The role of attacking players in football has evolved over the past two decades. Often, this either reflects or stands out from their respective teams’ playing style. While finding the back of the net will forever be the most valuable asset to a team, modern football demands more from a traditional attacker.
The Premier League has a wide variety of attackers with specific roles in their respective teams. Larger tactical trends, such as inverted wingers and gegenpressing, which have come more into fashion, have permitted the more creative attributes of an attacker to assume a more conventional role.
This data analysis aims to investigate critical statistics in the playmaking of attackers in the 2020/21 Premier League season thus far. Our data includes traditional number 9s, wingers, as well as number 10s. We employ a k-means clustering algorithm to divide the attackers into different cohorts.
This analysis is relevant with respect to quantifying an attacker’s contributing output in their team. Furthermore, the clusters suggest a potential for positional changes tactically, as well as provide a foundation for scouting by other clubs.
Tactical relevance and exploratory data analysis
Traditional roles in football involve midfielders taking the majority of the creative responsibility by creating key chances and being the primary assisting figure in a possession chain. However, assists in the Premier League are currently dominated by attackers.
Tottenham Hotspur‘s Harry Kane has been the epicentre of headlines with an outstanding eight assists in eight matches. There are several reasons for the Englishman’s newfound creative dimension, resonant of Dennis Bergkamp. For instance, his routine vertical rotations with Son Heung-min have become a deadly tactic employed by Jose Mourinho.
In the above position, Kane drops as deep as the halfway line. His ability in half-spaces permits him to consolidate his body positioning as he exposes Southampton’s high defensive line with a correctly weighted through pass to Son. The awareness of his counterpart’s positioning is integral in breaking lines and scoring.
Additionally, Spurs’ disciplined midfield three provides a foundation for their attackers to be dangerous during transitions. Pierre-Emile Hojbjerg, for instance, places in the top five for total passes per 90 with 72.16. Tanguy Ndombele further tops the league with the most second assists (three). This implies that Tottenham’s midfielders play more of a metronomic role that encourages their front three to engage in the more tangible creative actions.
Tottenham’s example captures the underlying tactical trend of a playmaking attacker. This is in an era where accommodating a traditional number 10 comes at a high cost. Therefore, attackers that are versatile enough to embody traits of a number 10 while successfully doing their respective positional duties i.e. of a number 9 or winger are highly valued.
The aforementioned tactical trend provides an insight into an uneven distribution of attackers in the league. The following visualisation sheds light on the attackers’ creative output with respect to their team’s style of play.
The concentration of players on the bottom left in conjunction with lines for the mean and median quartiles point towards a handful of standout players. The players are divided into four clusters. Kane and Jack Grealish fall in the top 1% in terms of creative output.
This is especially relevant considering that their respective teams don’t rank highly for their total passes per 90. While Spurs are mid-table, Aston Villa rank 18th for total passes per 90 (316.59). This implies that Kane and Grealish produce high creative output despite an overall lower volume of passes made by their respective teams.
Attackers in the yellow and red clusters must be analysed with more granularity. We shall do this by measuring creativity in terms of ball progression and defensive output. The former comments on the attacker’s creative output on the ball. The latter comments on the attackers’ role in their teams’ pressing – a weapon widely accepted by analysts to be an effective chance creator in transitions.
A major aspect of creativity and chance creation involves breaking defensive lines. This can be achieved either by passing the ball in between the lines or by physically progressing the ball by dribbling. Mapping a player’s output through the metrics of progressive passes and progressive runs illustrates his style of breaking defensive lines. Consider the following image.
Creating the above spectrum provides insight into the different types of attacking players. Hakim Ziyech and James Rodriguez are notably inclined towards making more progressive passes (10.49 and 7.93 respectively). Conversely, Pedro Neto and Richarlison have an affinity to progress the ball through a high amount of progressive runs per 90 (4.04 and 3.78 respectively). Grealish is, once again, an outlier, accompanied by Adama Traore.
The quality of both metrics must be separated from their volume. In terms of physically breaking the lines, consider the following image.
Quality of output separated from its volume can be observed in the distance between Grealish and Traore in the above visualisation. Through these metrics, Pedro Neto and Liverpool‘s Mohamed Salah are clustered with Grealish. Moreover, while Wilfried Zaha and Sadio Mane record a similar volume of dribbles, they are clustered with the likes of Richarlison and Arsenal’s Nicolas Pepe. A granular difference in output is irrelevant individually, but tends to add up in the long run.
