"Football Freakonomics": When Good Stats Go Bad

In the third segment of “Football Freakonomics,” Dubner examines how impressive stats in the NFL are often indicative of bad results. For example, we all want a quarterback who throws for big yardage. But for all the times a quarterback threw for 400 yards or more last season, how many of those games did his team actually win?


I would imagine that the answer is the team had less wins when the quarterback throws more. The offensive aspect of the game contains two methods for gaining yardage: through the air or on the ground. Usually a team isn't successful when only one method is used, so I would assume that an offense who has a QB throwing 400+ yards also has a lacking running game. In this situation, we can assume the running game is failing, and it is a part of the game that winning teams are -- more often than not -- keeping the balance of run vs. pass.


Actually it has more to do with the scoreboard... Teams with the lead rush more (it eats more time off the clock), and teams that are losing pass more (to conserve time on the clock).

The rushing correlation is perhaps more famous -- "Team X is 4-0 when they rush the ball 30 or more times" and other such nonsense. They rushed the ball 30 times because they had the lead, and teams with the lead usually win.


Another problem with looking at an individual's stats is that they miss a lot of intangibles.

A really great wide receiver can be completely shutdown by simple double teaming. But, that leaves one less defender to cover everyone else, and so the rest of the offense should be more productive.


While MarTay Jenkins does hold the record for kick returns and kick return yards, that is NOT a coveted record. And why are you comparing him to a wide receiver and a running back? Incidentally, he gained over 2400 all-purpose yards that year. That 2402 yards ranks 14th in all purpose yards in a single season . His yards/return that year ranks 236th. Weak example. Try average punt length for a better example. Bad teams tend to punt from deep in their own territory so their punts tend to go farther -- They don't have to try to avoid kicking them into the endzone.

We only notice return guys on bad teams? Yeah, nobody noticed Devin Hester the year he went to the superbowl, right? And Gale Sayers, surely nobody has heard of him since he played for a good team. Seriously, do you guys even pay attention to what you're saying?

A+ for identifying that passer rating is terrible, F for suggesting that Culpepper has a higher passer rating because the teams he played for were bad. Lets start with this: Did you not notice that passer ratings have been increasing steadily for decades now? Did you not consider how that might affect comparisons of players from different eras? Did you not notice that 28 of the top 30 in career passer rating started their career after the "Mel Blount rule" in 1978? (Otto Graham and Roger Staubach are the exceptions)

Crediting or blaming Culpepper for his team's wins and losses... Sloppy. Where's your evidence?

Seriously, what the heck happened to you guys? These are TERRIBLE.


Neil (SM)

They did not even suggest that Culpepper's passer rating had anything to do with being on a bad team. They said that his passer rating was unimportant because it's wins that count, and his teams were not winning.


That is nonsensical. Does that mean Dan Marino's numbers were unimportant because he never won a Super Bowl?


What a ridiculous premise. What does 400 yards have to do with anything? It's a loaded argument because of it's dependency on unique game conditions. NFL statistical prowess is always contextual.


Don't forget teammates when judging a quarterback. Big Ben has worked with amazing defenses his entire career. Put him on a team with a moderate defense and there's no way he wins as much as he does. This is only one example.


This video is ridiculous. Yes, I want a quarterback who will win games, but as football is a TEAM GAME I want a good passer first. If I've got a good defense and a good run game then I have no need to have a good quarterback as was the case with Trent Dilfer. Despite what the Mad Dog has to say, I don't want a quarterback who is 5-30 with 5 picks no matter what he does in the fourth quarter, because most of the time if my quarterback is that bad the game isn't 10-10 going into that quarter.

Jared Doom

(1) I think Dubner is tearing down a bunch of Straw Men here, unless he's assuming the audience is a bunch of casual football viewers.

A 400 yard game (with knowing nothing else) by a QB would lead me to the following conclusions:
(a) The quarterback is likely good. Mediocre quarterbacks don't often have 400 yard games.
(b) His team was behind for at least a good portion of the game. Teams behind need to pass a lot because it is the fastest way to decrease the point differential. The need to pass decreases as the point differential (relative to the offensive team) increases.

It should not be surprising when a QB with over 400 yards passing loses.

(2) Passing efficiency (offensive yards per pass attempt) is BY FAR the best predictor of a team's future probability of winning games: It is also highly correlated with passing yards, so, yes, you do want a quarterback who throws for a lot of yards.

(3) An amazing analysis of why teams win is here: http://www.advancednflstats.com/2007/07/what-makes-teams-win-part-1.html

This article points out (as you've done indirectly here) that # of passing attempts is correlated with losing and # of rushing attempts is highly correlated with winning. One of the best examples of causation/correlation confusion ever.

(4) The website on which this article resides has mountains of original research on this and similar NFL phenomena. (http://www.advancednflstats.com). HIGHLY RECOMMENDED.



Wins are primarily a team stat, and QB rating, yards, and TDs are primarily individual stats.

Many times, a great QB on a bad team will lose more than he wins. That doesn't make him a bad QB. In many cases, a great QB is the only reason the team is competitive, even when they lose.

For example, Cam Newton of the Carolina Panthers is having an outstanding season, according to most experts, broadcasters, and fans alike. They were the worst team in the league last season, and now they can play head-to-head with any team in the NFL. They are only 2-6, but they do not get blown out every week anymore. Most games are close, and they often lose games in the 4th quarter because teammates are not playing at Cam's level. The overall team still has too many holes. A cornerback may blow his coverage let a receiver catch the winning TD pass. A linebacker may miss a simple tackle that results in a long run and propels the opponent to victory. A breakdown at any position can cost a team a victory.

Therefore, a QB's win-loss record is more useless than his individual stats. Once you move a good QB from a bad team to a good team, his record will almost certainly improve, while his individual stats may not change at a similar proportion.

There are exceptions, but we are talking about the rule, which is clearly correct a large majority of the time.