by Aaron Schatz
Now that we're down to just two conference champions, there's no point in doing a whole table of "postseason weighted DVOA" with all 32 teams. The Patriots and Giants are the only teams worth talking about, and besides, the weighted DVOA for the Giants still looks ridiculous because they were so mediocre from Week 10 through Week 16.
So let's talk about the Giants and the Patriots -- well, mostly the Giants.
First, here are the ratings for the conference championships.
| DVOA: 2007 Week 20 | ||||
| TEAM | TOT | OFF | DEF | ST |
| NE | 41% | 18% | -14% | 9% |
| SD | 21% | -6% | -29% | -2% |
| NYG | 48% | 10% | -46% | -9% |
| GB | -40% | -25% | 12% | -3% |
Many readers have asked for VOA as well, without opponent adjustments. Here you go, although the adjustment for "fumble luck" remains.
| VOA: 2007 Week 20 | ||||
| TEAM | TOT | OFF | DEF | ST |
| NE | 29% | 9% | -11% | 9% |
| SD | -22% | -11% | 9% | -2% |
| NYG | 28% | 8% | -29% | -9% |
| GB | -40% | -29% | 8% | -3% |
We know the Giants are playing better now than they did before Week 17, but we're still left with the question of how many weeks are worth using to judge the Giants when we look at matchups for the Super Bowl. Does every single game before Week 17 no longer count? Is four games really enough to judge a team? If we want to also look at games before Week 17, how many should "count," and how important are they relative to the last four games? There's really no clear answer here.
As an experiment, however, let's look at just the last four weeks of the season. Now, we're left with one other issue here, which is that the opponent adjustments are based solely on the regular season. New England's performance in Week 17 is being judged as if they played a Giants team that was completely average -- but we think the Giants are much better than that now. So I've gone back and re-done opponent adjustments (for all 32 teams) so they now include every game from Weeks 1-20, rather than just Weeks 1-17.
So, with that in mind, here are the DVOA ratings for New England and New York only over the last four weeks: Four games for the Giants, and three games for the Patriots.
| DVOA Weeks 17-20 | ||||
| TEAM | TOT | OFF | DEF | ST |
| NE | 44% | 45% | 1% | 0% |
| NYG | 43% | 28% | -8% | 6% |
Yes, that's correct. The Patriots have a higher DVOA rating than the Giants even if we only measure the past four weeks. It's really not statistically significant, but the difference becomes more substantial if we take out the opponent adjustments, which are still heavily rewarding the Giants for the very act of staying in the game with the Patriots.
| VOA Weeks 17-20 | ||||
| TEAM | TOT | OFF | DEF | ST |
| NE | 30% | 39% | 9% | 0% |
| NYG | 17% | 24% | 13% | 6% |
None of this is meant to belittle the accomplishment of the Giants over the past four weeks. They've done something extremely rare: put up four straight games with a DVOA rating over 40%. Here is a table showing every team that has a streak of four or more games with a DVOA over 40%, along with the average rating over the streak, and the rating for each game.
| Streaks of 4+ Games with DVOA Over 40%, 1996-2007 | ||||||||||||||
| Team | Year | Weeks | Avg. DVOA |
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 |
| NE | 2007 | 1-12* | 71% | 63% | 98% | 78% | 66% | 51% | 41% | 75% | 114% | 66% | 78% | 54% |
| STL | 1999 | 11-15 | 70% | 82% | 106% | 54% | 63% | 43% | ||||||
| TB | 2002 | 2-6 | 60% | 86% | 63% | 42% | 64% | 46% | ||||||
| IND | 2007 | 3-8* | 59% | 43% | 72% | 64% | 63% | 51% | ||||||
| KC | 1997 | 12-15 | 78% | 48% | 56% | 109% | 99% | |||||||
| PHI | 2004 | 10-13 | 77% | 52% | 62% | 95% | 98% | |||||||
| IND | 2004 | 10-13 | 75% | 73% | 98% | 64% | 64% | |||||||
| GB | 2003 | 15-18 | 71% | 41% | 117% | 78% | 46% | |||||||
| OAK | 2002 | 16-20* | 59% | 42% | 69% | 72% | 54% | |||||||
| GB | 1997 | 13-16 | 53% | 43% | 47% | 65% | 56% | |||||||
| NYG | 2006 | 5-8 | 53% | 46% | 59% | 65% | 40% | |||||||
| NYG | 2007 | 17-20 | 49% | 47% | 55% | 46% | 48% | |||||||
| *includes bye week | ||||||||||||||
One of these accomplishments sure stands out, doesn't it? Of course, since Week 13 the Patriots have fewer games over 40% (three) than the Giants do -- but two of those three games happen to be the Patriots' two postseason wins.
Notice who else shows up on our list of teams with four straight games over 40%? Hey, it's the 2006 Giants. These wins came primarily against poor teams (they beat Washington, Atlanta, Dallas, and Tampa Bay) but after opponent adjustments, the streak is awfully impressive. What happened to this "hottest team in the NFL" after Week 8? They followed their four-game win streak with a four-game losing streak and ended the year 8-8. Whoops.
If we loosen our restrictions to look at three-game streaks, we'll find other teams that went on surprising late-season runs but didn't finish up with a Super Bowl title. The 2002 Jets are a good example. The Jets snuck into the playoffs at 9-7 by winning their last two games 30-17 over New England (DVOA: 46%) and 42-17 over Green Bay (DVOA: 78%). In the first round of the playoffs, they annihilated Indianapolis by the ridiculous score of 41-0, which had a one-game DVOA of 130%. The streak isn't as long as the Giants' streak, but the wins are far more dominant. What happened to the Jets in the divisional round? Oakland 30, New York 10.
There's one other question about the Giants: Should we have seen this coming, either objectively or subjectively?
Objectively, I don't really see the evidence. There have been plenty of comments in our discussion threads, saying that the Giants' postseason run shows a major flaw in our DVOA formula. None of these comments, as far as I can tell, give any suggestions that would improve things. Any change that is intended to raise the Giants' rating for the 2007 regular season has to work for every other team since 1996. If we really want to use this to improve the ratings, we need to find other teams that had the same markers and were also underrated by the formula. I'm all about the suggestions, and I'm always looking to make the ratings more accurate, but comments that say "The Giants totally prove DVOA sucks" don't really get us anywhere.
(As an aside, people around the Interwebs do seem to have an awfully hard time telling the difference between hardcore statistical analysis of the regular season and subjective preseason predictions that are meant to be somewhat pithy.)
Subjectively, there seems to be a general meme in the press that the Giants really had a great passing game all along, they just needed to cut down on the mistakes. This is a fun bit of 20-20 hindsight, but to how many other mediocre offenses would this statement also apply? Isn't that exactly what happened to the Washington Redskins after Todd Collins came in to replace Jason Campbell at quarterback? How did that work out come playoff time? Don't the Arizona Cardinals have the talent to become a top offense "if they just cut down on the mistakes?" How about the Chicago Bears? The other popular idea is that a team's performance on the road means a lot more than their performance at home, and the Giants were 7-1 on the road during the regular season. As I explained a week ago, that didn't really do much for the 2001 New York Jets.
The Giants have been outstanding for the past four weeks. They are going to go into Arizona and make a game of this thing, no doubt about it. But nothing the Giants have done over the past four weeks invalidates the idea that they were a mediocre, overrated team from Week 1 through Week 16.