*All data correct as of 22/06/2020, after each team completed 30 Premier League Fixtures.
If you follow football statistics of any kind you will be aware of the concept of Expected Goals (xG). This statistic first got its break in the gambling world where companies would use a team's xG value to calculate the quality of their performance, finding it a much greater indicator of form than other more conventional statistics and often better than observing previous results. In its most simplest form, xG puts a value to each shot taken within a match that indicates the chance that the attempt will result in a goal. A number of variables are considered to give a shot its xG value such as position, assist type and the type of shot (volley, header, etc…) to name a few. Historical data collected from thousands of matches and shots is used to create an understanding of how likely it is that similar shots will end with a goal. As a basic example, if an analyst were to study 1000 shots from the penalty spot which came from a cross and found that 10 of these shots resulted in a goal then an xG value of 0.1(xG) could be assigned to this type of chance. Expected Goals can provide useful information on a team's performance in a game by calculating an xG value for both sides in a match. The xG value is an accumulation of the xG value of all their shots taken within the match. For example if a team takes 3 shots with xG values of 0.2, 0.4 and 0.5 then the teams xG total for that game would be 1.1(xG). A book by James Tippett, ‘The Expected Goals Philosophy’ provides an in depth look at all aspects of the metric Expected Goals and is highly recommended if you want to develop a greater knowledge around the statistic and both its history and future within football.
Further to the above mentioned uses of Expected Goals, it can be used to provide detail of how clinical a team has been in front of goal. After 30 Premier League games, Champions Liverpool had scored a total of 66 goals this season, however, had accumulated an xG value of 58.1(xG) showing they have been outperforming what would have been expected of them in terms of goal scoring. In contrast, Norwich have underperformed their expected total of 37.1(xG) only scoring 25 goals in their first 30 games. The graph below displays both the total number of goals in the first 30 games as well as the percentage difference (xG%) between goals scored and xG of all 20 Premier League teams.
To begin with, the outliers in this graph do not provide any immediate shocks. Both Manchester City and Liverpool are out in front with regards to the total number of goals scored and have a respectable xG% to accompany their goal tally. Leicester City’s campaign so far has no doubt seen them exceed expectations and the same can be said for their efficiency in front of goal, as they are able to boast the highest xG% in the league. They have netted 59 times this season, outperforming their Expected Goals total of 48.8(xG) by 17.3%. North London rivals Arsenal and Tottenham are the only other teams in the league who have recorded a greater number of goals than their xG total, overachieving based on this metric by 6.5% and 12.8% respectively. Norwich unsurprisingly find themselves isolated in the furthest bottom left of this graph. As is to be expected when recording the lowest number of goals this season they have underperformed their xG total of 37.1(xG) by some margin.
The biggest surprise displayed within this graph however, comes within the lower left region. One of the great success stories this season has been newly promoted Sheffield United who are in the process of competing with a number of well-established Premier League sides for a spot in next year's Champions League. Despite this, their ability to turn chances into goals is amongst the worst in the league, scoring 36.8% less than they would have been expected to. Whilst this hasn’t proved detrimental with regard to league standings, it is unlikely that such poor efficiency in front of goal will be sustainable as a long term characteristic of Chris Wilder’s side.
This graph displays the goals conceded by each Premier League team for their first 30 games during the 2019/20 season. Whilst the first graph didn’t particularly offer an explanation for Liverpool’s extensive lead at the top of the League, it becomes clear to see why they find themselves winning their first title in 30 years, when analysing the amount of goals they’ve conceded. Liverpool had conceded just 21 goals at this stage of the season, however this figure is accompanied by an xGA% of -32.8%. Such a large differentiation shows Liverpool have perhaps been fortunate to concede so few goals this season with their xG against (xGA) tally sitting at 27.8(xGA). A combination of both elite shot-stopping from Brazilian goalkeeper Alisson Becker, resolute defenders and a lack of prolificacy from opposing attackers has allowed this differentiation to form and has perhaps provided Liverpool with the slice of fortune needed to finish top of the pile.
An explanation for the minimal impact of Sheffield United’s lack of efficiency in front of goal is offered within this second graph. After 30 games, they had conceded the 2nd fewest goals in the league despite an xGA of 37.9(xGA), so although they underachieved with regard to scoring goals, a parallel with their oppositions offensive efforts has allowed the Blades to enjoy a season far better than many would have imagined possible. Another team whose defensive efforts have seen results greater than expected are Crystal Palace with an xGA% of -34.6% who make up a cluster of 4 teams to have achieved an impressive defensive record whilst significantly overachieving in relation to the goals they would have expected to concede. At the other end of the graph sit the teams who perhaps haven’t been quite as fortunate. Southampton and Everton appear to be cruising towards a mid-table finish moving into the final games of the season, but the graph above suggests that they may have been unlucky with the amount of goals they have conceded. Southampton saw the biggest differentiation in goals against v xGA, conceding 52 goals when they would have been expected to concede 10 goals fewer (18xG%).
Whilst the metrics xG and xGA can provide insight into where teams are performing in comparison to an average, there is a lack of context to explain why a team may be under/over performing from the expected results. Further analysis into a team's attacking and defensive efforts in front of goal will be able to determine more beneficial conclusions as to the causes of these differentiations. Whilst these specific details remain unclear from the above graphs, they may somewhat offer an idea of the stability of a team. For example, Leicester City this season, had scored 17% more and conceded 32% fewer goals than their expected tallies would have predicted. Whilst this leads to an excellent run of form as these differences are maintained, over time it would be expected that the differentiations will reduce which would likely see a drop in results. As it happens, Leicester City failed to win their first 4 games upon the resumption of the season with 2 draws and 2 defeats, both of which came by 1 goal. So, as metrics such as Expected Goals are now growing ever more popular within the footballing world and as their implementation expands, it offers the exciting chance for our understanding of the game to follow suit.
*All data is collated from Wyscout, after each team completed 30 Premier League fixtures.
About the Author: This article was written by Axl Rice, Head of Academy Analysis at Nottingham Forest Football Club. Check out other articles that Axl has written by clicking here, including an analysis of Stade Rennais' surprising Ligue 1 success this season.