Does Liverpool Deserve to be Champions? Analyzing the Premier League Season Using Expected Goals

Charles Carey

*All Statistics via Statsbomb and Fbref.com

Another captivating Premier League campaign has come to an end. This season was truly unlike any other. It lasted 352 days, which was by far the longest in the history of the Premier League. And, if you didn’t hear, for the first-time ever, Liverpool was crowned champion of the Premier League.

To the naked eye, Liverpool’s road to becoming champions seemed like a cakewalk. Liverpool won their first eight fixtures. They were unbeaten in the league until late February. Before the lockdown, they were on pace to break the record for most points in a season and become the quickest team to clinch the Premier League title with seven games remaining.

However,  let’s look at the season from an entirely different perspective - let's look at it through the lens of expected goals. Expected goals (xG) is a metric that measures the likelihood that a player will score a shot. If, for each match, we use expected goals instead of actual goals, another season unfolds – one that doesn’t lead to Liverpool’s title. Statisticians use multiple variables to pin a value to each shot, including, the location of the shooter, type of assist, and shot type. Soccer statisticians regard xG the best metric for estimating how many goals a team should have scored in a match. For this study, each match xG were compared instead of actual goals.

*The chart on the left shows the actual Premier League table, while the one on the right shows the table when xG is used instead of goals.

As shown by the charts above, the season outcome was very different when using xG instead of the team's actual goals. When a team had more than .5 xG than the other side, the match was a win. If the difference in xG was less than .5, the match was a draw. If the difference in xG was .5, possession statistics were used to decide which team won or if it was a draw. The chart below demonstrates how xG and possession decide each match.

The most unexpected finding from this study was that Liverpool would not have won the league using xG. The goal of this post is not to discredit Liverpool for winning the Premier League. On paper, they were nearly the greatest team ever to grace the Premier League and broke multiple records en route to clinching the Premier League earlier than any team has done before. 

However, xG tells us that luck was on Liverpool’s side more times than not. For instance, in Matchweek thirty-seven, Liverpool beat Chelsea five to three. However, if we look through the lens of xG instead of actual goals it reveals that Chelsea should have won the match 2.3 to 1.2. xG is only an approximation of how the game should have scored. For this match the scoreline should have been 2-1 in Chelsea’s favor since matches cannot be won by a decimal score. Chelsea had more and better opportunities to score, but they were unable to take their chances successfully. Had this, and other matches, not gone in Liverpool’s favor, their season would have gone differently. 

Looking into individual players, Liverpool also benefited from plenty of heroic performances. Players such as Sadio Mané and Trent Alexander-Arnold provided more goals and assists than expected. Alexander-Arnold had an expected assists (xA) total of 9.6 but finished the season with 13. Also, Sadio Mané had an xG of 13.4 for the season but scored 18 times. The incredible performances of these players, along with plenty of others, are what led Liverpool to glory this season.

Liverpool was not the only team to outperform their expected finish. Teams such as Tottenham and Newcastle, performed much better than what their xG showed. Players such as Harry Kane and Son Heung-min outperformed their expected stats by a wide margin even though both missed significant game time. Newcastle benefited from exceptional goalkeeping from Martin Dúbravka, who had a save percentage of 70.5% and 134 total saves throughout the season.

While these teams had luck on their side this past season, most did not. The chart above shows that Manchester City had the opportunity to win their third straight Premier League title. Throughout the season, Manchester City dominated games that they eventually lost. For example, in Matchweek five Manchester City lost to Norwich City 3-2. However, Manchester City had 69% possession and had an xG of 2.2. This is greater than Norwich City’s 1.6. Matches like these were ultimately City’s downfall this season. If we look further in depth into this match, forward Sergio Agüero missed multiple chances close to the goal that should have put his team ahead. Performances like this and the consistent underperformance of Gabriel Jesus are mainly why Manchester City was unable to win the Premier League again.

Chelsea was yet another team that should have finished higher in the standings based on xG. However, Chelsea’s main issue this season was goalkeeping. Chelsea had by far the lowest save percentage in the Premier League. Goalkeeping is an issue that Chelsea must address this summer if they want to improve for next season. Southampton underperformed as a team throughout the season as well. Their lackluster season was mainly due to the lack of production from their attackers, other than Danny Ings, and conceding twenty shots during the season from individual defensive errors alone.

Perhaps the most disappointing finding from this study was that Watford nearly avoided relegation. In the xG table, Watford finished comfortably above the bottom three, and only two points behind Tottenham. Watford’s disappointing season was mainly due to costly individual defensive errors. Watford led the league in own goals scored and conceded nine penalties during the season. To go with this, Watford, as a team, had a G – xG (goals minus xG) of -10.7 throughout the season. Thus, Watford should have scored at least ten more goals during the season. It is always sad to see teams get relegated, but it is much worse when it turns out they should have prevented it.

Looking towards next season, xG projects Liverpool to regress. It is incredibly unlikely that Liverpool replicates how clinical they were in front of the goal this season. If they create a similar amount of chances as this season, their total goals should decrease. While regression is not guaranteed, it is rare that teams sustain consecutive years of significant overperformance. For example, Manchester City outperformed their xG by twenty-six goals in 2017-18 and 13 goals in 2018-19. This past season Manchester City only outperformed their xG by 8 goals. In the past two seasons, Liverpool outperformed their xG by 16 goals in 2018-19 and 15 goals this past season. The Premier League average for xG overperformance (G - xG) was +1.6 goals this past season. Over time, Liverpool’s xG overperformance should be near this number. For more information on which teams overperformed this season, or in the past, check out these datasets. 

Teams typically outperform their xG tally by attempting, and scoring, many difficult goals. Liverpool scored fourteen goals this past season from outside the penalty area, and scored many goals from barely inside the box. For a more in-depth analysis of Liverpool’s recent shooting charts, check out this article. Over the past few years, Liverpool attempted fewer outside the box shots. However, the goal-scoring rate of these shots remained above-average. Liverpool has many skilled shooters, but according to statistics, the long-range shots they scored in the past are unlikely to be scored again. The decrease in xG for next season is because most Liverpool players outperformed xG, meaning their finishing from this season was highly unsustainable. 

Whether it be an entirely different table or a twenty-point decrease from the champions, this article demonstrates how expected goals and real performances can tell two very different stories in sports. Some fans may feel cheated knowing their team should have done better this season, while others may feel fortunate that their team did so well. A goal like this can demonstrate how the randomness of sports is what makes following it so exciting.