In this week's Varsity Numbers, Bill Connelly takes a page out of baseball's playbook and attempts to isolate power from efficiency.
25 Jul 2005
by Aaron Schatz
For the past few months, five years worth of DVOA have been available on our website: 2000-2004. But since the end of last season, you may have noticed me occasionally mention ratings from 1998 and 1999. I was not able to fully analyze those seasons until this past off-season, but now we can unveil the 1998 and 1999 DVOA ratings for each team. We'll start with 1998, and do 1999 in a couple weeks.
In 1998, just like last year, a few teams really stood out, with Atlanta and Denver at 14-2 and Minnesota at 15-1. Which one of those teams topped the league in DVOA?
Surprisingly, none of them did. According to DVOA, the best team over the course of 1998 was the 12-4 New York Jets.
But wait, there's more. The 1998 Vikings, who set an all-time NFL record for offense by scoring 556 points, were not the number one offense according to DVOA. They were ranked third, behind Denver and San Francisco.
1998 also brought us one of the greatest fluke teams of all time, the 1998 Arizona Cardinals, who went 9-7 despite getting outscored by their opponents 378-325. They rank 25th in DVOA below three different 4-12 teams. One of those teams is the St. Louis Rams, so we can finally answer the question: Was there any indication that the Rams were going to burst from out of nowhere to win the next year's Super Bowl?
Before we discuss 1998 further, let's show the numbers. These are the Football Outsiders team efficiency ratings for 1998, measured by our proprietary Value Over Average (VOA) system that breaks down every single play and compares a team's performance to the league averaged based on situation in order to determine value over average. (Explained further here.)
DVOA represents adjusted statistics. OFFENSE and DEFENSE DVOA are adjusted for opponent quality and to consider all fumbles, kept or lost, as equal value. SPECIAL DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. NON-ADJ TOTAL VOA does not include these adjustments. DVOA is a better indicator of team quality. VOA is a better indicator of actual wins. WEIGHTED DVOA gives a stronger consideration to games late in the season. Remember that, as always, defense is better when it is NEGATIVE.
The Jets are on top of the ratings thanks to adjustments for strength of schedule. The two teams that met in the Super Bowl are actually one-two in non-adjusted VOA. On one hand, this could be an indication that the strength of schedule adjustments in DVOA are too strong. On the other hand, four of the top ten teams of 1998 came from the AFC East. But these were actually not the teams that beat the Jets. The Jets went 7-1 in division games, with the one loss coming by just one point, 24-23 to rookie Peyton Manning and the Colts in Week 11. That game deserves a special mention: The Jets let Manning march the Colts 80 yards in the final three minutes for the winning touchdown, including an 18-yard pass to Marshall Faulk on 4th-and-15, and then Vinny Testaverde fumbled a snap with 11 seconds left at midfield. Yikes.
But I digress. The other Jets losses came to San Francisco (5), St. Louis (19), and Baltimore (23). A lot of their rating comes from a mid-season dismantling of Atlanta, 28-3, in a game where Chris Chandler was injured and the Falcons had 97-year-old Steve Deberg and future Arena hero Tony Graziani at quarterback. (Deberg had not played a down since 1993!) The Jets were beating good teams and losing to bad teams, but those losses came early. They started out 2-3 before finishing the year with 10 of 11 wins, the Colts game mentioned above being the only loss. They were well-balanced, finishing in the top five on both offense and defense. But you have to look at the ratings with the knowledge of how much that Week 8 game affected both the Jets and Falcons. Remove Week 8, and Atlanta is now the top team in the league with a 35.0% DVOA, while the Jets drop to second with a 31.9% DVOA.
OK, you say, if we're going to consider Atlanta to be a better team than DVOA says because they had to play a game without Chris Chandler, why don't we give credit to the Broncos for the games they had to play without John Elway. Elway missed three games in his final season, and played sparingly in two others. If you only consider the games Elway played, wouldn't Denver rank as the best team in the league?
Actually, no, because guess who ranks as the best passer of 1998, according to value per play (DVOA) rather than total value (DPAR, or Defense-adjusted Points Above Replacement, which is explained here). Yes, that's correct, it is Bubby Brister, with 44.4% DVOA on 140 pass attempts. Elway is third, with Vinny Testaverde in between. So the Broncos didn't lose anything in the games where Elway was on the sidelines. They won 13 straight games with a mix of Elway and Brister behind center, but their two losses against the Giants and Dolphins were both games with Elway healthy. (The fact that those two losses came in Weeks 15 and 16 explains why the Broncos rank just fourth in WEIGHTED DVOA.)
More important than the quarterbacks was Terrell Davis, who dominated running backs in 1998. His 65.7 DPAR rushing was almost twice as valuable as the second-best running back of the year, Jamal Anderson, who had 34.3 DPAR. Davis also had a 23.2% DVOA, making him the only running back with at least 120 carries and a DVOA rating above 10%! (Compared to recent seasons, 1998 was a tough year to be a running back.) Combine the great passing with the great running and you get the best offense in the league -- even better than the 556-point Vikings. Strength of schedule was part of this: Denver's faced the sixth-hardest schedule of opposing defenses in the league, while the Vikings ranked 21st in average defensive opponent. Remove opponent adjustments and the two offenses are tied at 29.1% VOA.
