Who can beat Kevin De Bruyne in terms of passing style?
Bruno Fernandes? Henderson? Kovacic? Grealish? Or John McGinn? We use tracking data to find out.
Who is the best passer of the ball in the Premier League? When it comes to midfielders there is (according to the numbers) probably only one correct answer: Kevin De Bruyne.
Last season (and the season before) De Bruyne passing and dribbling topped all expected threat — which measures how a pass increases the chance of a team scoring––models per minutes played. Here is an attack map of his most valuable passes.
This is the reason De Bruyne employed a data analytics company in order to help negotiate a new contract. The numbers are guaranteed to make him look good…
But De Bruyne can’t be best at every aspect of passing the ball?… or can he?
To find out, at Twelve football, we developed an expected threat model which uses tracking data to give context when a pass is made, along with event data to know where the pass moved the ball. These models take the basic expected threat idea to a new level: they include the position of the opposition when a pass is made, the speed of teammates, and context around the type of pass (cross, long ball, cutback). They allow us to find out what makes a pass valuable .
Before we start looking at individual players, lets take a peak under the hood of our model. Below is a visualisation of a very good pass.
This pass breaks two lines of defence (red attacking, blue defending) and is received near to the goal. Our machine learning model evaluates this as a 0.112xT pass (it increases the chance of a team scoring with 11.2%). But why does the model think it is good? Part of the answer to this question can be seen from the numbers in the figure above: these tell us how much closer the pass moved the ball to the goal, various measurements of how many opposition players the ball went by. Each of these numbers, and combinations of them, are then weighted (based on training a machine learning model) to find out how important they are in predicting a goal as a result of the pass.
For the pass in the figure, perhaps unsurprisingly, it is closeness to goal and number of opposition players taken out by the ball, which makes it so valuable. We do similar weightings for the speed of movement of the players: successful passes made when everyone is running towards the goal are almost 3 times as likely to result in a goal as passes made when players are stationary.
So… what we can then do with this model is look at which players produce more value in different situations. The top 5 attacking midfielders in terms of increasing value in terms of moving the ball to more dangerous areas last season were:
1, Bruno Fernandes (3029 minutes)
2, Kevin De Bruyne (836 minutes)
3, Jack Grealish (634 minutes)
4, Martin Odegaard (908 minutes)
5, Mason Mount (696 minutes)
So Bruno Fernandez takes this one. He was also top in terms of playing past opposition players (a term sometimes referred to as packing). Here is his radar, showing how he ranks in all passing attributes.
De Bruyne also played in a central midfield role (1141 minutes) last season and here he is in fourth place overall.
1, Jordan Henderson(1034 minutes)
2, Mateo Kovacic (1797 minutes)
3, Thiago (1920 minutes)
4, Kevin De Bruyne (1141 minutes)
5, Ilkay Gundogan(2082 minutes)
In this position Henderson remains formidable: topping the packing and location ranking, and also ranking highly in making high velocity passes (successful passes when both teams are moving at high speed during, for example, counter attacks).
Jack Grealish plays both attacking midfield and left winger and excels in both positions. But comparing his radar with De Bruyne we see exactly how well the pair complement each other. Grealish excels at balls when the players are running fast (velocity) and cutbacks. De Bruyne tops the rankings for crosses, long balls and through balls.
So who are the (maybe not so well) hidden gems in the Premier League? The one that stood out for me was Aston Villa’s central midfielder, John McGinn. Fourth on the list for value produced by long balls, he isn’t quite at De Bruyne levels, but given he plays in a non-big-six team, he is performing very well. Dani Ceballos was good at Arsenal too. And if you believe the rumours, he might be on his way to a new club.
In scouting, context is everything. We need to be able to rank players in terms of how they contribute to a team. At Twelve football, we have developed a whole range of metrics, which combine a measurement of value of an action (based on expected threat) along with context (what was happening off the ball when an action occurs) and are now supplying these to clubs. We believe that we are the only analytics company who are offering easy to use combined tracking data and event data player rankings.
OK. Thats enough of the self-promotion. The most important thing to say, is that has been lots of fun to work on this with Jernej Flisar at Twelve … and there is a lot more to come!