The Momentum Myth

The concept of momentum is bandied about during the post-season as a way for the media to explain and analyze the post-season developments and individual post-season series. Momentum is used as a default explanation for why any given team is winning any given series. When announcers cannot explain outcomes any other way, or want to infuse drama into an otherwise one-sided series, they can talk about momentum, shifting momentum, deflating momentum and other clichés.

While it is possible that momentum is a real thing, and that even at the Major League level, the results of a game played the previous day will have some real but undetectable impact on the next game, the data for this is scanty. Moreover, when momentum is used to explain everything in the post-season, it loses its meaning. More simply, if momentum can shift several times in a series, game or even an inning, it is not really momentum at all, but something else. For example, the San Francisco Giants surprised many this year by winning the World Series with a remarkable 11-4 record in the post-season. A lazy person’s explanation for this would be that as the post-season went on, the Giants gained increasingly more momentum. This is untestable, tautological and is not a real explanation of the Giants success. Moving away from the momentum paradigm might shed light on the real reason for the Giants success, which as everybody knows was largely their extraordinary pitching.

Teams advance in the post-season only after winning the previous series, so at any given time all of the teams in the playoffs have some degree of momentum making it possible to easily craft momentum based narratives. However, the data, at least at the level of individual games, suggests that momentum is not much of a factor. In the 2010 post-season, for example, there were 32 games played. Of those 32 games, seven were the first game of a series, so these can be eliminated from the sample. Of the remaining 25 games, the team that won the previous game won 13 games or 52% of the time. If wins were randomly distributed in these 25 games we would expect the team that won the previous game to have won either 12 or 13 games so, in this post-season, momentum had little discernible impact within each series. Data from other recent post-seasons is not radically different either.

Predicting the winner of a post-season game based on who won the previous game is about as effective a method of doing this as flipping a coin would be. Other methods for predicting post-season outcomes are comparably dubious. For example, home field advantage is very elusive in the post-season. In 2010, the visiting team won 19 of the 32 post-season games. This is somewhat due to the Rays-Ranger series where the home team lost every game, but even excluding that series, home teams were only 13-14 in the post-season. More predictably, the team with the better regular record won 19 of 32 post-season games this year. This is a bit surprising given the long season in which rosters change so much.

It is unlikely that analysts will stop talking about momentum because it makes for such an easy story line. If a team loses, one can point to the moment when the momentum shifted, if the team wins one can attribute it to momentum. This is probably unavoidable, but it should not be confused with a rational way to explain or predict post-season outcomes.