"Football Freakonomics:" Is Momentum a Myth?

In the first segment of “Football Freakonomics” on the NFL Network, Dubner examines the phenomenon of momentum and whether we can actually prove its existence in football games. Here’s a taste of what he found in the data: since 2007, immediately after a long kickoff or punt return, NFL teams are nearly four times as likely to score a touchdown on the next play than they are on a given play from scrimmage.


Ben M. Schorr

Momentum *IS* mostly in our minds, but your mental approach to the play is important in determining outcome. Things like clarity, focus, confidence...when you FEEL unbeatable you often are.

When your confidence is shaken because the other team is rolling it creates uncertainty, doubt, hesitation...all things that tend to lead to poor play.

Kevin

Immediately after a long kickoff or punt return the distance between the starting point of a play and the goaline is shorter than the distance between the average starting point of a given play and the goaline.

Fail on demonstrating momentum using the logic used in this paragraph.

Adam Roberts

I really like it - I love to see more emphasis on applying statistics to sports in a public setting. The twilight-zone music was a bit much though. Probably pretty solid for the uninitiated.

Joe Westhead

Rugby World Cup winning coach Clive Woodward wrote in his autobiography of how they used statistics to show that for every 'phase' of play an attacking team has (sort of similar to a down in American football, except there are no limits to how long you can have the ball), the much higher the likelihood of a try (equivalent to a Touchdown).

I'll try and find the exact quote but it was certainly an interesting passage.

jonathan

Did you not know that Mike Reid, the Cincinnati Bengals Pro Bowl DT in the 1970's, did a study that showed a team is more likely to score a TD after having a TD scored on them and a FG after a FG is scored on them. Reid was also a pianist and composer. To round it out, he's in the Nashville Songwriter's Hall of Fame after writing something like a dozen country #1's. He showed momentum is bunk a long time ago.

It just seems unfair not to mention Reid's contribution, especially since it was done while he was playing at a very high level in the NFL.

David L

Some of the "data" cited in the video is pretty bogus. For example: if an offense converts 3 consecutive first downs, there is a good chance that they are a better than average offense (or that they are facing a worse than average defense), and therefore it's no surprise that they are more likely to get that 4th first down than the average team is to get a first down on a random set of downs. So the effectiveness of the offense (or ineffectiveness of the opposing defense) is a confounding variable with respect to that particular line of reasoning. On the long kick/punt return example, that confounding variable is less of an issue because there is going to be less correlation between a teams offensive and special teams effectiveness (different personnel, coaches, etc). But did the analysis control for proximity to the end zone? Average starting field position is something in the neighborhood of 70 yards from the end zone. Average starting field positions under your "long kickoff / punt return condition" is probably going to average something like 50 yards from the end zone. Now I don't know the exact number but I can guarantee you that 50-yard plays are a LOT more common than 70 yard plays.

That said, it's also going to be very difficult to extricate the psychological elements of what we call "momentum" in a football game from other factors. Even with the psychological aspect removed, each play, and each series, is not an independent data point. Offensive and defensive coordinators adjust their game plans as the game develops on the basis of what has happened. If you're gouging a team on the ground, for example, you might force them to bring a safety into the box, leaving single coverage in the passing game and increasing your win probability on an independent basis. On top of that, an offense that is really moving will tire the defense out more than it tires the offense out, since defenders are responding to what offensive players do, which takes more energy (i.e., covering a WR is a lot more physically taxing than running a route). So a long drive is, on a field-position adjusted basis, going to increase your chances of scoring by tiring out the defense. So how do you separate tactical and/or physical momentum from psychological momentum within a large sample size?

Read more...

MattieShoes

This lacks a conclusion. It physically hurts to see statisticians fall back on "if you ask me". If I wanted an opinion, I could have asked a hundred million fans.

It also totally ignored painfully obvious things like, "Good teams are more likely to get three consecutive first downs, and that might affect the likelihood of getting another."

If you're going to do stats pieces about football, either take a position with stats to back it, or man up and say "Yeah, even with all the stats in the world, I just don't know."

KevinB

I've always thought momentum was just more inane commentary from announcers.

Joshua Northey

Wow that was very unenlightening. Why don't you spend some time following football analytics before you pretend to practice it?

This is starting to remind me of Gladwell, where when you veer into topics I actually know well, your argument suddenly seems childish and slapdash. It makes me question your other arguments in areas I don't have as much background in.

Benny Donovan

For this debate to be worthwhile we need to be able to quantitatively measure the behaviours and cognition levels. Furthermore, it is complicated by the fact that one needs to consider the athleticism and skill levels.

We are way off in our ability to effectively measure all these elements without disrupting the game it's self.

assumo

I always hear annoucers say, "the momentum is shifting," as if they were meteorologists predicting a cold front. I don't believe that it is anything you can predict, or that it's a quantity that you can drop into a bucket, but I do believe that confidence affects sports greatly. Teams play harder when they have a shot at winning. That's part of the coach's job- to make the team beleive that is they keep running and doing what they are supposed to do,they can catch up when they're down big. Of course, the team has to buy what the coach is selling, and it's a lot easier to play motivated when your gaining ground rather than losing it.

ryan

Regression TOWARD the mean, not "TO" the mean. To the mean would indicate everyone has the mean score on the next measurement.

Ulysses

To be completely unfreakonomical and illogical, having played sports and watched it all my life, momentum/psychology are both real and huge in sports. I can't prove it but then again neither can anyone else. And I say this as a fairly new-school supporter of data in sports (particularly baseball, where it's most useful [in football, stats are much less useful])

HDD

Agree this is not up to Freakonomics' usual high standards.

But why is it so hard to prove/disprove momentum?
Just compare two sets of stats:
1) The winning percentage (WP) of all teams down by 4-6 points with 5 minutes left in the game; and
2) The WP of teams that had been down by 20+ points at halftime, but trail by 4-6 points with 5 minutes left in the game.
(I think it's fair to assume a team that has outscored its opponents by 14+ points in the last quarter and a half "has the momentum", if it exists.)
If the WP is more or less the same, no momentum. If WP increases significantly for #2, there might be momentum (though it doesn't control for things like weather changes or injuries to star players causing the change).
Problem (more or less) solved, no?