Friday, January 15, 2010

2009 NFL regular season wins and stats correlations

With the post-season underway, I thought I'd have a little fun running correlations between stats and wins for the 2009 NFL regular season. There are plenty of professionals at NBC, FOX, CBS, and ESPN (just to name the major broadcasters) who slice and dice this stuff for a living, and I'm not interested in trying to give them a run for their money by controlling for other variables, matching up output with active players for various games, and the like. Just the raw correlations for entire teams here:

Points scored.88
Team passer rating.81
Total yards gained per play.81
Yards gained per pass play.80
Total yards gained.77
1st downs.70
Turnover ratio.69
Pass yards gained.68
3rd down conversion %.64
QB hits allowed(.53)
Sacks allowed(.53)
Time of possession.46
4th down conversion %.25*
Pass attempts.14*
Run attempts.12*
Yards gained per rush play.09*
Offensive penalty yards.04*
Run yards gained.04*
Pass/run ratio.02*
Points allowed(.68)
Run yards allowed(.58)
Passing yards allowed per pass play(.57)
Total yards allowed(.56)
Total yards allowed per play(.54)
Opponent's passer rating(.47)
1st downs allowed(.45)
Sacks made.41
Run yards allowed per play(.33)
4th down conversions allowed(.30)
Passing yards allowed(.24)
3rd down conversion % allowed(.23)*
Defensive penalty yards.14*
Special Teams
Average kickoff (kicking team).32
Average net punt yards (kicking team).27*
Average kickoff return yards allowed.15*
Total penalty yards committed.11*
Average kickoff return yards gained.05*
Field goal %.03*

* not statistically significant at 90% confidence

Points scored is the best predictor of a team's win-loss record. That's hardly surprising, since it's almost like saying a team's number of wins is the best indication of its success. But when considered along with other statistics, it does indicate how good offense (and passing, specifically) has come to supercede in importance good defense in posting a winning record. Every year, though, Chris Collinsworth or Boomer Esiason will remark in the course of a playoff game how crucial a good defense is to move forward in post season, insinuating that it matters more then than it does during the regular season. The NFL's official site archives stats going all the way back to 1932 (in a limited capacity), offering a way to empirically test that assertion for someone interested enough in it to make the required time investment.

The passer rating system as a measurement of QB performance has its critics. It doesn't account for yards gained on the ground or its corollaries, scrambling ability and sacks taken. But it's almost as useful as points scored in predicting a team's record.

Most teams win with their passing games. Of the top ten rushing teams, only four are in the playoffs. And the two odds-on favorites to win the Super Bowl, San Diego and Indianapolis, are #31 and #32, respectively. In contrast, eight of the top ten offenses in terms of total passing yards made the post-season, and only one playoff team, the improbable Jets, can't move the ball through the air.

Although special teams are often accorded one-third of a team's total equation by media figures and putative insiders, the layman's tendency to deemphasize them in favor of offense and defense looks pretty reasonable. It is nearly impossible to predict a team's level of success by looking only at its special teams statistics.

A defense that is worn down by being run at throughout the course of the game fares worse than one that gives up a lot of total yards in the air, statistically speaking. But this is surely obscured by the fact that teams that are winning run while those playing from behind pass.

Turnovers are huge. Penalties are not. Fourth down attempts are too infrequent to be of great consequence in aggregate, but the ability to convert on third down is important. Again, though, better teams are going to have more third and short situations than the Rams or Lions are.


sykes.1 said...

Someone did this a few years ago and concluded, like you, that football is a passing game. Again, a famous defensive player from 20-30 years ago said that no running back would ever beat them.

I think the reason for the emphasis on running is that very few teams, even at the pro level, have a great passing quarter back, and so they have to make do.

Jokah Macpherson said...

The WSJ had a article within the past two weeks talking about the same thing - how offense, and more specifically, passing offense, is a greater predictor of NFL team success than defense. It may be that all things being equal, a passing play is just superior to a running play. The running back is fair game as soon as the handoff takes place, but while the receiver may get plowed over by a safety, this is not going to take place until he has at least had a shot to catch the ball for positive yards.

Audacious Epigone said...

I was going to point out that of the four teams going into the division championship games next weekend, the starting quarterbacks' passer ratings were 1st (Brees), 2nd (Favre), 3rd (Rivers), and 6th (P. Manning) during the regular season, and also that the Ravens, exemplars of the great defense/solid run game formula, underwhelming loss to the Colts illustrated how a high-powered passing game is a necessary ingredient for any truly top-tier team. Then the Jets go and vindicate what the Collinsworths and Esiasons (or today, Simms) inevitably assert each year--defense becomes especially important in the post-season.

This is why I remain only descriptive rather than attempting to be predictive when writing about the NFL!

Anonymous said...

great read. I would love to follow you on twitter. By the way, did anyone learn that some chinese hacker had hacked twitter yesterday again.

Steve Sailer said...

Yes, it seems like NFL running backs aren't as big stars as they used to be relative to quarterbacks.

A 0.09 correlation between yards per running play and wins is extremely low. I bet in the 1970s is more like 0.5.

A big thing NFL teams have done is cut down on the number of interceptions. Joe Namath threw 220 interceptions compared to only 173 touchdowns and had five years of 20 or more interceptions, 5.8% of his throws. Philip Rivers has only has been intercepted on 2.4% of his passes.

I wouldn't be surprised if running back fumbles haven't declined as much, since not fumbling is a simpler coaching problem to solve than not getting intercepted.

Still, you'd think there would be some way to make running the football more meaningful to wins in the NFL.

Steve Sailer said...

I went and did 2008's correlations to see if passing dominated rushing as much.

Not really.

Net Yards / Pass Attempt (i.e., yards lost from sacks subtracted from yards gained passing) correlated with wins 0.55

Yards / Rush Attempt = 0.14

So, the passing v. running gap wasn't as great in 2008 as in 2009.

dk said...

The NFL has been adjusting the rules on pass protection the last few years.

1. Any contact downfield is flagged
this makes it extremely hard to do man coverage on the #1WR. And increases YPC average.

2. There is a presumption of guilt on Des Pass Inter.

3. You have to be real careful trying to bat down a pass around the QB. If your hand hits his helmet you are now flagged for 15.
So qbs have less of a risk of the ball being fumbled/batted in the air for a int.

I wouldnt use the raven colts game as a example of passing game dominance. 20 points isnt exactly awe inspiring. And the colts have the #6 scoring defense in the league.