Writers of Pro Football Prospectus 2008

03 Feb 2009

The Unprecedented Roller Coaster Ride of the 2008 Cardinals

We know that it was a surprise to see the Cardinals play well in the playoffs after they crashed at the end of the regular season. But was this turnaround really as unprecedented as we made it out to be?

Oh, yes. Yes, it was.

From Week 13 through Week 16, Arizona had three games with a single-game DVOA rating below -60%. That's very, very bad. Only three other teams this year had three games that bad over a four-game period: Detroit, Kansas City, and St. Louis (twice). You may notice these are the three worst teams of 2008 according to both total DVOA and win-loss record.

Then, from Week 17 through Week 21 (the Super Bowl), Arizona had a single-game DVOA rating above 35% for five straight games. A streak this good is just as rare. Only one other team this season managed a DVOA rating above 30% in at least five straight games: the New York Giants, from Week 7 through Week 12.

Has there ever been a team that followed up a streak that bad with a streak that good? Certainly not during the DVOA Era, which goes back to 1995 -- even if we loosen up the requirements a little bit.

I went looking for teams that put up
1) at least three games with DVOA below -40% over a four-week span, and then
2) at least three games with DVOA above 40% over a four-week span.

Only one team since 1995 qualifies: the 2008 Arizona Cardinals.

Then, I changed the baseline to three games below -30% followed by three games above 30%.

Now two teams qualify: the 2008 Arizona Cardinals and the 2003 Minnesota Vikings. The Vikings had three horrible games below -50% between Week 10 and Week 13, then managed DVOA above 30% in Weeks 14, 16, and 17. Unfortunately, they blew their playoff spot when Nate Poole caught a game-winning touchdown in the final minute of that Week 17 game, and Arizona beat the Vikings 18-17.

What happens if we change the baseline to three games below -25% followed by three games above 25%?

Now four teams qualify. Along with the 2008 Cardinals and 2003 Vikings, we have:

1) The 2007 New York Giants, who had three games below -25% between Week 12 and Week 15 and then five straight games above 45% from Week 17 through Week 21, thus winning the Super Bowl, and

2) The 2003 St. Louis Rams, who were all over the map all season. The Rams had DVOA ratings below -25% in Weeks 9, 11, 12, and 17 but DVOA ratings above 25% in Weeks 6-8, 10, 13, 15, and 16.

The Cardinals' performance is far beyond this baseline, and yet there are still only three other teams since 1995 that had a similar run of good play after a run of awful play.

If we change the baseline to three games below -20% and then three games above 20%, we get a baseline matched by roughly two teams per season. The other team for 2008 was Dallas, which had three games with DVOA below -20% from Week 6 through Week 9 (the Brad Johnson games) followed by games with DVOA above 20% in Weeks 11-13.

Posted by: Aaron Schatz on 03 Feb 2009

19 comments, Last at 04 Feb 2009, 8:07pm by Jetspete


by RickD :: Tue, 02/03/2009 - 1:44pm

The high variance in DVOA would seem to point to a lot of competitiveness in the NFL overall. If we believe (as I do) that the Cardinals didn't radically change over a two month period, then the real difference between their level of play could be viewed as relatively small. And yet a small difference in their level of play yields a huge difference in outcome.

Another way of saying this is:
The NFL is very unforgiving of sub-otpimal play on the part of any team. Look at what happened to the Giants down the stretch and in the playoffs. Two months ago, they were playing at a higher level than anybody (and certainly a higher level than either the Steelers or Cardinals managed on Sunday). And yet a slight fall-off (precipitated by the absence of Eli's favorite receiver) and suddenly they cannot win a single playoff game.

