How bad analytics built a Tottenham team that is heading for relegation

There's a funny, instructive little story in Michael Lewis' "Moneyball" that no one remembers because it doesn't involve Billy Beane and therefore was never recreated on a movie screen by Brad Pitt.

In the late 1970s or early 80s, the Houston Astros commissioned a study about what might happen to their team's performance if they moved the outfield fences closer to home plate. They wanted to move the fences in because they figured it would lead to more home runs, and because fans love home runs, they figured they'd sell more tickets. Except, given the types of hitters and pitchers on Houston's roster, the study's authors found, moving the fences in would actually lead to more losses for the Astros.

So, Houston's decision-makers looked at the data, and they decided ... to order that the study never be made public. They'd already made the decision to move the fences in and only wanted data that would support their choice.

I was told a similar story about a professional soccer club by someone who has been working in the industry for more than a decade. The team commissioned him to work up scouting reports for three different players. He broke down each player in detail, and his conclusion for each one was the same: You do not want to sign any of these players. The club responded by asking him if he was able to send over positive scouting reports for each player; they'd already committed to signing all of them.

In both stories, the organizations wanted to use data, but not to make better decisions. They wanted it to justify the decisions they had already made.

• Why missing the Champions League might boost PL teams
• Premier League without set-piece goals: What would the table look like?
• 2026 World Cup squads ranked: All 48 national teams

Now these might sound like stories from a simpler time. Almost every baseball team is being run with way more advanced analytical models than the public can access. And soccer data is everywhere now; Amazon is powering Bundesliga broadcasts and "expected goals" have become part of the common language for virtually every English-language broadcaster.

Yet, while baseball teams have mostly moved beyond using numbers to reiterate and justify their own ingrained biases, soccer clubs have not. They're still not even close. Don't believe me? All you have to do is take a look at the team that, reportedly, was considering telling its own fans that it had "redefined what a modern football club can be."

In other words, all you have to do is look at Tottenham Hotspur.

What we know about how soccer works

Perhaps the core insight of soccer's analytics movement is something that everyone already knows: The best team doesn't always win.

This is essentially what expected goals tells us. At almost any point in a given season, a team's expected-goal differential is a better predictor of future performance than any other top-level number like shots, goals or points. If the best team always won, then past wins would immediately tell us who the best teams are, and then those past wins would predict the future.

Instead, it would appear that the best teams are the teams that accrue the greatest proportion of expected goals in their matches. If we simplify that idea down beyond the abstraction of an ever-updating algorithm that awards every attempt in a given match a specific conversion probability, then the best teams are simply the teams that create better chances than their opponents.

This is something that anyone who has played or watched the sport for long enough really does understand on a deep level -- whether or not they're willing to admit it. But by acknowledging this, we're accepting that there's a large amount of randomness inherent to the outcome of a given soccer match, because there's a large amount of randomness inherent to kicking a bouncing ball, with a misshapen foot, past the one guy on the field who is allowed to use his hands.

Now, the Premier League season isn't that long, and each season consists of something like 20 different team-level experiments. So over a decade, we get 200 different little experiments. And over these 200 different seasons, we'd expect there to be a couple of examples where the randomness boosts, or punishes, a team for an entire season.

That is exactly what we see. Here's every Premier League season since 2010, arranged by how much a team underperformed or overperformed its xG differential:

That team all the way to the right is Tottenham in 2016-17. And if you had to pick anyone to occupy the far-left spot, Tottenham in 2025-26 would seem like a pretty good choice, right? For one of the 10 richest teams in the world to be in a relegation battle with six games left to go in the season, surely "historically bad luck" would have to play a role?

Nope. That's Sheffield United in 2023-24.

This season, Tottenham aren't an outlier at all. Their goal differential (plus-11) is actually slightly better than their xG differential (plus-15.13), but not that much.

How, then, does a team with what's estimated to be the ninth-most valuable roster in the world actually become one of the worst teams in the Premier League? One possibility: You measure the things that you think matter -- and not the things that actually do matter.

Tottenham's major issue: They can't pass

Usually, soccer is a complex, dynamic game where individual qualities are impossible to extract from the interdependencies of roster construction, managerial instructions and on-field interactions. But sometimes you get a team like Tottenham, where the diagnosis is pretty simple: These guys can't pass.

