I added a team EPA visualization that's limited to the current playoff teams that might shed some light on where teams stand on offense and defense. What I find particularly useful is the week filter slider, which shows how teams have performed over recent games. You can find it on the tab titled '2011 P.O.'
A few things that struck me:
- The NO offense has outdone the GB offense, particularly in the 2nd half of the season.
- GB's defense is actually much better than its total yards allowed would indicate.
- The PIT offense has faded in the later weeks.
- The BLT offense has come on since week 9, but its defense has faded.
- The NYG defense is over-rated, even over the recent several weeks.
- DET appears to be the most balanced playoff team--solid on both sides of the ball, but the defense has faded in the 2nd half of the season.
- DEN is by far the weakest team in the playoffs.
- HST's defense alone keeps it a threat in the AFC.
- NE's defense is a -7 point defense, but its offense is a +14 pt squad.
- ATL and CIN are mediocre playoff teams as they go, but are opposite sides of the coin.
- SF is all defense.
"GB's defense is actually much better than its total yards allowed would indicate."
That will happen when you intercept 5% of the passes you see, and return several for touchdowns.
I didn't look ay of that up, so those "stats" of mine could be way off.
Brian, does taking the last 8 or 10 games versus taking all of the games increase accuracy of prediction? I know you're usually against weighing newer games more heavily, but what about simply throwing out the oldest. I guess its a matter of sample size vs freshness of data, and I know you've touched on it before, but I'm wondering what your thoughts are about it these days
Also it would be cool if you could make a team ranking (like the weekly team rankings) that we could manipulate like this as well. I'd like to see GWP and how each team stacked up against the rest if we cut out the first 4 games.
Because this is _not_ limited to only playoff teams, I got more confirmation of what eyeballs and basic stats were saying: Arizona's D improved immensely from the first seven weeks (avg dEPA of 9.07) through Weeks 8-17 (-1.42).
@Anonymous (comment #3),
Aaron Schatz of Football Outsiders commented the other day that with their rating system, total season rankings do a slightly better job than their weighted rankings (iirc, last 8 games are full strength, games 5-8 are about 50% strength, they don't count the 1st 4 games at all at this point). He also said that the only time that weighted works better is for a particular team that might have had bad luck, injuries, etc in the first half of the season, where their weighted formula probably indicates the true "NOW" strength of the team. His comment seemed to indicate that the good teams tend to play good all year, and that the bad teams tend to do the same. Sometimes their wins and losses don't reflect that. [Obviously CHI & HST didn't play to their full potential this last month or so because of QB injuries, for example--but the general rule is what I said earlier.]
OTOH, if I wanted a "true" weighting for DEN, personally I would throw out everything for when Orton played, and only include plays where Tebow was the QB. It also gets rid of most of Brandon Lloyd's plays, since he's not on the team anymore.
To sum up, throwing out old games doesn't seem to help for the NFL as a whole, but I bet it helps for certain teams--like HST & DEN in this year's playoffs.
Thanks for the insightful comment Joseph, although personally I have a hard time trusting anything coming out of Football Outsiders considering they've shown time and time again that they can't grasp simple statistics
It makes sense that Detroit's defense has faded in the second half: on the line, Suh missed almost 2.5 games, and Fairley and Jackson have had injury issues as well, and the secondary, which wasn't that talented to begin with, has been hit pretty hard by injuries. Delmas has been out since getting injured early on Thanksgiving, the game that saw Rashied Davis get some snaps at corner. They've also been forced to use special-teams players at CB and S, and keep in mind that this is on a team whose coverage units have been awful.
The defense is much better than it has been for quite some time, but you can only turn over so many spots on the roster, and the Lions have a ways to go before they can lose a couple of key defenders and not have the unit go up in smoke. (It's also true that their second-half schedule included two games against GB - even if one didn't feature Jennings or Rodgers - and one against NO, but even opponent-adjusted numbers show a decline. They had one bad game before the bye, against Good Chicago, and then three stinkers after the bye.)
how is it that the Giants appear so pedestrian on the visual when they rank #6 on the team rankings? On every visual they compare unfavorably with Atlanta, yet you have them as solid favorites. Especially, why are they barely above average on offense, when you rank them #4 on offense in the team rankings?