The Cardinals got hot at the right time. As for the Steelers, they played at a pretty consistent level the entire year, and they were the only team among the elite group who managed to do that. Good for them.

by bravehoptoad :: Tue, 02/03/2009 - 4:15pm

Baltimore Ravens? Weren't they a pretty consistent team, too? When did they have a big drop-off?

by Bobman :: Tue, 02/03/2009 - 4:57pm

My guess is Flacco's "rookieness" left them pretty inconsistent over the first half of the season. I know the Colts waxed them something like 30-3 (back when Indy was struggling mightily amid their 3-4 start), and if they had the chance to play in the playoffs it would not be the same thing. Not by a long shot. So that's a pretty high level of variance within one season. My impression of Balt is that they righted the ship and were stable and formidable down the stretch, probably coinciding with Flacco's maturation. No drop off, but not quite consistent over 17 weeks. Actually, kinda like Indy: up-and-down for 7 weeks, then world-beaters for 10.

by joepinion (not verified) :: Tue, 02/03/2009 - 3:08pm

This is a good posts. It indicates that the league isn't going totally insane, but that the Cardinals are a one-of-a-kind team.

by ammek :: Tue, 02/03/2009 - 3:53pm

Interesting that the five examples given are heavily biased to the second half of the regular season and the postseason. This suggests one or more of: injuries; broad-scale coaching adjustments; teams/individuals "getting it" where a new scheme or adjustments are concerned; increased intensity/urgency (own and opponents') as the season progresses; a significant lag in opponent adjustments in DVOA in the second half of the season.

The last of these can be cleared up easily enough: do the same outliers creep up when using Weighted DVOA?

Also, from an unscientific point of view, this year's big streaks and swoons all seemed to take place near the end of the year: San Diego, Philly, Minnesota, Houston and the AFC East went on uncharacteristic win binges, while Tampa, Denver, Green Bay, and the rest of the NFC East slumped. Do the stats bear this out, or is such streakiness entirely usual and/or due to opponent adjustments and/or a figment of my imagination? And how does 2008 compare to previous years?

by Waverly :: Tue, 02/03/2009 - 5:13pm

Since Weighted DVOA is supposed to depend on recent games, looking for interesting cases of consecutive games based on WDVOA might not give you the results you want.

by Will :: Tue, 02/03/2009 - 7:56pm

By definition, these streaks have to come later in the season. Since we are looking at four week spans, the earliest the "uptick" could happen are weeks 5-8.


by ammek :: Wed, 02/04/2009 - 4:28am

Yes, but the earliest the "uptick" did happen is Week 13, in Aaron's four examples.

by ammek :: Tue, 02/03/2009 - 4:56pm

Looking back at the 2003 Vikings, it's somewhat surprising to see them as the Cardinals' closest kin. I suppose the three "horrible" games they played from Weeks 10-13 were the losses to San Diego, Oakland and St Louis; Minny's dismal DVOA here must have a lot to do with opponent adjustments since (if you'll forgive my use of conventional stats for a moment) the Vikes outgained all three of their opponents and had more first downs. They did, however, lose the turnover battle 11-3; this was the infamous Culpepper Fumble Season, and the Vikes were employing something called Kelly Campbell opposite Randy Moss at WR.

All three of the losses were on the road; indeed, the Vikes lost their last five road games in 2003, while going 6-2 at home (including both of the heavily positive DVOA wins mentioned in the article). For me, the most urgent offseason project for FO is to incorporate home/road adjustments into DVOA; how much of the Vikes' three-game DVOA slump was due to playing three road games in four weeks (including two on the west coast)? Whether there's a correlation or not, DVOA ought at least to be able to guide our thinking here.

Finally, to reiterate something I mentioned in my previous post, it's rather surprising that the 2003 Vikings' bad/good split is skewed to the second half of the season. Minnesota started the season 6-0, then lost four straight, before splitting its last six. From a purely win-loss perspective, I'd have expected the DVOA streakiness to come during the first half of the season, rather than the second. (Again, a home/road adjustment would make a difference, since the Vikes scored all three of its road wins in the first five weeks of 2003.) Is this late-season drama just random, or does variance generally increase in the closing stages of the season?

PS As you may have gathered, this is the kind of article I like best on FO. Please keep 'em coming during the offseason, to give us strength during the annual Favre Retirement Snorefest.

by Bobman :: Tue, 02/03/2009 - 4:59pm

OMG!!! Brett Faavre might be retiring this offseason? How can you have kept this a secret all this time? This is BIG News! Does Peter King know? Somebody better call him right away.

by ammek :: Tue, 02/03/2009 - 5:18pm


And be thankful I didn't mention the soon-to-be-daily Brady's Ligament Update Zzzzzzarama.

by ammek :: Tue, 02/03/2009 - 5:26pm

Sorry to hog the discussion but...

