"Football Freakonomics": When Good Stats Go Bad

The following is a cross-post from NFL.com, where we’ve recently launched a Football Freakonomics Project.

What do Dan Marino, Jerry Rice, and MarTay Jenkins have in common?

Yes, wise guy, they all played in the NFL. But beyond that? They all hold all-time single-season records.

+ Marino (among his other records) passed for 5,084 yards in 1984.

+ Rice (among his many other records) gained 1,848 receiving yards in 1995.

+ Jenkins had 2,186 kickoff-return yards in 2000 for the Arizona Cardinals.

But Jenkins, unlike the other two, won’t be getting a call from Canton any time soon, even though he set a second record that season – for the number of kickoff returns, with 82. Eighty-two kickoff returns! That’s an average of more than 5 a game.

Care to guess the Cardinals’ record in 2000? They were 3-13. Yes, it’s great to be a kickoff returner when your team is getting kicked off to over and over and over again.

 And so it is that MarTay Jenkins is the poster boy for our latest Freakonomics Football video, “When Good Stats Go Bad.”

Most of us are by now pretty savvy about which stats are meaningful in football and which ones aren’t. But as the border between real football and fantasy football shrinks ever thinner, it’s easy to lose sight of the fact that big numbers can mean big trouble. How happy were Raiders fans last season to see their team ranked second in the league in pass defense? Pretty happy – until they realized that was because their run defense was so porous that opponents didn’t need to bother passing. Same story for the Buffalo Bills.

Perhaps the most easily misleading stat is big-yardage passing games for quarterbacks. Just as a swinging bunt looks like a line drive in the box score, a massive passing game looks awesome on your stat sheet – until you realize it’s more likely to indicate a bunch of worthless garbage-time yards.

Last season, QB’s who threw for 300 or more yards in a game had a combined record of 47-49. QB’s who threw for 400 or more yards were 3-11. This season, we’ve already seen 10 400-yard-plus passing games – although, interestingly, none since Oct. 9. (Can anyone say “lockout defense”?) The combined record of those quarterbacks was 3-7, with Tom Brady accounting for two of the wins and Aaron Rodgers for the third.

My favorite example of when good stats go bad is career fumble recoveries. Who do you think might top that list – maybe Jason Taylor or Jim Marshall or Kevin Greene? All those players are on the list but in fact the top 21 slots are taken up by quarterbacks, the same guys who dropped the ball in the first place.

 


Tom Woolf

I remember an old NFL record that might still stand - number of your own fumbles recovered. Roman Gabriel of the LA Rams (told you it was old) recovered 4 of his own fumbles in one game.

BSK

What is the point of this? Any serious football fan, or fan of any sport, is increasingly aware that context matters and that few numbers should be taken at face value. Freakonomics is attempting to solve a problem they are creating. No one is lumping Jenkins in with Rice or Montana. You do and then say, "It's crazy to lump them in together and here is why!" Pure silliness. For real football analysis, head to Pro-Football-Reference.com or FootballOutsiders.com.

Erik Jensen

It is not pointless to claim that stats are misused, even by so-called experts. Conventional wisdom is that you need to win the "time of possession" to win games. I'm sure that the stats agree that the winner of time of possession (usually the team with more running attempts) usually wins the game, but this is confusing correlation with causation. You don't run a lot to win, you run a lot because you are winning. Another bad statistic is head to head record over, say, the last 10 years. There might not be a single player left from 10 years ago, but announcers (so-called experts) will provide these stats as if they are meaningful.

Advanced NFL Stats is also a good site for football nerds.

Nathan Stockstill

Speaking of context, how much of it is due to the way the game is played? For instance, has passage yardage always correlated negatively with victory, or is it just how the game has been played recently?

Bob

You guys should leave the football stuff to Brian Burke over at advancednflstats.com.

Tom

I want to add in one of my favorite players, Brian Moorman- punter for the Buffalo Bills.

Neil (SM)

Didn't we already do this one?

http://www.freakonomics.com/2011/10/30/football-freakonomics-when-good-stats-go-bad/