The performance and potential of soccer teams and individual players are increasingly being defined in probabilistic terms rather than simply by counting the actual fruits of their labours, namely goals scored, allowed or created.
Are soccer teams reliant on individuals?
The point can be illustrated by the perception of two of Manchester United’s highest profile players from 2016/17, Zlatan Ibrahimovic and Paul Pogba, with the weight of expectation apparently lying heavily on both at particular stages of last season.
Expected assists and goals for both players and teams and their more persuasive predictive power for future performance allows bettors to make a more informed judgements.
The former experienced what by his very high standards might be considered a goal drought during the mid-season period (he went six games without scoring a goal between September and November). Additionally, Pogba’s return of five Premier League goals and four assists could also be considered poor reward from such a high-value purchase.
The importance of process, the ability to create or get onto the end of chances, even if few were actually being converted was implicitly recognised by Ibrahimovic in a TV interview during his dry spell.
In a short timeframe, it is likely that the quality and quantity of chances that a player is involved in will not be accurately reflected in his more visible, tradition statistics of goals scored or created.
He might be on a hot run and goals and assists will flow or, as Ibrahimovic found, he may not be returning the number of goals that would be typically expected for the type of chances he was involved in. It is likely that a player in this situation will in the future perform more in line with the quality and quantity of his chances, rather than mere goal counts; which are partly down to randomness.
Using expected output to analyse player performance
In short, we would expect a player’s actual record to regress to one more in line with his expected output. This was the case with Ibrahimovic and by the end of a season cut short by a serious knee injury, United’s target man boasted actual goal figures that were very close to his expected figures based on a probabilistic model of the likelihood of his attempts turning into goals.
Paul Pogba underperformed last season compared to his underlying stats.
Expected goals models (amongst other outputs) are now commonplace in soccer. Some use different variables and data but they are effectively trying to do the same thing - use a model to predict an outcome.
In Paul Pogba’s case, we should not only take into consideration his expected goal contribution, but distribute the credit that goes to a player who creates the chance and the player who attempts to convert the opportunity.
Around a quarter of goals scored in the Premier League are solo goals; meaning they are scored with no assistance from a teammate. Similarly, the larger proportion of goals where an assist is involved demands that the creator receives some credit for that assist.
However, just as Ibrahimovic recognised his process was sound and goals would eventually return at a rate more befitting of the chances he was involved in, a central midfielder, such as Pogba is also dependent upon the partly random nature at which the chances he creates are finished by others.
Pogba's actual expected involvement in those 56 probabilistic Manchester United goals accounts for over 12% of the total. Zlatan’s involvement amounted to over 18%.
To view Pogba’s first season on his return to Manchester, we can look at his cumulative expected goals and assists as well as his raw actual return. He scored five actual goals, three being solo efforts and provided assists for four others that were scored by a teammate.
His expected record, based on a model such as one previously explained in Betting Resources suggests five expected assisted goals scored, two solo efforts and five further assists for his colleagues.
So it looks like Pogba, just as Zlatan did earlier in the season, underperformed compared to his underlying stats and a less extreme return in the future may bring about some improvement in his actual goals scored.
How team performance relates to individual performance
To get a more rounded view of Pogba’s big money move we should also look at his expected stats in the context of those of his team, Manchester United. United were out-scored in actual goals by each of the teams who finished above them, as well as Everton and Bournemouth. These stats point to a more defensive set up, in which Pogba played a part.
In the context of playing for United last season, the attacking process of Paul Pogba was above average for a player who played 2610 minutes.
Their expected goals scored were in the region of 56 and if we make adjustments to Pogba’s expected numbers to prevent double counting of assists, his actual expected involvement in those 56 probabilistic Manchester United goals accounts for over 12% of the total. Zlatan’s involvement amounted to over 18%.
Even for two players who respectively played over 70% of United’s available Premier league minutes, their importance to the attacking process for Mourinho’s side is evident - in the context of playing for United during the 2016/17 season, the attacking process of Paul Pogba was above average for a player who played 2610 minutes.
Premier League clubs and reliance on individuals
While United were not quite a one man team with regards to attacking intent, the loss of Ibrahimovic and the return to Everton for Rooney (another player who perhaps with more playing time would have approached Pogba’s 12% contribution) does reiterate the importance of their recent signing, Romelu Lukaku.
The knock on effect for Everton is that 25% of their goal involvement needs to be replaced with the departure of Lukaku, a figure that will rise to 40% if Ross Barkley moves to another club. Lukaku's percentage contribution to his team's attaking process is one of the many reasons why he is one of the favourites in the 2017/18 Golden Boot Winner betting.
Other teams who were reliant upon a single player in 2016/17 and now about whom transfer speculation abounds include Arsenal (Sanchez 25%), Swansea (Sigurdsson, 23%) and Stoke (Arnautovic, 19%).
25% of Everton's goal involvement needs to be replaced with the departure of Lukaku, a figure that will rise to 40% if Ross Barkley moves to another club.
Liverpool have half a dozen players who have demonstrated above average expected attacking performances in 2016/17 and appear the best among the top sides to withstand injury or suspensions, as do Spurs (Kane, Alli and Eriksen), Southampton and to a lesser degree, Chelsea and Manchester City.
The increasing availability of expected assists and goals numbers for both players and teams and their more persuasive predictive power for future performance allows bettors to make a more informed judgement about the impact of transfer moves or team news, particularly if placed in the context of the preferred attacking profile of their team.