Despite getting walloped by — squints, removes glasses, squints again … puts glasses back on — Watford over the weekend, Liverpool are going to win the Premier League. (Exiting the FA Cup at Chelsea on Tuesday will do little to stop this, either.) FiveThirtyEight’s projection model gives Jurgen Klopp & Co. a greater than 99 percent chance of winning the title, and it puts their average outcome at 101 points, which would be a league record.
Depending on the market, you could bet 10 grand on a Liverpool title right now, and were it to happen, you’d be rewarded with a whopping $10 in winnings. Insofar as the human brain or sophisticated computers are able to conceptualize potential outcomes, the only outcome is Liverpool lifting the Premier League trophy at some point over the next couple of months.
There are plenty of reasons that even though it’s only the beginning of March, even an awful result doesn’t register as a speed bump.
— Euro 2020 Way-Too-Early Preview: Assessing the field 100 days out
— Meet Hout Bay United, Jurgen Klopp’s “other” club
— Ogden: Liverpool’s season in danger of being good, not great
Liverpool are the only big club in Europe to truly invest in a top-down analytical approach. They moved quickly for one of the best managers (Klopp) in the sport. They’ve acquired perhaps the best attacking trio (Roberto Firmino, Mohamed Salah, Sadio Mane) in Premier League history, and all for cut-rate deals. They’ve understood the value of set pieces. They’ve spent big on a top keeper (Alisson) and a top central defender (Virgil van Dijk), and both that keeper and that defender now have a claim to “best in the world” titles at their respective positions. They’ve also unearthed a dominant pair of fullbacks (Trent Alexander-Arnold, Andy Robertson) who have been able to carry a massive creative burden.
What isn’t clear, however, is how the midfield contributes to all of this. They clearly do. You don’t gobble up as many points as this team has over the past two seasons if 30% of your outfield lineup isn’t contributing in some way. But while the rest of the roster is unarguably dotted with world-class players, the picture in the midfield is a little murkier.
Naby Keita was supposed to be the midfield version of the Liverpool approach — analytically adjacent, high-action, progressive play — and he has been phenomenal for short stretches, but he hasn’t been able to stay healthy. Before the season, maybe you could’ve argued that Fabinho was the team’s world-class workhorse in the middle, but Liverpool have actually been better without him this season: their goal difference and their expected goal difference per 90 minutes are both higher over the stretches when the Brazilian hasn’t featured this campaign. Other than Keita’s low-minute, high-production, per-90 efficiency, none of Liverpool’s other midfielders create many chances, complete many passes into the penalty area or score many goals, according to the site FBRef.
Other, more holistic player valuations agree. Per the player-contribution ratings from consultancy 21st Club, Liverpool have five of the 15 best players in the world: Van Dijk (no. 5), Robertson (9), Alisson(10), Salah (11) and Mane (12). No midfielders! Compared to players at the same position, only one of LFC’s midfielders — Giorginio Wijnaldum, tied for fourth — ranks in the global top 10.
Liverpool’s success presents an intriguing question about the present and the future of the sport: Might the “metronome” sharp-passing midfielder represent something like soccer’s version of the midrange jump-shot artist or the high-usage running back, aka a beautiful, breathtaking player archetype with a deep tradition that doesn’t actually affect winning all that much?
The transfer market certainly doesn’t value midfielders the same as everyone else. Of the 20 most expensive players of all time, the only true midfielder is Paul Pogba. The data don’t value them, either — at least, not yet.
The first tiny step in soccer’s movement toward a more objective understanding of the game was expected goals, a statistic that both put a number on the process most managers care most about — did we create better chances than our opponent? — and created some new knowledge. One bit of info was that finishing skill provides only a marginal amount of value: most professional soccer players convert their chances at roughly the same rates. More important is the ability to get on the end of the chances in the first place.
On top of that, xG also provided a couple of seemingly obvious insights: shots with your feet are more likely to go in than shots with your head, and shots closer to the goal are better than shots from farther away. Teams might not be consistently acting on this information, but it has had an effect, as the average shot distance in the Premier League is declining with each passing season.
Liverpool have pretty clearly worked off of this information. Salah, Mane and Firmino all produced impressive expected goal and expected assist numbers prior to arriving at Anfield. If everyone were doing the same thing, there’s no way that the trio could’ve been had for less than a combined €120 million. By looking at these more stable indicators, Liverpool were able to get a better assessment of player quality than their competitors.
