Several years ago, I wrote about the importance of continuity in a team’s lineup over the course of the season. The piece has since been taken down (it will soon be republished on this site), but the thrust of the argument was that the most successful teams in Major League Soccer were able to identify a core group of players who played a significant amount of a given season together. Teams that couldn’t, or didn’t, coalesce around such a core were less likely to be successful.
Over the past several weeks, I’ve been re-visiting that thesis using some alternate strategies to see if they continue to hold true.
The chart at the top of this piece compares two values: lineup continuity and points per game. Lets look at each in turn.
The instrument I’m using for this is a Herfindahl-Hirschman Index (HHI), which you can find defined via Google (or just start with Wikipedia). I came across its application to soccer via a piece by Dan Altman examining league quality, and realized that it could probably be applied to the question of how teams distribute playing time along their roster.
In this approach, players competing for time on the field can be seen as a market, and the number of minutes played by a given player is analogous to that player’s market share.
I’ve made a slight change to how this index is presented here, however. Normally HHI figures are displayed from 0 to 1, with a value of 1 indicating a monopoly while a value of 0 indicates an infinite number of small competitors. With eleven players on the field at once, however, a soccer team will only ever get to 0.090909(…) as an HHI score – and that will only happen when the same eleven players play an entire season, every minute of every game. As such, I’ve recalculated HHI to use this as “100%”, with all other values correspondingly lower.
Points Per Game
While the Herfindahl-Hirschman Index is not universally known in soccer circles, Points Per Game is much more familiar. Given that the season is still in progress, with various teams having played between five and 11 games, it makes sense to try and find a common reference frame for competitive success that is less biased by the number of games a team has played. While I’ve considered using goal differential in this role, for now I’m sticking with the more results-oriented PPG.
Right now, there exists a relatively weak – but still noticeable – relationship between lineup continuity and competitive success. My first look at this issue, from late April, found an r-squared value of 0.3986, but that value has now dropped to 0.2406.
The experiences of teams like Portland and Dallas help explain this drop. The Portland Timbers have had the most consistent lineup in the league so far, but have earned a mediocre 1.3 points per game and currently lie outside the playoff berths in the Western Conference. Meanwhile, FC Dallas is one of the best teams in the league at earning points (2.0 points per game), but have changed lineups relatively frequently.
Despite these outlying data points, however, the experience of many teams around MLS continue to support the thesis that a stable lineup correlates well with competitive success. At the lower left of the plot are the struggling trio of Montreal, Philadelphia and New York City – each with a rotating cast of players and few points earned. On the opposite end of the spectrum, then, you have elite teams like New York Red Bulls, DC United, and the Seattle Sounders.
The next look at this metric will come sometime in June, as the season nears the midway point. It will be interesting to see whether outliers like Portland and Dallas continue to buck the trend, or whether they fall in line with the expected narrative.