yards. Sometimes, the statistics completely agree with conventional wisdom about players. Sometimes, the statistics completely contradict the conventional wisdom about players. Nowhere is this more true than in the comparison of the Baltimore rush offense and defense.
Notice how the Ravens are ranked #3 in the league in line yards to the left, but are middle of the pack to the middle and right? Baltimore's left tackle is the great Pro Bowler Jonathan Ogden, of course. This statistic definitely agrees with Ogden's greatness.
Flip to the other side of the ball now. Notice how the Ravens are second-to-last in preventing runs to the left? OK, of course defensive players move around more than offensive linemen, and the Ravens run a 3-4, but who are the two linebackers on the right side of the Ravens' defense (meaning the offense's left side). Yes, Ray Lewis and Peter Boulware. Yes, like the statistic that shows that the Redskins pass defense is much better on Fred Smoot's side than on Champ Bailey's side, this seems to make no sense.
What could be the explanation? If you have an idea, please share. Could Lewis and Boulware really not be that great defending the run? Is RDE Adalius Thomas that bad? If line yards don't really reflect quality of the players on that side, how come they so clearly reflect the quality of Ogden?
Another strange note on the Ravens' defense: Here's another highly-regarded defense that wasn't good in POWER situations (third or fourth down, one or two yards to go). Isn't it odd that Tampa Bay, Pittsburgh, and Baltimore were the three worst defenses at stopping first downs in these short-yardage situations?
On the other hand -- and this has Ray Lewis written all over it -- Baltimore was one of the league's best teams at preventing success on passes to running backs.
Baltimore's pass defense was great in the red zone, but below-average at midfield:
Opponent's goal line to opponent's 40: -5% VOA allowed
Between the 40-yard lines: +18% VOA allowed
Baltimore's 20-yard line to 40-yard line: -24% VOA allowed
Red Zone: -50% VOA allowed
For some strange reason, Jamal Lewis has a lot more success catching the ball on the right than on the left, even though he has more success running left than running right.
Chris Redman's DVOA for the first three quarters, believe it or not, was actually above-average: +8% DVOA. In the fourth quarter, he divebombed to a -57% DVOA.
|Westbrook, Michael (FA 03)||-45.0%||26||94||-45.7%||-15.5|
|Kelly, Reggie (ATL 02)||7.1%||14||20||162||5.8%||16||2.0||16|
Other important additions: QB Carson Palmer (R1), CB Tory James (OAK), LB Kevin Hardy (DAL), DT John Thornton (TEN), K Shayne Graham (CAR), LG Eric Steinbach (R2), WR Kelley Washington (R3)
Other important losses: OLB Takeo Spikes (BUF), SS Cory Hall (ATL), FB Nick Luchey (GNB), K Neil Rackers (FA), OT Richmond Webb (PIT)
The Cincinnati Bengals defense is a good example of how our VOA statistics measure defensive ability better than the usual NFL rankings, by taking into account what situations each team faced.
According to our rankings, Cincinnati ranked #31 in defensive efficiency in 2002. According to NFL rankings, Cincinnati ranked #17 in total defense in 2002. Why the difference?
First of all, the NFL rankings are based on yards. But they don't take into account that the more plays a defense is on the field, the more yards it allows. Cincinnati was on the field for 976 plays in 2002, one of the lowest totals in the NFL. Sort the official NFL stats by yards per play, and the Bengals move down to #24 in the NFL. But there are more adjustments than just number of plays; there are also the types of plays. Here are two examples why the NFL official stats say the Bengals defense was better than it really was.
49% of third down plays in the NFL last year were third-and-long, defined here as needing seven or more yards for the first down. Because the Bengal defense was so bad, only 40% of third down plays against Cincinnati were third-and-long. So a third down pass for 4-6 yards against Cincinnati was probably a success, gaining a third down. A similar play against an average team was less likely to be a success -- even though the defense allowed the same yardage -- because the offense was more likely to fall short of the first down. The Bengals were giving up the same amount of yards, but more first downs.
