Why do we put so much stock into the concept of momentum in sports? Outside of winning and losing, no abstract concept seems to stretch further across the entire spectrum of world sports than momentum. Conduct a Google News search for momentum in sports and you’ll simultaneously see just how meaningful a target it is and how loose its application can be. Baseball teams want to carry the momentum to the final game of their series. College football teams want to carry the momentum from the end of their previous season to this upcoming year. NFL teams want to carry their momentum from the preseason into the regular season. And that’s just on the first two pages of results as I write this. For every usage of the term as it applies to the physical concept involving mass and velocity, there are 30 articles that seem to describe taking some prior moment of success or tipping point and using that as a springboard to future success. It would seem like overkill if it weren’t so damn insidious.
As you probably suspect, I’m a little skeptical of momentum, so much so that I started to use a name for it last February: nomentum. I can’t prove that momentum does or does not1 exist in sports, because it’s an arbitrary, abstract idea that you can mold into just about anything you want to tell the story you’re looking to tell. Of course, you can also say the same thing about statistics, but there are stricter concrete rules that come into play when using statistics; even an untrained eye can recognize a small sample size or an arbitrary endpoint chosen to create a misrepresentative idea. There are really no rules for momentum. You just need to have something good — not even good, actually; just something meaningful — in the past and the hope of possibly succeeding in the future. And the more I look, truthfully, the less of an argument for momentum I see.
I’ve written about momentum here and there in the past, but it really started to become a point of critical contention for me after last year’s Super Bowl, when the momentum argument became the de facto story line for America’s most-watched sporting event. It was simple: The Ravens played great early, but then the blackout happened and the Ravens gave the momentum they had to the 49ers, who promptly went on a massive comeback. It didn’t pass the smell test for a number of reasons, which I wrote about at length the morning after the game.2
I think I made a reasonable case against the concept of momentum in that game, but to be fair, one of the problems with momentum is that it’s often used to explain anecdotal situations. You can’t prove that momentum exists with one game. Not only is it an impossibly small sample, it runs into the issue of confusing correlation and causation; in other words, the 49ers might have played better after the blackout, but it doesn’t mean they played better because of the blackout. To get some real insight into whether something like momentum really exists in certain situations, you need to look at the concept over the bigger picture. That’s beyond the scope of a post-game wrap-up or an in-game graphic, so I ran a few studies myself.3 In each case, momentum did not fare so well.4
The Statistical Cases Against Momentum
Going For It
One of the pet peeves that came up in the “Thank You for Not Coaching” section of my articles each week during the NFL season is how coaches are excessively conservative near the opposition’s goal line. Despite the numbers often being in favor of going for it on fourth-and-short,5 teams usually find themselves opting for the “sure thing” and coming away with guaranteed points by kicking a field goal. One of the arguments against going for it in that situation is that you can give the momentum to the other team by kicking, something that came up in a farcical Chiefs-Broncos discussion late last year that I’ll bring up again later.
Do teams really exhibit any sign of momentum when they get the ball back from the opposition after a big stop or play by their defense? Well, that’s something I can test, thanks to the wonderful new Drive Finder at Pro-Football-Reference.com. In this case, I want to see what teams do when they get the ball back really deep inside their own territory, so I’m starting with drives that begin inside a team’s own 5-yard line. I don’t want to include blowout situations or spots where teams have to go for it because they’re either desperate or in a game situation where they have no choice, so this study only included plays in which the two teams were within 14 points of one another, and in which the defensive team took the ball over with more than five minutes to go in either half. Since Pro-Football-Reference.com’s play data goes from 1999 to 2012, I’ll use those 13 years in my study.
There are three notable ways in which a team can take over the ball that deep in its own territory. Most common, of course, is to take over inside the 5-yard line after a punt. While I believe there’s some momentum to be gained from downing the ball deep in enemy territory or bringing back the coffin corner game, it hardly gets the sort of reaction that more dramatic events receive. A team would presumably get more momentum from a takeaway that gives it the ball deep inside its own territory, with momentum ranging from a forced fumble at the goal line (lots of momentum) to a bomb on third-and-long that some defensive back who should have dropped the ball tracks down (a moderate amount of momentum). The most consistently momentous of the three, though, would be stopping the opposition on fourth down when they choose to go for it and keeping them off the scoreboard altogether. In these game situations, those plays are always enormous.
So, how do these defenses do after regaining the ball and going on offense?
[Editor's note: The above table had incorrect results in an earlier version of this piece. For a discussion of the corrected statistics, read today's post.] That’s pretty stunning. As it turns out, teams actually score fewer points on those possessions that begin with the most momentum, the stoppage on downs, and do their best work taking over after a punt.6 I can throw up a number of qualitative explanations for the output — the offense almost always gets a commercial break after the punt that allows it to prepare for the upcoming series, the big stop is produced by the defense, which then somehow absorbs the momentum without providing it to the cold offense going onto the field, or the coaches are possibly more likely to be aggressive after a punt than they are after getting lucky with a stop on fourth down or from a turnover — but I expect that this is mostly just sheer chance, and that there’s just not evidence of momentum meaning anything at all in terms of those drives.
The Hot Second Half
NFL teams also see momentum as an asset that can carry over from year to year, with a blazing end to the season serving as a promising building block for the subsequent campaign. Fans of teams like the Bengals and Redskins, each of whom finished 7-1 during the second half of the 2012 campaign, have a right to remember their blazing winters and hope that they’re going to produce scorching falls in 2013.
