With or Without You Analysis of Real Madrid

In hockey, ‘With or Without You’ (WOWY) analysis measures how players perform (using shot differential) when they’re on the ice with and without a particular player. In football, we have far fewer shots, so when looking a data from a single season, we couldn’t draw any meaningful conclusions because of sample size issues. However, if we compare how the team performs with and without the player, the sample size should be sufficient, providing the player hasn’t played too many or too few of the possible minutes played.

This immediately creates a problem that WOWY analysis is supposed to eliminate – differentiating between players who are responsible for an improved shot differential and those who are fortunate enough to play with those players, but don’t have an impact themselves. This is a significant problem – looking at the chart below, Ramos has clearly performed better than Pepe, but Pepe still has the 4th best shot differential on the team. Pepe played more than 80% of him minutes with Ramos and yet Ramos’ shot differential was almost twice as large. In the 459 minutes Pepe played without Ramos, his shot differential would have ranked 4th worst in the team.  With that swing, we can probably say that Pepe is just a passenger and benefits from playing with more talented teammates – but the sample size is hardly definitive and this is an extreme example.

Two more key pieces of context missing are game state and quality of competition. For the latter, I’m comfortable saying that it’s not a concern for a player who’s completed more than 1250 minutes. It’s unlikely a player would play more minutes and be poor enough to merit intentional sheltering from stronger opposition. As you can see here, the spread of quality of competition will naturally narrow as the minutes increase (with a larger dataset, it would resemble a funnel plot). The relevance of game state will be similarly linked to minutes played (we’re only comparing within a team), although I don’t know at what point it becomes an issue.

This was the main reason I picked Madrid for this analysis – the player with the most minutes played fewer than 80% of possible league minutes and there were 21 players in all who had played more than 20%. Even among other big squads, this is rare. I excluded players who fell outside those (20-80%) parameters.

Bubble size is minutes played. N.B. The scale of the y-axis is inverted. The embedded chart shows minutes played (x-axis) and relative shot +/-  per 90 (y-axis).

Bubble size is minutes played.
N.B. The scale of the y-axis is inverted.
The embedded chart shows minutes played (x-axis) and relative shot +/- per 90 (y-axis).

Ronaldo has the greatest impact on shots for, as you would expect, but also has the second greatest influence on limiting the shots against – something Madrid’s other attackers seem to have little impact on. This could be because teams defend deeper and so shoot less to cope with Ronaldo, or it could be because there are a finite number of events that can occur in 90 minutes, and with Ronaldo on the pitch, more of those opportunities are taken by Madrid. Ramos, along with Pepe (a passenger?) are the only defenders to have a significant positive impact on shots for. Albiol was as bad as Ramos and Ronaldo were good, performing poorly in both measures. Despite Madrid having signed two players (Modric and Illarramendi) to replace Alonso, he is still one of the strongest, notably outperforming Modric.

Varane, Marcelo, Kaka and Carvalho all played fewer than 1250 minutes and so could have been affected by quality of competition or game state.

Rel Shot +:-

This shows the shot +/- relative to the rest of the team – in absolute terms every player had a positive shot differential. This is useful to compare players in the same position – the difference between Ronaldo and di Maria, for example, is almost 13 shots/90, whereas, little is gained or lost whether Ozil or Modric is played. I included Casillas and Lopez, not because they have an influence on their shot differential, but because they roughly represent the first and second halves of Madrid’s season.

It’s the transfer season and Madrid have already spent nearly £60m, also selling a handful of players. Interestingly, their acquisitions and sales line up rather well with what we see in these charts. The four worst players in relative shot differential are Albiol, Carvalho, Essien and Kaka. Essien’s loan has expired, Carvalho has moved to Monaco and they managed to get actual money for Albiol. Having played fewer than a thousand minutes, the time has come for Kaka to choose between money and the chance to play in a ‘home’ World Cup.

If we look at the 4 players with the most minutes and a negative shot differential, we have Arbeloa, di Maria, Khedira and Higuain. So far, Madrid has bought Carvajal (Arbeloa), Illarramendi (Khedira), and Isco (di Maria). It’s been widely reported that they are trying to sell Higuain to fund a move for Suarez. They’re spending a lot, but in the right places, while shedding salary from their poorest players.

With the arrival of Illarramendi, Madrid try again to solve their midfield problems. Throughout last season, Mourinho tried (largely in vain) to find a permanent spot for Modric – a player whom he had specifically requested. He played just under half of his minutes with Ozil and the same with Alonso, but just 21% with both. Similarly, he played just 21% with Khedira and Alonso. Combined, the four of them were substituted a total of 58 times in the league. This meant that Modric didn’t complete more than four games with any of his midfield partners. He managed 10 games with more than 60 minutes with Ozil and 9 with Alonso. He played just 4 games with the pair of them for more than 60 minutes. The four players were also frequently subbed at half time – clearly, while they are all talented, they don’t fit together.

The addition of Illarramendi and Ancellotti makes the situation more complex – instead of fitting four players into three positions, we now have five and there is a new decision maker. At Chelsea, Ancellotti used a combination of Essien, Mikel, Ramires and Lampard. Two box-box midfilders, a defensive and an attacking midfielder. At PSG, he relied heavily on Matuidi, with Verratti, Chantome and Motta providing the support. The constant between the two is an energetic box-box presence. Alonso is the player who best fits that description, with Khedira bringing energy without having a significant impact at either end of the pitch. Illarramendi is the natural fit beside him – a talented passer (enough to minimize the effectiveness of man-marking Alonso), who will circulate the ball and effectively screen the back four. With Ozil likely to be 1st choice at no. 10, Modric could once again be the odd man out, with Khedira shuffling further down the pecking order.

Potential Madrid XI for 13/14:


With a larger dataset, WOWY data could be used as it is in hockey – to differentiate between players who are ‘driving the bus’ and those just along for the ride, as well as identify combinations that show tangible chemistry. Even with the limited dataset I have used, it is still useful, largely confirming the performances you would expect and aligning nicely with Madrid’s transfer activity.


Leave a comment

Filed under Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s