Offenses average fewer yards per play when they are winning, especially in the second half. In 2002, during the second half of games, the average NFL offense gained 5.3 yards per carry. If that team was winning by more than a touchdown, the offense gained 5.1 yards per carry. If that team was winning by more than a touchdown and playing the Bengals, the offense gained only 4.7 yards per carry. And how often in the second half were the Bengals playing down by more than a touchdown? Over half the time, compared to only 21% of the time for the NFL as a whole. So the Bengals defense gets credit for all those extra plays where the defense allowed fewer yards because of opponents running out the clock -- credit that, frankly, it doesn't deserve.
Are you as surprised by am at the low number of double-digit yard runs by Corey Dillon?
While Dillon gets better on third down, Jon Kitna gets worse. Too bad the Bengals face more third-and-longs than they do third-and-shorts:
|First Down||Second Down||Third Down|
|C. Dillon rushing DVOA||-2%||+7%||+50%|
|J. Kitna passing DVOA||+32%||+16%||-20%|
Chad Johnson was thrown twice as many red zone passes as any other Bengal, but he had negative value in the red zone: -61% DVOA, compared to +20% DVOA on the rest of the field. Can you say "double coverage?"
One possibly underrated Bengal is cornerback Jeff Burris. The Bengal pass defense allowed only +16% VOA on passes to his side (right) as opposed to +48% VOA on passes to the left and middle.
Cincinnati's pass defense: Very good -26% VOA allowed in the red zone, crummy +28% VOA otherwise.
|Campbell, Mark (BUF 03)||-35.1%||48||46||185||-33.8%||48||-20.7||46|
Other important additions: OLB Barry Gardner (PHI), C Jeff Faine (R1), LB Chaun Thompson (R2), RB Lee Suggs (R4)
Other important losses: OLB Dwayne Rudd (TAM), C Dave Wohlabaugh (STL), LB Earl Holmes (CLE), LB Jamir Miller (FA), CB Corey Fuller (BAL)
I bought William Green in our fantasy auction last year, expecting to get the next Jamal Lewis. Instead, I got the next Ron Dayne. Around mid-year, I dropped him. Then, in week 11, Green made Browns fans giddy and yours truly frustrated by suddenly breaking out of his shell. Inserted into the starting lineup, Green immediately went from pathetic to, if not Portisesque, at least above average. Here are his adjusted line yards and DVOA numbers:
|ALY left||ALY middle||ALY right||DVOA|
|W. Green weeks 1-10||2.25||2.26||2.76||-55%|
|W. Green weeks 11-17||3.48||3.18||4.20||+7%|
By the way, that #30 ranking running up the middle is reflected in the numbers for both Green and Jamel White. It's an adjusted number, meaning that it already takes into account that runs up the middle average fewer yards per carry than runs to the side. This could be a good reason to spend your first round pick on a new center. Be wary, Rams fans, because 2002 Browns center Dave Wohlabaugh is now anchoring your line.
Yes, now that we've fixed the Tampa black hole, Holcomb rates higher than Couch, but he still doesn't come out above average. Interestingly, both Holcomb and Couch were far better than an average quarterback with their team down by more than a touchdown:
|Passing DVOA||T. Couch||K. Holcomb|
|Browns losing by 8 or more||+11%||+75%|
|Game within a touchdown||-24%||-20%|
|Browns ahead by 8 or more||-112%||+3%|
Of course, Couch only threw four percent of his passes last year with the Browns ahead by 8 or more, so that's a very small sample size, but in that small sample size he was awful.
Strange trend: Kelly Holcomb's passing value was +52% DVOA in the second quarter but -23% DVOA in the other three quarters.
Take a second to appreciate the work -- even though it was on a smaller number of passes than most top receivers -- of Dennis Northcutt. He caught more than his share of long passes, and hardly ever missed a pass thrown his way. 80% of passes thrown to Northcutt were complete, compared to only 55% of passes thrown to all other Browns receivers.
Jamel White had the best receiving DVOA in the league (+86% DVOA) of any player thrown more than 10 passes in the DEEP zone (his own goal line to 20-yard line).
Cleveland was a bad team to run against if you needed that one last yard, and a good team to run against if you wanted a long run of a double-digit yards. Only Tennessee was better in POWER situations, third or fourth and very short, and only Minnesota allowed more rushing yards over ten yards past the line of scrimmage.
