The Geek's Guide to the NFL Playoffs
As we've talked about before, the football analytical world had a ways to go to catch up to the analytical baseball world a few years ago. Between the stuff we've done and the work of many others like Albert Breer and Michael David Smith, we're closing this gap in a hurry.
Now, to the arguing: I'm sure there's a lot that we'll agree on in this discussion, but I don't see how the Giants are going to upset the Cowboys. Even with a banged-up Tony Romo, the Cowboys still look like a team that can win a high-scoring shootout. I don't think the Giants can do the same.
The problem is New York's pitiful receiving corps. When I'm evaluating teams and players, I rely a lot on a statistic called Yards Per Attempt. I've found that YPA—total yards gained divided by total pass attempts—offers the best accounting of the efficiency and explosiveness of receivers and quarterbacks. For wide receivers, an excellent season-ending YPA is 10, and a poor one is around six. I know it's a small sample size, but in the two games since Jeremy Shockey left the Giants lineup with a broken leg, Steve Smith and Plaxico Burress have been abysmal. In the last two games of the year, Smith averaged only 5.8 YPA, and Burress averaged only 5.3. They both kept up that anemic pace in the wild-card game against Tampa Bay, with only 61 yards between them on 11 pass attempts.
For a receiver, a low YPA can be a symptom of poor play or of playing with a bad quarterback. In the Giants' case, the team's receivers typically have low YPAs because of Eli Manning's inaccuracy. The YPAs for Burress and Smith, though, are lower than their teammates' numbers, and based on what I've seen on tape, it is more their fault than Eli's. Smith simply doesn't get open. And besides the second game against the Eagles, Burress has disappeared since Week 6, perhaps due to a lingering leg injury. It is very hard for an offense to get moving if its receivers are putting up low YPA numbers. While Amani Toomer and backup tight end Kevin Boss are both producing at a much better rate, the Giants can't beat Dallas with those two alone. They'll need Smith or Burress to step up, and I don't see that happening.
On to the Jaguars and their surprising quarterback David Garrard. I think there are two reasons for his statistical improvement. The first is Reggie Williams' morphing from a potential draft-day bust. Williams put up an eye-popping 10.3 YPA this year. That level is usually reserved for the most dominant receivers in the league. Williams accomplished this largely due to his 12 receptions of 20 or more yards. He became the vertical threat that Jacksonville lacked in 2006. For the Jaguars to upset New England, they need Williams to catch a bomb or two.
The second reason for Garrard's surprising season is his decreased interception total. Garrard threw only three picks in 325 attempts this year, by far the lowest interception percentage of any NFL starter. I think this improvement in Garrard's game is mirage based largely on luck. I track how often a quarterback forces a pass into coverage—I call this my "bad decision" metric. Garrard has historically posted a high bad-decision percentage, and his bad-decision percentages this year are just as high as his historical numbers. So, Garrard's low interception total this year has nothing to do with increased savvy in the pocket; he's still taking plenty of chances and still forcing passes into coverage. Don't be surprised if the Patriots' Asante Samuel makes him pay for some of those risky throws.
I think the Jags-Patriots game will come down to one matchup: any Patriots wide receiver against Jaguars cornerback Brian Williams. As I pointed out in an ESPN.com column (subscriber only), Williams had the following stat line against the three highest-potency offenses Jacksonville faced in 2007 (two games against Indianapolis and one against New Orleans): 17 passes thrown at him, 16 completions allowed for 262 yards, and a touchdown. That is a 15.4 YPA—5 yards worse than what will normally get a cornerback demoted out of the league. To be fair, 14 of the passes and completions were to Reggie Wayne, and few people are able to stop him. But even if New England can't get Randy Moss lined up against Williams regularly, Donte Stallworth and Wes Welker should both be good enough to test Williams. If he fails the test, Jacksonville's season ends this weekend.
You mentioned that the Colts secondary doesn't look too good in your game-charting metrics. It's worth mentioning that the Colts play Cover Two, a defense that often hides players' true abilities. Nick Harper did a very good job for the Colts in 2006, but unless you took a really detailed look at his performance, it was hard to tell that he was playing so well. On the flip side, Jason David parlayed a decent 2006 season into a huge contract in large part because he wasn't asked to stretch his coverage skills. While my stats indicate that the 2007 Colts' secondary is roughly equal to the 2006 version, my scouting eye says Kelvin Hayden and Marlin Jackson are playing much better collectively than David and Harper. Jason David was a coverage liability (I had him ranked as the worst CB in the NFL in 2005), and the Colts often had to work around his weaknesses. They don't have to do that now. Now, back to your mention of Bob Sanders: Did your numbers say he was as good as he is generally regarded?
The only matchup we haven't discussed yet is Seattle-Green Bay. I keep hearing how great the Packers receivers are at getting yards after the catch. Part of me assumes this is because they almost never run the ball and they do throw a lot of short passes—ones where you'd expect there to be lots of opportunity to run after the catch. I'm also a big matchup guy and wonder how the Seahawks defense fares in YAC yards allowed. Got any ideas on this one?