Over the past season in the Premier League, Olivier Giroud, Wayne Rooney, David Silva and Danny Ings all scored 10 goals from open play. Yet anyone who watched these players would probably characterise their playing styles as largely differing from one another, and would likely not consider them equally skilled at scoring goals.
Obviously this is a bit of a contrived example – the idea that goals total does not fully describe a player isn’t a revolutionary thought. Quantitatively, however, it can be hard to put a finger on what exactly makes Giroud’s 10 goals different from Rooney’s or how much of Ings’ production we can expect to translate to his first season at Liverpool. One of the ways we can differentiate between how two players create and score their chances is by examining how those chances are created.
In examining the origin of chances, I looked at 14,252 shots from open play over the past two seasons in the Barclays Premier League. Of these, 11,622 are given qualifiers in Opta’s data which indicate that they were assisted. At a glance, a league average assist rate of 81.55% seems rather high. Here it’s instructive to consider how the data is recorded. The qualifier for an assisted pass is given for any pass which precedes a shot directly. This means that there is little subjectivity at play – a player either received a pass and attempted a shot before they ceded possession of the ball, or they did not. However, it also means that passes which might not subjectively deem assists are given as such.
Consider last season when Stoke City’s Eric Pieters’ header found Mame Biram Diouf on the edge of his own box following a Manchester City corner, before he embarked on a 70 metre solo run to score (see below). If we’re thinking of assists as passes which meaningfully contribute to a player’s ability to shoot, this does not meet our criterion.
Fortunately Opta analysts also record a qualifier indicating whether an assist was ‘intentional’. This introduces an element of subjectivity into the data, as judgement of intentionality has no definitive marker in the way that a completed pass does. There are 5,935 shots from our sample which were deemed to have been intentionally assisted, for an intentional assist rate of 41.64%. Amongst 40+ shot seasons for players over the past two years, intentional assist rates range from 8% to 78% as opposed to 62-93% for all assists.
Though there is certainly subjectivity involved in crediting intentional assists, the descriptive statistics surrounding intentional assists consistently capture what we’d think of as assist more accurately. With this in mind, my analysis from here will refer to intentional assists only.
Looking at the rate at which a player’s shots are assisted allows us to characterise their playing style and contextualise raw shot numbers. Over the last two years, here are the top and bottom five players with at least 40 shots in a season ranked by the assist rate on their shots.
Assisted shots: bottom five
|Andros Townsend||Tottenham Hotspur||2013||49||8.16%|
|Jason Puncheon||Crystal Palace||2014||53||11.32%|
|Charlie Adam||Stoke City||2013||44||13.64%|
|Riyad Mahrez||Leicester City||2014||50||14.00%|
Assisted shots: top five
|Nikica Jelavic||Hull City||2014||41||78.05%|
|Kevin Nolan||West Ham United||2013||41||70.73%|
|Roberto Soldado||Tottenham Hotspur||2013||44||70.45%|
|Peter Crouch||Stoke City||2014||45||66.67%|
Shots do not occur in a vacuum. In fact, assisted shots tend to be of higher quality and find the net more often than unassisted shots. The conversion rate of assisted shots from open play is 14.09%, more than double that of unassisted shots at 6.71%.
Players with 40 or more shots in the Premier League, 2013/14 – 2014/15
This suggests that assists help put players in positions to take high quality shots. This information makes Olivier Giroud’s 10 goals from shots which were assisted 60% of the time look markedly different from Nacer Chadli’s 10 goals from shots assisted nearly half as often (32.56%).
For a demonstration of how team environment matters to a high assist-rate player, look no further than Roberto Soldado. In his final year at Valencia, Soldado had 14 goals from open play and 24 in total, good for 3rd in the league amongst players not named Ronaldo or Messi. He did so being assisted on 72% of his shots, a rate higher than all but two players in the last two Premier League seasons. Though Valencia weren’t generating a vast percentage of their shots via assists (36.96%), in moving to Tottenham he joined a team assisting shots at a rate a full five percentage points lower in his first season at 31.79%. His assist rate stayed high at Tottenham at 70%, but, reliant on assists in a team not primarily concerned with generating them, his performance suffered. The quality of his assisted chances at Valencia, as measured by the average expected goal value for his assisted shots, was 0.194. At Spurs the following season, it dropped to 0.162, and his conversion rate on assisted shots plummeted from 22.22% at Valencia to a ghastly 6.45%. Though luck (both good at Valencia and bad at Spurs) had its role in Soldado’s precipitous drop-off, he seemed a mismatch for the offensive system at Tottenham from the start.
We can also see the effects of within-team style changes on players who are dependent on assists. In the 2013/14 season Aaron Ramsey scored nine goals from open play. During this season, 54.76% of Ramsey’s shots were assisted, and Arsenal as a team assisted shots at the highest rate in the league (46.95%). In the most recent season (2014/15), however, Arsenal’s assisted shots moved nearer to the average rate over the two seasons, as they assisted on 40.12% of shots. Introducing a high-volume player in Alexis Sanchez who took a lot of unassisted shots certainly contributed to this. For Aaron Ramsey, this change in Arsenal’s source of shots (perhaps combined with his being deployed more on the right rather than centrally) meant being forced to create more shots unassisted, and his goalscoring dropped off in turn.
A case to watch this season will be that of Christian Benteke. For the past two years, Benteke’s assisted rates of 60% and 61.22% highlighted his role as a lone target for a team which tended to assist shots at a rate nearly half that in general. It is clear Benteke was successful in dispatching his mostly-assisted chances despite his team’s style of attack, but it’s interesting to consider whether this success will translate to his new club, Liverpool. Liverpool have actually assisted shots at rates not dissimilar from Aston Villa over the past two years (around 34% in both years for Liverpool, compared to 31% and 37% for Aston Villa). However, while Benteke thrived on assists in spite of his environment, Liverpool haven’t had a player take even 25 shots from open play in a season while being assisted at a rate higher than 50% (Jordan Henderson in 2013). Given this, as well as the departure of Raheem Sterling, provider of 24.38% of Liverpool’s open play shot assists, it’s possible that Christian Benteke will be forced to adapt from the steady diet he’s been used to, and if so, it’s uncertain how he’ll fare.
While I’ve examined some cases of players who rely on assists to generate their shots struggling to cope with creating their own opportunities, having a high assist rate doesn’t mean that a player isn’t a capable striker or a valuable player. Diego Costa is an example of a high assist rate player who moved with great success to a team which had previously generated few assists (though the simultaneous arrival of Cesc Fabregas didn’t hurt in this regard).
What a high or low assist rate ultimately means is that we shouldn’t count on a player’s performance, good or bad, translating to a different team with a different style, or even the same team with different personnel.