At Gradient Sports, there is a team of people who watch every Premier League game and grade every pass a player makes on a minus-2 to plus-2 scale. Here's how they describe the process:

For example, consider a centre-back passing the ball on the halfway line. A routine, unpressured pass to an open teammate would receive a 0, as this meets the expectation of our expert Grading team. A precise, line-breaking pass under pressure would receive a positive grade. Conversely, an underhit pass to a teammate -- even if completed -- would receive a negative grade if it falls below the expected standard. This reflects our focus on evaluating performance rather than just outcomes.

The grading process is guided by detailed frameworks designed to minimise subjectivity and ensure consistency. Once raw grades are collected, they undergo multiple layers of quality control, including senior review of flagged actions, consistency checks, ongoing analysis, and dedicated quality assurance processes.

Based on this process of evaluating passing, here's where Tottenham's five best passers rank in the Premier League season:

1. Cristian Romero: 19th
2. Mickey van de Ven: 87th
3. Destiny Udogie: 152nd
4. Kevin Danso: 167th
5. Mohamed Kudus: 186th

Passing is the fundamental skill in this sport. The average Premier League team attempts 450 passes per game. Nothing else comes close: in a single game, the average team attempts eight shots, crosses the ball 18 times, tries to dribble past defenders 18 times, attempts 16 tackles, and makes eight interceptions. If you can't pass the ball, then nothing else matters. It's the force at the heart of the game that gives everything else meaning.

So, how the heck does one of the richest teams in the world -- one that purports to be the modern example of what a soccer club is -- build a team with only two of the 150 best passers in its own league?

play

1:35

Will Tottenham get relegated from the Premier League?

Janusz Michallik debates Tottenham's Premier League survival hopes after their 1-0 loss to Sunderland.

The rise of the wrong analytics

Over the past few years, a new set of numbers has emerged in the soccer world. Rather than quantifying the things that lead to winning, they quantify the things that scouts and coaches have always seemed to value: Who is big and who is fast? Who looks good? Who would be unstoppable if I could teach him how to play?

A number of companies, like Gradient and SkillCorner, now offer a spate of physical metrics that show how often a player is running -- in and out of possession, at top speed, at high speed, etc. I don't fault any of the companies for doing this; it's good that these datasets exist. One of the things that's been missing from soccer data since the start is anything that tells us what everyone else is doing off the ball. The average player has possession of the ball, at most, only for a couple minutes per game, and most soccer data is only quantifying that tiny snapshot of time. It doesn't come close to telling us everything, but it is telling us the most important things.

Used properly, this off-ball, physical data can be incredibly powerful. If you're running a team and you can figure out how to combine these physical metrics with what drives winning and scoring goals, then you've created a new, much more holistic understanding of player value, and you'll have a leg up on anyone who is only using passing and shots to quantify performance. But that's really hard, and since it's really hard, it's not really happening.

Instead, as a source who has worked with a number of Champions League clubs put it to me, the physical metrics are allowing clubs just to confirm their own biases -- the same biases we've been talking about in this battle between scouts and stats since "Moneyball" was written. Except, now we have new stats that say the scouts were right.

How else to explain what happened with Spurs?

What Tottenham have is a roster filled with explosive athletes who can run. Using their physical metrics, Gradient created an "athleticism" score that's a combo of endurance, explosiveness and speed that adjusts for position and size. It's on a 1-100 scale. Tottenham have seven players at a 90 or above and five of them -- Wilson Odobert, Lucas Bergvall, Archie Gray, Dominic Solanke, Conor Gallagher -- were signed after Johan Lange became the club's technical director in October 2023. The first four were the four outfield players signed during Lange's first summer in charge.

You can't build a roster that can't pass unless you're systematically focused on a set of alternative player attributes that creates an institutional blind spot. Given that Romero -- by far their best passer -- was signed in 2021, and James Maddison, who has been injured all season but is easily their other best passer and was signed in the summer of 2023, the ignorance of what actually matters becomes even more stark.

One of the more memorable stories from "Moneyball" is the one where Billy Beane is arguing with his scouts, who are obsessing over how big a guy's butt is, what kind of face he has, or if his girlfriend is attractive. Beane keeps coming back to the question, "But can he hit?" Eventually he becomes enraged, and yells to everyone in the room, "I repeat: We're not selling jeans here."

I've heard it suggested that having someone who understands data and giving them an actual voice in your club is valuable simply because of all the things they'll keep you from doing, by reminding you to keep the main thing, the main thing. But can he hit? At Spurs, though, it seems like a new set of numbers might've blinded the club into thinking that they actually were in the business of selling jeans. What they really needed -- and what would've saved them from relegation -- was someone who kept asking a simple question:

But can he pass?