Who was Liverpool’s best player during their ‘19-‘20 Premier League season?
Based on an analysis of ESPN’s player ratings through the season, I have an answer: Mane. Van Dijk. Salah. In that order.
The wait for a trophy is over.
Liverpool just marched through one of the most impressive seasons in Premier League history. Over the course of the season they amassed a staggering 99 points, one shy of the record 100 point mark set by the 2017–18 Manchester City team. They wrapped up the title with a record 7 games remaining, giving Liverpool fans plenty of time to celebrate their first league title since 1990 and ending a 30 year wait. In addition to winning the Premier League, Liverpool added the UEFA Super Cup and FIFA Club World Cup to the trophy case.
With this season over and the next one starting in less than two months on September 12 due to the scheduling changes around COVID-19, I thought it would be fun to do a quick analysis of the Liverpool team of 2019–20. This incredible team, built carefully over the past several years by the masterful and charismatic Jurgen Klopp has the potential to add even more trophies in the future. But, in a team full of stars, who shone the brightest?
Liverpool’s team is scary deep.
Liverpool’s talent across the pitch runs deep in nearly every position. In the ESPN FC 100, which ranks the top 10 players at each position across the globe, Liverpool had a staggering 8 players represented, including 4 ranked first in their respective positions. To add to this wealth of riches, Liverpool’s coach Jurgen Klopp was named best in the game. Below are the Liverpool stars and their ranking from that list:
- #1 Goalkeeper — Alisson
- #1 Right-Back — Trent Alexander-Arnold
- #1 Center-back — Virgil Van Dijk
- #1 Left-back — Andrew Robertson
- #1 Coach — Jurgen Klopp
- #2 Wing — Sadio Mane
- #4 Forward — Mohamed Salah
- #5 Forward — Roberto Firmino
- #6 Central midfield — Fabinho
While individuals can certainly be difference makers, football is a team game at its core. Some of the most important contributors on the pitch go about their craft without ending up on the score sheet or in the highlight reel. Offensive players pester defenses with their runs to create space for teammates. Defenders battle things out with the opposition to keep them from finding their rhythm. While it’s easy to track goals, assists, and other scoring based metrics, these data points miss these subtle contributions and narratives that occur during the game. Many great and vital players make their presence felt, but that rarely show up in the box score.
Newer stats centered on passes completed and accuracy have been aimed to capture the contributions that non-scoring players have on the game. But even these measurements are imperfect. A 15-yard pass completed in space is very different than a 15-yard ball threaded through the opposition’s defensive line to the striding foot of an attacking player. Superior style or off-the-ball play are simply difficult to quantify with the measurements available.
So, we’re left with a dearth of data for understanding who is most important to Liverpool across the varying skills sets of its players. How do you judge the relative importance of Alisson, a world class goal-keeper against the scoring prowess of Mohamed Salah? Can one compare the offensive creativity of Roberto Firmino with the imperious defending of Virgil Van Dijk? Is the seemingly endless line-to-line hustle of Andy Roberston as valuable as the midfield distribution from skipper Jordan Henderon? It’s a difficult problem.
The data set
After each Liverpool match, an ESPN writer completes a player rating report based on observed performance. While 100% subjective in nature and based wholly on the opinions of the writer, these summaries are convenient from a data perspective in that they regularly rate players on a 1–10 scale. They attempt to capture quality in a quantifiable way. These ratings, while certainly imperfect (is Roberto Firmino being rated against everyone else playing or how the writer expects Firmino to play?), capture the narrative of the game in a way that shots on target, goals, or other soccer metrics can miss. A total of 9 different writers published a player rating match summary (1).
ESPN produced 37 player rating reports for Liverpool during the 2019–20 Premier League campaign, one for each game played except for the July 5, 2020 match vs. Aston Villa (I’m not sure why). Starting players are given a 1–10 rating assuming they are on the pitch long enough to merit a rating. For example, a starter might be injured in the first minute of the game, wherein the writer would likely rate the player as N/R (meaning “Not Rated”). Substitutes are also given a 1–10 assuming they are on the pitch long enough to have a quantifiable impact. For substitutes coming on late in the game or not impacting play meaningfully during their time on the pitch, they receive a rating of N/R.
Even though the writer for each player rating post might change, there is at least a basis that we could be standardized against — how that player performed relative to their teammates on the pitch. This relative scale exists across writer, pitch conditions, whether the game was home or away, or game outcome.
A simple quantifiable approach.
