What’s Real Madrid’s best XI under Zidane?


As we close the chapter on Zidane’s first 12 months in charge of the club, we review which player or combination of players have performed the best in net, defense, midfield, and offense. The methodology for assessing performance is based on the average goal difference or GD (goals scored less goals conceded) and shot on target differential or SoTD (shot on targets for less shots on target against) for each unique positional combo — how the team does while they’re on the field. For a more detailed overview of the methodology, see this article where positional combos were introduced.

Although this provides an interesting more team-oriented view than traditional individual statistics, it still suffers from the same issues that most statistics in sports and specifically team sports have (sample size and unequal/different environments/conditions being the two major examples). One way to address sample size is to limit the analysis to positional combos that have played at least 180 minutes (equivalent of two full games) -€” given rotations and injuries, combinations don’t play together too consistently.

In order to create a somewhat even playing field for considered combos (as some would still have played considerably more than others), GD and SoTD are normalized by calculating their per minute equivalent (i.e. +4 GD/SoTD in 180 minutes and +8 GD/SoTD in 360 minutes will produce the same value of 0.022). The resultant values, GD per minute or GDMA and SoTD per minute or SoTDMA is a comparatively fairer representation of performance than raw GD and SoTD. To standardize the data further, only Champions League and La Liga matches are considered as the other competitions (Copa del Rey, European Supercup, and Club World Cup) can significantly skew results.

The size of the bubbles denote minutes played and the color of the circles represent strength of opposition based on average league standings – the spectrum goes from 1 (darkest red) to 20 (darkest green). The analysis captures all official league and CL matches managed by Zidane in the 2016 calendar year.

 photo Keeper1_zpsoebbfgy6.png

For keepers, goal difference is a stronger and better indicator on its own as there is much less of a direct contribution to shot performance (although this assumption could stand to be tested given the recent focus on GK participation in build-up). Keylor Navas comes out on top but by a slim margin. The team averaged a positive GDMA of 0.019 with the Costa Rican compared to 0.018 with Casilla. The average strength of opposition for Navas was 9.15 (slightly stronger than Casilla’s 9.41) meaning quality of teams faced did not play a big role in Navas’ superiority.

Standout performance: La Liga 2016-March-05 Celta Vigo 7-1 Home Win | Score while on pitch [7-1]

 photo Defense2_zpsqv9zet2k.png

The unmatched and best defense is comprised of the Varane-Pepe center back pairing flanked by Marcelo and Carvajal. Going by minutes played, they have been Zidane’s third most favored defensive combo after the presumed starting quartet (Ramos taking Pepe’s place) going forward and the alternative with Ramos-Pepe n the heart of defense. The Marcelo-Varane-Pepe-Carvajal combo, benefitting from a relatively favorable strength of opposition rating, has registered an average GDMA and SoTDMA of 0.042 and 0.084 respectively. The nearest GDMA (0.033) and SoTDMA (0.049) values were achieved by two different combos emphasizing just how well the aforementioned best combo performed.

Standout performance: La Liga 2016-October-15 Real Betis 1-6 Away Win | Score while on pitch [1-6]

 photo Midfield3_zpsvxisfo7v.png

There’s also a clear winner when it comes to the midfield. Isco-Kroos-Modrić are far superior than all considered midfields. They stand at an impressive 0.033 GDMA and 0.064 SoTDMA. This far outstrips the numbers attained by Zidane’s undisputed preferred trio of Kroos-Casemiro-Modrić although, again, the strength of opposition comes into play as the latter midfield tends to play teams almost twice as good as Isco, Kroos, and Modrić -€” strictly based on league position on match day. There’s a strong argument that the football-centric (ball-focused, individual/collective possession control) style that the three very technically strong, press resistant players tend to employ can potentially work to great effect if fine-tuned further. As evidenced by their close to 300 minutes played together, they are functional and not completely defensively compromised in a competitive sense.

Standout performance: La Liga 2016-January-17 Sporting Gijón 5-1 Home Win | Score while on pitch [5-0]

 photo Offense4_zpsdawoanrk.png

The gap between the best offense under Zidane to date and rest is quite big, the biggest of any of the positions. By virtue of their automatic presence in the line-up and low substitution frequency, Bale and, especially, Ronaldo, were bound to feature in the best offense of the year. Although Benzema is the preferred striker and also has a level of minutes dominance, his relatively prolonged lay-offs in addition to the fact that there is a clear first rotational option (Morata) vs. Bale (i.e. can be either of James, Vazquez, Asensio, and sometimes even Isco or Kovacic with formational adjustments) mean there is a greater opportunity for other attacking combos that do not include him.

The manager’s best offense, by a distance, in 2016 was Ronaldo-Morata-Bale. Their strength of opposition in contrast to other offense combos was particularly advantageous. The toughest team they faced was Borussia Dortmund but only for a mere 6 minutes at the end of the game. Otherwise their opponents were ranked 8th, 9th, 15th, and 16th. That being said, it should be noted that a players can only play in the games they’re selected and the combo’s encouraging performances could justify further selection in the future. Morata’s intensity and speed adds another layer of complexity to the dynamism of Ronaldo and Bale. However, the struggle with interplay, that still exists to an extent even with a connector like Benzema as the striker, would be an issue.

Zidane’s best XI of 2016

 photo 2016BestXI_zpshkku0xkr.png

A note on the model

The idea behind using this analytical approach was to investigate the potential usefulness of team-based performance metrics. There is a myriad of information hidden beneath single-level individual statistics that can begin to be unpacked if the variables and factors influencing the numbers and raw productivity are identified and properly adjusted for. A striker can go on a run of 5 games without scoring and be doing excellently and similarly, a goalkeeper can have a save percentage of 80% but place the team in more risk due to poor clearances. Moving a step further, how a player performs changes based on who he plays with and who he plays against.

And what should be overlaid on this is how the player is used. He could be playing with the same players and against the same players as another option but is it in the same position? In the same role? The above only begins to scratch the surface of what a fully fleshed out comprehensive evaluation can look like but it at least shows that the possibility exists. How these ultimately correlate to the end result is the missing link right now. The additional work at this stage is data collection and manipulation as opposed to conceptualization. A coach should be able to quickly determine which line-up would work (at least which has in the past) best against a low-block defensive shape. If he wants to blitz them offensively, who are his best options? If he wants to hold the ball and be patient, who should be selected?