In terms of distribution, the following visualisation maps the quality of the attackers’ line-breaking passes.
Although Ziyech and Alex Iwobi have a notably low number of appearances, their style of chance creation is pass-orientated. Salah’s 1.26 key passes per 90 stand out from the likes of Riyad Mahrez, Roberto Firmino and Diogo Jota. This is noteworthy, considering his team’s relative possession-centric playing style. Furthermore, although Kane and Rodriguez play for teams with a comparatively lower tempo and pass volume, they record a similar output, matching the same value as almost half of Salah’s passes per 90.
The contrast in tempos of teams in the Premier League can be illustrated through an attacker’s distribution in the final third. This is illustrated in the following image.
The cluster with the highest creative output – with the exception of Grealish – consists of players playing for teams that like to dominate possession and have the majority of passes. In the second (red) cluster, Anthony Martial and Kelechi Iheanacho stand out. Their deep completions and forward passes are particularly impressive, with respect to their teams’ ball progression.
Both Manchester United and Leicester City play a mixed style in terms of their passing and pressing tempo. The above visualisation suggests that Martial and Iheanacho, who normally play central number 9 positions, have the potential of playing as pivots higher up the pitch. If paired with Edinson Cavani and Jamie Vardy, both players would thrive as second strikers with the right positional placement by their managers.
In a system-dominated Premier League, it is often relevant to consider the influence of the player’s team on the attacker, as it may provide insights into scouting them.
Jürgen Klopp famously mentioned how successful pressing and counter-pressing higher up the pitch is a highly potent playmaker in and of itself, and a team’s attackers are their first line of defence.
A successful offensive duel won by an attacker means that they either won possession for a transition or a set-piece in a dangerous area. An unsuccessful offensive duel may have the potential to trigger a counter-press by controlling the outcome of the second ball. Therefore, a high work-rate for the attacker contributes to creating chances for their team.
The above image maps said work rate of the attackers in the Premier League. Dele Alli’s outstanding defensive output as a supporting striker is caveated by his lack of game time. Moreover, despite failing to record significant attacking actions, Alexandre Lacazette has more successful defensive actions per 90 than Mane and Neto, with 6.59. A combination of the former and the latter can be observed in Jota’s impressive start to the season.
The following image compares the volume of an attacker’s attempted offensive duels to their success rate.
Crystal Palace’s playing style offensively and defensively can be understood by the offensive duels of Jordan Ayew and Wilfried Zaha. They are integral to not just getting in front of the opposition defensively, but also winning set-pieces on the break while on the ball.
Furthermore, although Mane and Salah engage in roughly the same amount of offensive duels, the Senegalese has won more battles this season. Finally, Grealish has one of the highest duel success rates higher up the pitch, which further underlines his importance to his team.
Reflections and scope
The goal of this data analysis was to identify different clusters of attacking players better suited to the different facets of chance creation. Our analysis had several findings.
Firstly, a high volume of attempted ball progression doesn’t necessarily result in key playmaking output. This was apparent through a rather metronomic output of several system players on one hand. On the other hand, players such as Kane and Rodriguez are the epicentre of key moments of their teams’ chances. Given his current form, Kane is arguably the most complete number 9 in the league. Despite placing around the median for his volume of passes and dribbles, his expected assists (xA) and key passes underline his inherent quality.
Moreover, Grealish is a creative output merchant. He has consistently placed in the top cluster of every key metric we analysed. His directness is essential in Aston Villa’s solid form this season. Grealish, often positioned in wide areas for his team, resembles a modern number 10 with the right statistical output. He could, on paper then, thrive on either side of a front three.
In the second clusters of each metric, there exist statistical similarities between the outputs of Mane, Salah and Pedro Neto. Neto has strung together more than five starts for the first time for Wolves. This has resonated throughout the team, as his interplay with Nelson Semedo on the right-hand side has been noteworthy. The Portuguese’s progress may be monitored by larger clubs as a result, in a manner similar to Liverpool’s signing of Diogo Jota.
For future analyses, our clustering algorithm could be further optimised by focusing on more granular hyperparameters.
In this exploratory data analysis, we highlighted how attackers in the Premier League contribute to a non-goal output through different key metrics related to chance creation.
While one section of our dataset function in tandem with their team’s style of play, certain outliers tend to define their team’s style of play.
The contrast between both types of attackers facilitates the potential for tactical innovations. These may be positional changes and integration into a wide variety of playing styles.