One of the most tangible changes Zidane made was the introduction of his rotation scheme that was far more intensive than past coaches. The manager has been credited with installing the current approach to allocating minutes to the squad which sees him deliberately rest key players when opportunities present themselves. Furthermore, Zidane has been unafraid to field fringe players such as youth/academy products or less prominent members of the roster.
“For you it might seem brave, but for me it is normal[.] I see them training every week, and when choosing the team I have the possibility to change many players.” - Zidane
However, despite the plaudits and acknowledgement the manager received from both players and the press, not everyone benefited or bought into his system. In 2016-17, players such as James Rodriguez, Coentrao, Danilo and even the incredibly exciting but ultimately unproven Mariano Diaz were perceived victims of Zidane’s choices. This season, criticism has emerged that Zidane’s rotational plan is less balanced with Ceballos and Llorente being particular points of focus.
“I expected to play more, it’s true. My intention is to continue working, get the most out of myself so that when the coach gives me opportunities I can take advantage of it.” - Ceballos
Real Madrid have played 48 games to date over the course of the 2017-18. This article will examine how the rotation in the current season compares to the previous two.
Distribution of minutes
After 48 games, there were roughly 4,500 minutes accumulated in total game time in all competitions in each of the last three seasons. The high points in terms of minutes played has dropped drastically going from 92% of minutes played (Ronaldo) in 2015-16 down to 76% (Kroos) in 2016-17 and 70% (Casemiro) in the current season.
With regards to distribution, the team minutes have been less spread out across the board. The more compressed the box plot, the less variance there is in minutes played by the players. The border of the shaded areas represent the median value between the lowest/highest point and the overall center of the entire distribution. Of special note is the dramatic effect Zidane has had on rotation of forwards.
General variance of minutes played
Another great measure of balance is standard deviation which calculates the average difference between each player’s minutes tally to the mean. It provides a good (albeit cursory) picture of how disparate the allocation of minutes has been. In theory, the more extreme or skewed the distribution (more players towards max and min), the higher the standard deviation.
The above confirms what many have believed. Zidane’s team selection has been more equitable as the overall standard deviation fell between 2015-16 and 2016-17 and slightly further in 2017-18. From a position perspective: goalkeeping and offense have benefited the most from the shift. Midfielders and defenders have actually become less evenly distributed over the last three seasons — supporting fan grievances about Ceballos and Llorente to a degree.
Squad utilization and concentration of games
In order to determine approximately how “fresh” or minutes heavy the match day team is, the percentage of total player minutes earned by players who participated in the prior match was calculated for each game day. The resulting output was then smoothed by doing a rolling 5-match average.
Again, the recurring theme is evident as although the beginning and end of the 48-game sample are relatively similar for all three seasons, there is a noticeable dip between around game-day 29 through 41 in 2017-18. This stetch amounted to 13 games (eight La Liga, four Copa del Rey, and one Champions League). Conversely, there is somewhat of an increase in this same period for 2015-16 as Real Madrid exited the Copa del Rey early in that season and there were presumably less opportunity/need for rotation.
The above charts provide a more pointed look at minutes concentration by tabulating the ratio of four or more consecutive appearances made in incremental groupings. The charts show that there has been a persistent shift to giving more match-day breaks/rests to players as the number of consecutive appearances has declined (especially in midfield and goalkeeping). Using the four game cutoff was arbitrary based on a general assumption that three consecutive games is relatively standard and because a huge portion of the sample fell in this range (three or less consecutive games), it would have clouded the results.
Attacking effect and utilization of substitutions
The below table provides a snapshot of how substitutes have been used in the last three seasons and their comparative attacking output.
The table shows that there hasn’t been any material differences in usage of subs over the three seasons. However, it seems that substitutes have generally contributed more under Zidane than in 2015-16 although there is no discernible trend or correlation.
Conclusion and summary
It is clear from reviewing the data from the three seasons to see the impact Zidane has had on the rotational and squad utilization scheme. Furthermore, despite protestations that the manager has been less balanced in minutes distribution in 2017-18, the evidence suggests that is not true. While there are certain positions that have not been rotated as much, player selections have not been less varied this season. As a summary of current state, the below depicts the trend over 48 games by game day for 2017-18.
Immediately apparent is that Zidane uses the Copa del Rey as a staggering mechanism which helps relieve some of the strain felt during a crunch period of the schedule. What’s also interesting to note is the symmetry that is visible — the players are somewhat grouped together in the minutes hierarchy by position suggesting more evenness in spread. Compared to other seasons, the number of players used in all competitions has remained largely the same (26 in 15-16, 27 in 16-17, and 28 in 17-18).
In conclusion, while the analysis is insightful and helps validate certain theories, it is important to remember that it is limited by data availability (scope/sample - could only do three seasons) and considerations. For example, injuries and suspensions weren’t isolated and as a result could be influencing the statistics in ways that were not recognized.