What happens when you reverse the equation: three excellent games followed by three feeble ones? Any more frequent? Early or late season? 2003 Vikes again? 2008 Cards?

by Bowl Game Anomaly :: Tue, 02/03/2009 - 5:35pm

I'm surprised the Marty Schottenheimer Redskins didn't make this list. They were atrocious for the first few games and then pulled together and became a decent team later.

(Formerly "The McNabb Bowl Game Anomaly")

by mlc0808 (not verified) :: Tue, 02/03/2009 - 6:58pm

I think this has a lot to do with the fact that DVOA doesn't have a metric to measure effort.

The Cardinals held a lot back the last 4 weeks of the season once they had their playoff spot set.

That they did indeed play poorly (as reflected by their DVOA the last month of the season) does not mean they weren't capable of playing better (as reflected by their DVOA in the playoffs).

That's not a fault of DVOA . . . there really are no statistics that can measure that reality.

by Will Allen (not verified) :: Tue, 02/03/2009 - 8:29pm

Boy, that 2003 Vikings team is the source of painful memories. If they hadn't blown that last game against the Cardinals in such an unbelievable fashion, I think they had a decent chance to do some damage in the playoffs. They would have played the Eagles first, and the Eagles stunk against the run by the end of that season, while the Vikings really ran the ball well.

Of course, the real deficiency with the Vikings then was cheapskate ownership, unwilling to spend money to make the team better. If they had simply signed Dre Bly that season, instead of letting him go to the Lions, there's a decent chance that their pass defense would have improved enough to make them a strong contender for the conference championship game.

by Eddo :: Tue, 02/03/2009 - 9:10pm

Not to pick nits, but the Vikings would have hosted a first round game had they beat the Cardinals to finish the 2003 season. The Eagles, however, had homefield advantage throughout the NFC playoffs, and thus a bye week. The Vikings would have hosted either the Seahawks, Cowboys, or Packers, all of whom finished 10-6.

by Henrik Madsen (not verified) :: Wed, 02/04/2009 - 6:52am

Very good article. It accurately captures the true uniqueness of the Cardinals this year. I believe every mediocre team is capable of one great game at any time, but only great teams can do it consistently. And the Cardinals really played very well in all four postseason games, even though they lost the Superbowl.

Usually the team with the early upset falls down to earth in their next game, and you end up being disappointed with a boring blow-out. Not this year. Looking back nobody can deny that the Cardinals totally deserved to represent the NFC in the Superbowl.

Still, there was luck involved - and lots of help from the Eagles. If Minnesota or the Giants beat the Eagles, Arizona goes to bad weather in New York for either the division or the championship round. I don't see them continuing their 40% DVOA streak there.

by Pete (not verified) :: Wed, 02/04/2009 - 8:51am

Does this mention other teams securing their position as a divisional champion so early in the season? I truly believe that this had an effect, no matter how small, on the outcome.

I also believe that New England went to the other extreme, trying to be the best team in history (and succeeding for all but 2 or so close victories and the loss at the Giants).

What is the lesson that most NFL teams are likely to re-learn? Many late-season games do not matter. If you are in the playoffs, then only worry about doing well in that single elimination tournament... Yawn.

by Jetspete (not verified) :: Wed, 02/04/2009 - 8:07pm

I just think you need to look at certain motivational factors when judging the Arizona "turnaround." The Cardinals lost a hard fought game to the Giants in Week 12 that cost them home field advantage. They had a 3 day, 3000 mile turnaround and lost to an Eagle team desperately in need of a win. That loss cost them any realistic chance at a number 2 seed. Having already clinched the division and a home game, what reasons other than pride did the Cards have to play for against the Vikes or Pats? Using that perspective, the turnaround makes sense. Werent the Cards an FO darling the early part of the season? I know Bill Simmons says that the NFL is not the NBA, teams just can't turn it on and off. Well, that is exactly what happened. The Cards turned it off, coasted in, then turned it back on.

Take a look, According to DVOA they were the third best team in the NFC and second in weighted after week 12. take away the post-plax giants, and the Cards WERE, who you THOUGHT they were!