Playing time is first chapter in any statistical analysis story of a player. To simply earn game time is an accomplishment in itself; before a player can impact a game, he must see the field. Repeated playing time is a positive indicator of a coach’s belief in that player’s ability.
Whereas other soccer stats are influenced by position (attackers score more goals, defenders make more clearances), minutes played matters everywhere on the field. Comparatively speaking, minutes played also yields a larger sample size than other metrics. Here are the minutes played for all MLS field players in 2014 and 2015 (all data from MLSSoccer.com). The median playing time was 907 minutes in 2014, and 988 minutes in 2015.
A few things jump off the screen. There is more turnover each season than I expected. There were 178 players who left the league after 2014, and 185 players joined or re-joined MLS in 2015. Stated another way, one-third of the players in 2015 did not play in the league in 2014.
It’s also rare that a player reaches 2000 minutes year over year: just 59 players out of our sample of 723 did so. Also intriguing are the players in the upper-left of the chart. These players, such as Kendall Watson or Andrew Jacobson, played more than 2000 minutes in 2015 after below-average playing time in 2014.
Tallying minutes played doesn’t provide a full picture, though. DaMarcus Beasley also appears in the upper-left of the chart, but his mid-season return explains the lower minutes played total in 2014. Additionally, given the scheduling conflicts with the FIFA calendar, national team players will never lead the league in minutes played, so there isn’t a perfect linear relationship where more minutes is always better.
Looking at minutes played accounting for player availability provides context for how a team uses a player. Here, I modified the two metrics used to evaluate John O’Brien’s career:
- Involvement = Minutes played / (Games played * 90)
- Contribution = Minutes played / (34 league games * 90)
Isolating the 2015 player pool and plotting the two metrics for each player reveals more detail. The medians are 73% of minutes for games in which a player appears (Involvement), and 32% of minutes for the full season (Contribution). The shaded areas represent the upper and lower quartile for each metric.
A few player clusters emerge from this analysis:
Key contributors in the upper-right of the chart have high Involvement and Contribution rates. These are “first name on the team sheet” type of players. They start, play 90 minutes more often than not, and appear in most games. Bobby Boswell and Tyrone Mears top this group – each played in 97% of the team’s available minutes in 2015. National team players such as Michael Bradley or Kyle Beckerman also have Involvement scores above 90%, but their Contribution rate is around 70%.
Mid-season acquisitions in the lower-right of the chart have high Involvement but lower Contribution rates. These are players like Didier Drogba and Anibal Godoy, who joined teams in the summer transfer window and started immediately. Also in the lower-right are self-explanatory injured players, where guys like Kevin Molino or Ike Opara were starters until missing the rest of the season.
Squad rotation players are in the outer-left group. These players have an Involvement rate below the median, but a Contribution rate above the median. Olmes Garcia is a good example of this type of player: he played in 33 games, but started 15. Teal Bunbury, Kelyn Rowe, Juan Agudelo, and Diego Fagundez all appear in this cluster, as each reached the threshold of 1700 minutes across 30 games played and 19 starts for New England.
Career Beginning or Ending are in the lower-left group. These are players who receive few minutes in a few games, and are probably starting out or hanging on. For example, LA had Todd Dunivant, Edson Buddle, and Ariel Lassiter in this cluster.
Filtering by team also adds retrospective insight. Portland had its starting back four and the tandem of Darlington Nagbe and Diego Chara in the top quartile for Involvement and Contribution. Supporters’ Shield winners New York had three midfielders and Bradley Wright-Phillips in the same category. Columbus had Kei Kamara, Michael Parkhurst, and three midfielders as key contributors. Recent work from Daniel Altman hints at the benefits of player familiarity.
The Involvement and Contribution rates provide a framework for future analysis. We can use these metrics to get a quantitative sense of player development, or at least player opportunities (more work is required to judge keepers.)
For future considerations, it’s also easy to filter by type of player to monitor progress season by season. What happens to those who are homegrown players instead of draft picks? Where on the aging curve or experience curve should players expect to breakthrough (on a surface level, it seems like the third or fourth season)? Here is an example of Matt Miazga’s evolution over the past three years.
Finally, with the framework’s simplicity, we can apply Involvement and Contribution rates to player performance in other leagues. I plan on exploring the international careers among those with MLS experience further in the future.