The next step after expected goals is to move beyond shots toward something that goes by a number of different names but is typically referred to as “Expected Possession Value” (EPV) or “goal-probability added.” As I’ve written about, it’s something that Liverpool’s Director of Research, Ian Graham, has spoken about openly. “We try to take whatever action a player does on a pitch — whether it’s a pass or a shot or a tackle if you’re a defender — and ask the question, ‘What was this team’s chance of scoring a goal before this action happened?’ And then, ‘What was this team’s chance of scoring a goal after that action happened?'”
Karun Singh, a software engineer in San Francisco, has created his own easy-to-grasp version of the idea called “Expected Threat.” “Given the ball at a certain location on the pitch, xT tells us the chances of a team going on to score in that possession,” he said. Essentially, he plots how likely a team is to score from anywhere on the field, with players receiving credit for moving the team to a higher value zone with a pass or a carry. Last season among first-choice starters, Alexander-Arnold and Robertson led the way for Liverpool in xT, followed by the front three and then … defender Joel Matip. Once again, none from the midfield trio.
The FC crew believe Liverpool’s focus was on the Champions League after their second defeat in as many games.
The intuitive logic of xT or the other EPV systems suggests that things that happen near the opposition goal, no matter what they are, have a larger effect on a goal being scored than a thing that happens farther away. And so, one of the common results from this kind of study is that most midfielders don’t produce a lot of value.
The midfielders who do are rare player types such as Kevin De Bruyne, who manages to rack up tons of assists and passes into the penalty area while somehow still functioning as a de facto midfielder. However, most midfielders complete a ton of actions, but none of them is likely to lead to a goal, meaning the difference between the best traditional midfielder and a league-average one ends up being quite small.
In a talk he gave at the Statsbomb Conference late last year, the company’s CTO, Thom Lawrence, neatly referred to the problem as “The Valley of Meh.”
“From the team point of view, you could look at the ‘Valley of Meh’ and conclude that most play in the middle of the field is actually within the margin of error, and so it doesn’t matter what you do, but we know that’s not really true. Every attack comes from somewhere, and it’s a product of volume and quality,” he said. “In terms of transfer fees, let’s say you can pay to get a player who’s 10 percent better than their replacement. Do you want that multiplier on lots of low-value actions in the middle of the pitch or on just a few high-value actions at the ends of the pitch? Both answers are probably valid, but it’s clear the market has always taken one side of that debate.”
Translating this back to Liverpool, though Keita is absolutely one of the few midfielders who bursts through these theoretical constraints — he leads the team in passes into the penalty area and through-balls per 90 minutes and is second behind TAA in expected assists per 90 — Liverpool’s first-choice trio of Wijnaldum, Fabinho and Jordan Henderson does seem to be a concession to this idea.
Rather than trying to create anything from a position on the field where, by definition, it’s hard to do that, Liverpool have assembled a collection of players who a) don’t provide many negative actions and b) allow their teammates to get up the field and create the threat. Of the players listed as midfielders on FBRef, Wijnaldum, Henderson and Fabinho all rank in the top 20 of pass-completion percentage.
“There’s nothing spectacular about the Liverpool midfield from a creative standpoint, but they also don’t recklessly give possession away,” said Mark Taylor, an analyst who has worked with Premier League teams. “That frees up the two fullbacks to create chances via carries or passes but also gives them the freedom to sometimes lose the ball in advanced positions.”
While the models are a step forward in connecting the data with reality, the EPV system still accounts for only what’s happening with the ball. The average player has possession of the ball for only a minute or two per game.
“I think a large part of midfielders’ value is derived from off-the-ball actions — positions that they take up to create space, opposition players that they mark or critical passing lanes that they block, covering for their defenders, etc.” Singh said. “For instance, I think a big part of why TAA and Robertson are able to create so much xT is because they can bombard forward knowing that they’re being covered by the midfield trio.”
But even if Liverpool’s midfield are showing the limitations in how value is quantified, they might also be showing other teams the most valuable way to deploy your midfield. Rather than asking them to try to control the game — a difficult achievement, given how far away these players are from goal — instead ask them to do all things that allow everyone else to control the game.
“If xG has slowly taught teams to take fewer long shots and try and work the ball further toward goal,” Lawrence said, “perhaps Liverpool are the archetype for what goal probability added teaches us: progress the ball out wide, make lots of cross-field balls to stretch the opponent, use your CMs to destroy any potentially valuable attacks through the middle rather than create and have possibly the most talented attack in the league.”