While games might go well or poorly for Liverpool, player ratings can reflect a relative performance of its players against the overall team’s average rating (generated by averaging all the ratings bestowed on players in that game). For example, in Liverpool’s first game of the season on August 9, 2019 vs. Norwich City at Anfield, 12 players received a rating. This included all 11 starters (Alisson — 7, Trent Alexander-Arnold — 7, Joe Gomez — 6, Virgil Van Dijk — 7, Andrew Robertson — 6, Fabinho — 7, Jordan Henderson — 7, Gini Wijnaldum — 5, Divock Origi — 8, Roberto Firmino — 9, Mohamed Salah — 8) as well as their substitute keeper (Adrián — 6) who came on after 39 minutes when Alisson was injured. Two other substitutes were not rated — Sadio Mane who replaced Origi after 74 minutes and James Milner who replaced Firmino after 86 minutes — given the limited action they saw.
The average player rating for the Norwich game was 6.9, based on the 83 total points given across the 12 players receiving a rating. We can assume that Firmino — who received a 9 — had an above-team average performance relative to his teammates. Wijnaldum, with a rating of 5, had a below-team average. The written notes for these scores reflect these assessments.
Data
Over the course of the season we can track how a player performed relative to the average rated Liverpool player. For example, below are the ratings for Salah and Van Dijk relative to the average each game. Note that Salah was rated in 32 games this campaign, whereas Van Dijk was rated in all 37 available (apart from the missing Aston Villa game).
Another way to understand this data is to look at the difference between the player and the average Liverpool player rating. Rather than only care whether a player receives high or low scores consistently, this performance can be compared to the relative performance of the team. For example, in the January 19, 2020 vs. Manchester United, Van Dijk scored a goal and put in a stellar defensive performance. He received a rating of 9 compared to the average team player rating of 7.3, representing an “above average” performance of 1.7 rating points. When looking at the data in this light, we can more clearly see the performances of Salah and Van Dijk in the context of the team.
Players with the highest positive impact
This analysis can be repeated for every Liverpool player throughout 2019–20 Premier League thus far. If we average the +/- rating points for each player, we can observe a ranked per-game performance indicator. Based on this analysis, Sadio Mane is Liverpool’s most impactful player, adding +0.67 rating points above the team average per game. That’s a significant lead over his next three regularly featured closest teammates: Salah (+0.45 per game), Alexander-Arnold (+0.20 per game), and Van Dijk (+0.18 per game). Curtis Jones also had a strong game above average score from the younger players getting minutes this season, but his metrics come from only rating performances.
If we sum up these +/- games over the course of the season, we can also observe the season-long impact per player. When we do this, the players who play more often and perform well regularly receive an additional bump. Sadio Mane’s impact is even more clear, as he racks up +22.2 cumulative rating points above the average Liverpool performer. Said another way, he has contributed an average of +0.67 points above the team average for 33 games this season. That’s impressive.
Salah (+14.0 over 32 ratings), Alexander-Arnold (+7.1 over 36 rating), Van Dijk (+6.8 over 37 ratings), and Firmino (+5.8 over 37 ratings). The only other players to be rated in over 30 games were Robertson (-0.2 over 35 ratings) and Wijnaldum (-10.8 over 35 ratings). In general Liverpool’s stars at forward and defense outperformed players in midfield based on this analysis, which matches with many commentaries. It was interesting to note that Alison — Liverpool’s star goalkeeper — didn’t do well across this analysis. Based on my readings of the commentaries, I think this is due to the fact that it’s hard for him to standout relative to the rest of the squad in a season were Liverpool was an offensive tour-de-force.
Zooming in on Mane’s game-by-game performance highlights just how vital he has been to Liverpool’s rampant 2019–20 season. Mane underperformed against the average Liverpool player in only 8 of his 33 games. In every other game, he outperformed the average Liverpool player rating.
If we look at games in this light — how often players under or over-performed relative to the average player, we can get an expectation of whether a play will bring an above-average performance. Once again, Mane is tops among regularly featured players. In a whopping 75.8% of rated games (25 of 33), he is rated above the average Liverpool player. That’s tops by over ten percentage points amongst a stacked squad. Other players performing above average in over 50% of games with at least 10 games rated include Salah (59.4%), Origi (54.5%), Van Dijl (54.1%), Firmino (54.1%), Keita (53.8), and Alexander-Arnold (50.0%).
Variability of Performance
Players who are regularly involved in scoring (Origi, Salah) tend to have higher variances relative defense or midfield players when it comes to their scores against the average. This likely reflects the fact that scoring a goal or directing play leading to a goal is the difference between what’s considered an average and a stellar game. Offensive playmerks are assessed more heavily on their ability to create goals.
