We recently realized that many are unaware/confused about Pass Maps or interprete them in a completely wrong way. This article hence focuses on how to read a Pass Map, what it tells, and what it doesn’t tell and thus how to effectively use a great mathematical tool to interpret the beautiful game.
To begin with, let us have an idea about what the 11tegen11 Pass Map, the one we frequently post, looks like. It basically shows the AVERAGE TOUCH POSITION of a player, represented by a Dot. The bigger size of the dot means more touches. The arrows indicate the lines of 5+ passes between the two players, arrowhead showing the direction. As the number of passes between two players increases, the thickness of the arrow increases. So that’s the vague idea.
THE AVERAGE POSITIONS
Suppose a player touches the ball at many different locations on a field. IMAGE 1 shows the example. So, when recorded it becomes a set of randomly scattered points. The average position of the player is determined from these positions. Other two important parameters are the spread and skewness of data. So, as in the example, full back’s data has much more spread than that of a Goalkeeper.
The triangles you see on the two sides of a circle, pointing outwards tell about the direction in which the data has more spread. So a Fullback’s circle having triangles above and below the circle means the data is spread vertically. So basically, he was marauding up and down the wing. The difference in role will imply difference in spread. An attacking midfielder, most probably one in a 4-2-3-1, may have some spread in all 4 directions, whereas a keeper’s position will usually have the least spread in data.
Now, think of Manchester United under Ferguson when Wingers always exchanged the flanks. In such a case, if they keep on doing it time and again, their circles will end up being more central. This is because, as both wingers have significant number of touches on other wing as well, their average position will shift towards the centre. A classic example is when Manchester City players switched position so often to compensate Walker’s red card, they ended up having the average positions in the centre. (Image 3).
And hence this DOES NOT necessarily mean lack of width.
Another problem- the positions recorded are the positions where the player touches the ball and/or passes it to other player. A player is on the ball for average 3 minutes in a game. What about rest 87 minutes? Data providers like OPTA do not track players in a way so that it can be used and Incorporated in pass map. The only solution is a Heat map in such case. The pass maps made in this way, thus actually DO NOT measure off the ball activities.
The second main component within the network refers to the passes between players. In the passmap the thickness of the arrow represents the number of successful passes. A thicker arrow equals more successful passes. There is a visual sense in the number of successful passes.
The total amount of successful passes is an outline of a team’s game play. In what way is the build-up play organized? Who is targeted in midfield? In answering these questions the pass map focuses on the aggregate number of passes. So, Barcelona’s pass map against Manchester United in CL 2011 final shows the incredible amount of passing in the midfield and thus tell us how Barça dominated the centre.
We do not however see the difference between dangerous passes and the ones that are not. It doesn’t differentiate through balls, crosses and direct passes. Moreover we take into account passes that were completed, but what about passes that were not? They stay out of the picture.
Most importantly, the arrows from one average position to other do not actually represent the direction of the pass. A player may make 5 different passes to a particular teammate, all in different directions, but in an actual pass map, it will only be represented by one arrow from one dot to the other.
But what happens when we decompose the relationship between passes and position? Do we get a fairer view of the average pass? Definitely! Does the image become clearer? Not really, because the passes are not completed between the player’s positions anymore. However, a positional and passing network is not meant to focus on only one player but to analyze the collective organization. This way it is impossible to get eleven players and their corresponding interactions into one clear image.
So, when we read a Pass Map these are the things to be kept in mind. A Pass Map tells only a limited number of things out of thousands of measurable parameters, and even less things about a player individually. However, since it speaks about the network as a whole, it tells us about the overall shape of the team, about the areas the team dominated and about the AVERAGE, OVERALL passing structure. Keep this in mind very well.
“This incredible fact that a discovery motivated by a search after the beautiful in mathematics should find its exact replica in Nature, persuades me to say that beauty is that to which the human mind responds at its deepest and most profound.” -Subramanyan Chandrasekhar
So, is there any ideal, beautiful Pass Map? Probably, I would say the 2011 CL final one for me, as the players are well connected, centre is dominated to an unprecedented level, and well, to my Mathematical eye, it looks good. In general, roughly, a good Pass Map has well connected players, good structure and effective usage of space. After 2014 we drifted more and more away from OUR style, and so did the Pass Maps. So what should we look upto under Ernesto Valverde? The Pass Map against Juventus, attached here, has so far been the best one. If Valverde manages to establish a proper controlling and compact midfield, his tenure can be termed a fruitful one. And he cannot afford to neglect this.