Speaking of those last important yards, the Cleveland run defense was good in the red zone -- and the pass defense was better. In fact, only Green Bay allowed less passing success in the red zone:
|VOA against||Cleveland pass defense||Cleveland run defense|
|Opponent's half of field||-4%||+10%|
|Midfield to Cleveland 20-yard line||+22%||+1%|
Only Tampa Bay and Seattle were better than Cleveland at preventing success on passes to tight ends.
|Randle El, Antwaan||-14.9%||10||37||-9.8%||-1.5|
|Fuamatu-Ma'afala, Chris (FA 03)||25.8%||23||115||22.4%||5.2|
|Randle El, Antwaan||11.3%||19||134||11.7%||2.9|
|Fuamatu-Ma'afala, Chris (FA 03)||-76.1%||3||12||-80.2%||-1.6|
|Randle El, Antwaan||-7.4%||66||65||544||-7.0%||65||-6.3||67|
|Doering, Chris (WAS 02)||-25.5%||91||42||199||-28.2%||92||-13.3||77|
|Mathis, Terance (FA 03)||-9.4%||70||40||289||-8.6%||69||-4.9||63|
|Riemersma, Jay (BUF 02)||18.6%||6||48||377||20.2%||6||11.9||5|
Other important additions: LB Clint Kriewaldt (DET), OT Richmond Webb (CIN), S Troy Polamalu (R1), DE Alonzo Jackson (R2)
Other important losses: S Lee Flowers (DEN), LT Wayne Gandy (NOR),
We always thought the XFL was a good idea (a spring football alternative, if not the Vince McMahon part) so it was nice to see Tommy Maddox have so much success at the helm of the Steelers in 2002. Now, with the 2003 season about to start, people want to know if Maddox is a one-year wonder. Looking at the past, there's no reason to think so.
Using Doug Drinen's awesome pro-football-reference.com database, I went back to find every NFL player since the merger who had missed two or more years and returned to the league. I used fantasy points as a guide, picking out players who had a layoff of at least two years between seasons of 50 fantasy points (according to the formula here). There were seven quarterbacks -- including Maddox -- and two running backs. Most are not very good comparisons to Maddox, but the one who is would seem to indicate continued success: Doug Flutie.
Flutie of course returned to the NFL in 1998 after eight years in Canada. Like Maddox, Flutie was the best player of a secondary football league -- albeit one with a bit more stability -- and like Maddox, Flutie ended up supplanting the player he was signed to backup. Flutie was great again in 1999, had limited playing time in 2000, and then spent 2001 starting in San Diego before moving back to the bench.
The others in alphabetical order:
Vince Evans was the Raiders quarterback during the 1987 strike games, and proved enough to Al Davis that he stuck around as a backup until 1995. Not really a good comparison to Maddox.
They play different positions but Maddox would love his comeback to be as good as Garrison Hearst's. Hearst, of course, spent four seasons as a starter with a breakout year in 1998, but then lost 1999 and 2000 to injury. He returned in 2001 as part of football's best running-back-by-committee and shows no signs of letdown.
Running back and return man Terry Metcalf (father of Eric) averaged over 1000 yards per season combined running and catching the ball from 1973-1977, missed three years, came back in 1981 as a third-down back for the Redskins with 595 yards receiving and 60 yards rushing, and then retired. Not a good comparison to Maddox.
Billy Joe Tolliver was in and out of the starting lineup in San Diego and Atlanta from 1989-1994, didn't play in 1995 and 1996, and then showed up again in 1997 before starting about half of New Orleans' games in 1998 and 1999. Not a good comparison, since he wasn't really gone, just stapled to the bench. I'm still working on the value of a replacement player in football, but Billy Joe Tolliver is pretty much it.
Doug Williams spent 1978-1982 in Tampa Bay and then disappeared for three years before re-appearing on the Washington bench in 1986. Then in 1987 he took over after a midseason injury to Jay Schroeder and led his team to a Super Bowl victory. He had a good half-year in 1988 but got injured and lost his job to Mark Rypien. This is Maddox's worst case scenario.
Danny Wuerffel was mediocre from 1997-1999, was mediocre on the bench in 2000 and 2001, and then returned to on-field mediocrity in 2002. Not a good comparison to Maddox.
The Steelers have replaced Jerome Bettis as the starter with Amos Zereoue, but they should still go to Bettis in the red zone. Bettis had a +37% DVOA in the red zone, while Zereoue had a -40% DVOA rushing in the red zone and an astonishingly bad -269% DVOA receiving. Outside the red zone,