Origi presents an interesting case for the club. While he did perform above the team average 54.3% of the time, his per game contribution is -0.11 points per game. How do we reconcile these data points? One way is to also look at the average points per game above or below the team average, as well as the variance of their per game performances. From this analysis, we can see that Origi’s performances tend to be fairly erratic across only 11 available data points, driven largely by one poorly rated performance relative to his teammates.
Salah — Liverpool’s goal-scoring machine and Premier League Player of the Season for 2017–18 — presents another interesting case. He performs above average 59.4% of the time, but his variance is second highest on the team. What can we read from this? At a simplistic level, Salah has more binary “on” and “off” games relative to other players. I think many followers of Liverpool’s campaign would agree with this assessment.
Fabinho, Henderson, Firmino, Van Dijk, and Robertson meanwhile have lower observed variances, particularly for players who’ve been rated in over 25 fixtures. How might we interpret this? One way is that they’re pretty consistent in what you’re going to get. For Liverpool fans, this sums up the stability Van Dijk has brought to Liverpool’s back line and Henderson’s consistent play — despite not having the highest above average contribution.
Mane interestingly has the sixth-highest variance amongst his data points. However, given how high his contributions are per game, this shows that even his “off” games are pretty darn good. Even when he’s not on the score sheet, he’s still harassing defenses and created value for his team.
Variance itself isn’t necessarily bad, particularly if a player has outlying great games. The ESPN writers don’t often give ratings of 9 or 10 to players, which they reserve for exceptional performances. Over the course of the season, only 8 Liverpool players received such a score. Among that group, Salah stood out with 6 such games, twice as many as Mane who had 3. This perhaps helps explain some of the variance seen above.
Who’s the barometer?
Finally, I thought it would be fun to identify if any Liverpool player acts as a barometer for the team’s average performance. Is there a correlation between individual ratings and the average team rating? While the data points are limited by the number of games and the rated appearances of players, correlation can (loosely) be measured for players with at least 10 data points. Intyerestly, Alexander-Arnold emerged as this player. His correlation was stronger than even that of manager Klopp (who also receives a score each week). If you want to get an idea how Liverpool is playing any given game, one may want to look at how Alexander-Arnold is performing.
Results.
So who is the most indispensable Liverpool player? Here’s my read based on this somewhat fun — and albeit questionable — analysis:
- Sadio Mane: Regular overperformer with the highest average rating points above average (+0.67 per game), nearly 50% higher than the next closest regularly featured teammate (Salah at +0.45). Over the season, his +22.2 cumulative performance points over 33 games is 58% higher than second place Salah (+14.1). He overperforms the average Liverpool player in over three-quarters of his games (75.8%), once again highest on the team by a significant margin.
- Virgil Van Dijk: Consistent overperformer with the fourth highest points above average per game (+0.18) among regular players. He underperforms relative to his teammates only 37.8% of games. His consistency is a huge asset. Along with Firmino, he was the only Red to have ratings across all 37 games. His variance is below average signifying he’s likely going to put in a solid performance. A true steadying force in the back — calm as you like. While not captured in the data, his defensive presence has probably allowed the younger Alexander-Arnold to shine higher up the pitch. And he can score those towering headers.
- Mo Salah: Second highest to Mane in both per game contribution (+0.45 points) and season-long contributions (+14.1% over 32 games). Performs above the team average 59.4% of the time. With his high variance in performance versus the team average, he occasionally brings the sublime game, captured by the 6 games where he received a score of 9 or 10 from ESPN writers. When he brings something special running down the wing, you can bet Liverpool is having a good day.
While the methodology on this exercise is far from perfect, this yields a result that at least feels reasonable. These are three of Liverpool’s biggest playmakers, regularly regarded as indispensable to the team.
Hope you enjoyed the analysis. I can’t wait to see the team back on the picth in two months. Until then, time to just enjoy being champions.
ESPN Liverpool Player Ratings:
10/27/19 vs. Tottenham Hotspur
11/30/19 vs. Brighton & Hove Albion
7/5/20 vs. Aston Villa (no player rating summary)
7/8/20 @ Brighton & Hove Albion
Endnotes
(1) ESPN writers included James Capps, Gustav Elvin, Tom Fenton, Nick Judd, Harry Kettle, Danny Lewis, Melissa Reddy, Shaun Reynolds, and Liam Wheeler.