Unfortunately, there’s scant evidence that a slow start and a fast finish are more meaningful in foretelling future success than a team with a similar performance record that did its best work during the first half of the season. In other words, if you finish 10-6, it doesn’t really matter whether you start 3-5 and finish 7-1 or start 7-1 and finish 3-5. I wrote about that research this past June.
There’s another argument that pops up with regard to late-season performance, though: You need to play your best football at the end of the season in order to be ready to succeed in the postseason. It’s the “peak at the right time” argument, a variation on momentum, where you’re carrying over your great work from the end of the regular season into the playoffs. I’ve mentioned that the 2012 Ravens are an obvious argument against peaking at the right time, since they finished the regular season by losing four of their final five games before winning four straight playoff games,7 but that’s only one example. I needed a larger sample, so I went back through 1990 and tried to fairly gauge whether each Super Bowl winner was playing its best football at the end of the regular season.
By my count, there are 10 winners who either played their best football at the end of the regular season or who kept dominating from the middle of the season without letting up: the Super Bowl winners from 1993, 1994, 1996, 1999, 2000, 2001, 2003, 2005, 2008, and 2011. The other 13 teams didn’t stand out in the same way. Twenty-three teams is still a pretty small sample, but it’s fair to say that there’s no clear indication that you need to play your best football in December to win in January. It doesn’t hurt, of course, but based on recent history, it’s far from a prerequisite.
The Overtime Winner
Let me switch gears for a second by changing sports. Living in Boston, surrounded by Bruins fans, I saw a number of momentum-shifting moments alter the Bruins’ season in dramatic ways during the playoffs. First, during their series with the Maple Leafs, the Bruins came back from down 4-1 with three goals in the final 11 minutes (and two goals in the final 90 seconds of regulation) of Game 7 to win 5-4 in overtime, clinching the series. Weeks later, they would lose a chance to play Game 7 of the Stanley Cup final by giving up a one-goal lead at the end of Game 6, with the Blackhawks scoring twice in 17 seconds to turn a 2-1 deficit into a Cup-clinching 3-2 win. In each case, momentum drove the Bruins; it helped them win in overtime against a deflated Leafs team, and it failed them when the Blackhawks burst their bubble at the end of Game 6.
I think you can make a pretty strong anecdotal case against, at the very least, that first playoff win. In that first series, the Leafs had clawed back from a 3-1 series deficit by winning two straight games, and then they had outplayed the Bruins over the first 49 minutes of Game 7, creating an extremely comfortable three-goal lead for themselves. The Leafs had all the momentum; if momentum were really such a driving force in how teams play, how could the Bruins overcome that enormous momentum gap and come back to win? And while the memory will likely suggest that the Leafs had no hope of winning that game in overtime after blowing the lead, the teams skated for 6:05 of overtime before Patrice Bergeron scored the series winner, and the Leafs had a number of chances in sudden death before going down. (Cue Sean McIndoe, cursing at nobody in particular.) What would have happened if a puck deflected a couple of inches one way or another? Would all that momentum have been for naught?
In any case, I think it’s an interesting topic to study: Are hockey teams that tie the game up in the final two minutes of regulation more likely to then win the game in overtime? First, I went through the regular-season box scores for each game over the past five years that went to overtime, and tracked the outcome of those that saw one (or both) teams score during the final two minutes of regulation. That gave me 260 games, of which 151 went to a shootout without any scoring during the five-minute overtime period.8 In the other games, the team that scored within the final two minutes of regulation9 went 52-57 in overtime, winning 47.7 percent of the time. That hardly suggests that their game-tying goal meant much for overtime at all. If you want to limit the study to goals from the final minute of action, teams who lit the lamp with 60 seconds or less to go went a virtually identical 31-34 (yes, also 47.7 percent) in overtime.
Now, what if I include the playoffs? That’s where it gets tricky, since overtime can presumably go on forever,10 which makes momentum pretty hard to link. Is it fair, for example, to say that Daniel Alfredsson’s goal with 29 seconds left in the third period of Game 3 against the Penguins in May gave the Senators the momentum to score a game-winning goal nearly eight minutes into the second overtime period? Probably not.
If I follow the same rules from the regular season and only consider a “momentum”-driven goal to be one scored during the first five minutes of overtime after a goal during the final two minutes of regulation, there just aren’t many games to look at. Even going back all the way to the lost season of 2004-05, there are only 12 such playoff games, but in those games, the team that scored at the end of regulation went 9-3 in overtime. At the same time, though, there are some games that raise questions about what momentum really means. Take the Hurricanes-Devils game from May 2006, in which the Devils took the lead with 21 seconds left only for the Hurricanes to equalize with three seconds remaining before winning it 3:09 into overtime. Is it fair to say that the Hurricanes were able to ride the power of momentum to an overtime win when the Devils couldn’t even keep that same power of momentum going long enough to hold a lead for 21 seconds? In any case, combine the regular season and playoff studies, and you find that teams who have scored at the end of regulation are a combined 61-60 in games that were decided during the subsequent five minutes of overtime. I don’t think that suggests that momentum means very much at all heading into hockey OT.
For whatever these studies are worth, none of them definitively proves that momentum doesn’t exist. What they each suggest is that, in a place where it seems obvious that there would be some record of momentum, little or no record of momentum exists. That can be up to a fault in the study’s methodology or the circumstances of the sample, among other things. In any case, if momentum were quite as obvious as the sports world makes it out to be in these situations, it would appear in one of the studies above.
Tomorrow, I’ll move away from the numbers and take a look at the underlying logic behind momentum, and how it can be used and abused to distort history and performance on a theoretical level.
Correction: An earlier version of this story had incorrect results in the table detailing the points per drive after possession changes.