The other issue is long plays. The Vikings led the league with 52 plays of 25+ yards. They had 22 offensive plays of 40+ yards; no other team had more than 16 plays of that length. Long plays like this are discounted in DVOA, because as plays get longer and longer they actually have a diminishing value when it comes to determining future performance. Often the difference between an 80-yard play and a 40-yard play isn't that the first offense is better; the difference is that the first team was on its own 20 and the second team was on the opponent's 40. But with a high-powered offense led by some rookie receiver named Randy Moss, perhaps the 1998 Vikings were the exception to this rule.
Denver's defensive ranking is worth a little discussion as well. How does a team that ranks 8th in points allowed and 11th in yards allowed rank a dismal 24th in defensive DVOA? Well, this is what happens when you play in a division with two of the more unbalanced teams in league history. We all know about the 1998 Chargers, with a dismal offense quarterbacked by Craig Whelihan and Ryan Leaf, 29th in the league in offensive yards per play but second in the league in defensive yards allowed per play thanks to a defense featuring Pro Bowlers Junior Seau and Rodney Harrison, for whom we have the utmost respect. Fewer people remember the 1998 Raiders, the first Jon Gruden team, with an offense led for most of the year by the immortal Donald Hollas. I mentioned this team two weeks ago when discussing the Dolphins -- they went 8-8 despite getting outscored by their opponents 356-288, and the next year they brought in Rich Gannon and suddenly had one of the best offenses in the league.
The Broncos also didn't play any hard offenses from outside their division. The only top ten DVOA offenses they faced were #6 Dallas and #9 Jacksonville.
The 1998 Arizona Cardinals are discussed in depth in an essay in Pro Football Prospectus 2005, so I don't want to write too much about them here. But they were a colossal fluke, despite the fact that they managed to win a playoff game in a huge upset over Dallas. The article in the book says the Cardinals were ranked 21st in DVOA for 1998, but since we turned in the final manuscript I fixed some mistakes in my 1998 data and the Cardinals now rank even lower, 25th. The Cardinals went 6-3 over their final nine games, but all six wins came against teams that were 6-10 or worse, and all six wins came by either two or three points. The next year, they were back to 6-10, and they haven't been in the playoffs since.
Three teams stand out in the table above for having WEIGHTED DVOAs much higher than their full-season DVOAs. Two of those teams really improved in 1999: Tennessee, which went from 8-8 to 13-3, and Washington, which went from 6-10 to 10-6. The third team was the New York Giants, who went 7-9 the following year, but at least were good in the first half (5-3 over the first eight games of 1999, 2-6 over the last eight).
But the team we all want to know about, of course is St. Louis. As you can see above, the Rams were #19 in DVOA despite being 4-12. Then again, the Bears also had a DVOA much better than their 4-12 record, and they didn't get much better in 1999. But look closer at the 1998 Rams, and there clearly are signs of something stirring.
Start with the defense, which had an above-average DVOA in 1998. But that rating, by itself, doesn't explain why we should have expected the Rams to take a big defensive leap forward. In the San Diego chapter of Pro Football Prospectus 2005, you will find a long article explaining how third down performance tends to fluctuate far more than first and second down performance. A team that is strong on first and second downs but poor on third downs is likely to improve in the following year. Well, guess what:
|1998 STL Defense by Down||DVOA||Rank|
Defenses with a similar trend include both the 2000 and 2002 Patriots, as well as the 2001 Falcons and the 2003 Chargers. So yes, this is important. But the 1998 Rams offense actually had the opposite trend -- better on third down than first or second, they would have been projected to decline in 1999 without a major influx of new talent. And a major influx of new talent is exactly what they got.
As for that other 1999 turnaround offense, the Oakland Raiders:
|1998 Oakland QBs||DPAR||Rank||DVOA||Rank||Pass||Yards||Yd/Pa||TD||INT|
|1998 Kansas City QBs||DPAR||Rank||DVOA||Rank||Pass||Yards||Yd/Pa||TD||INT|
Yeah, I think I would want that Gannon kid.
A few more facts about the best and worst individual players of 1998:
|1998 Buffalo QBs||DPAR||Rank||DVOA||Rank||Pass||Yards||Yd/Pa||TD||INT|
|1998 San Diego QBs (and friends!)||DPAR||Rank||DVOA||Rank||Pass||Yards||Yd/Pa||TD||INT|
|Bobby Hoying (PHI)||-57.5||49||-64.7%||48||265||759||2.9||0||9|
|Barry Sanders 1998||DPAR||DVOA||Yards||Runs||Yd/R||Suc. Rate|
But if you were wondering if Barry Sanders was in fact the king of boom-or-bust:
|Most Runs for a Loss, 1998||Most Runs of 10+ Yards, 1998|
|Barry Sanders||97||Terrell Davis||50|
|The Long Lost Good Eddie George||84||Jamal Anderson||45|
|Curtis Martin||77||Barry Sanders||44|
|Jamal Anderson||76||Emmitt Smith||40|
|Marshall Faulk||76||The Long Lost Good Eddie George||38|
1998 numbers are not yet on the site for individual positions or team splits like offense and defense, but they will go up after I do the article on 1999 sometime in August. Feel free to ask questions about 1998 in the discussion thread or in email. As with previous years, I'll collect the questions and do either a special 1998 mailbag, or a 1998-1999 joint mailbag.
87 comments, Last at 03 Aug 2005, 9:54pm by Björn