How Poor Analytics Created…

There’s an interesting but often overlooked story in Michael Lewis’ “Moneyball” that doesn’t feature Billy Beane, thus it’s rarely retold in films.

In the late 1970s or early 80s, the Houston Astros commissioned a study to assess how relocating their outfield fences closer to home plate would impact team performance. Their expectation was that this change would result in more home runs, enticing fans and boosting ticket sales. However, the study revealed that, given the skills of their current roster, moving the fences in could actually lead to more losses for the Astros.

Despite the findings, club executives chose not to disclose the study results. They were already determined to pursue the fence relocation and sought only data that supported their pre-existing decision.

A parallel situation occurred at a professional soccer club, shared by an industry veteran. The team had tasked him with creating scouting reports for three players. His comprehensive assessments concluded that none of the players would be worthwhile signings. When he presented these findings, the club inquired if he could provide positive reports instead, as they had already committed to signing all three.

In both scenarios, organizations wanted to leverage data, but not for better decision-making. Instead, they sought to validate choices 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

---

These stories may feel outdated, but today’s baseball teams utilize advanced analytical models that aren’t readily available to the public. Similarly, soccer data is now ubiquitous; with Amazon enhancing Bundesliga broadcasts and concepts like “expected goals” becoming commonplace in English-language commentary.

Nonetheless, while baseball has evolved past merely using statistics to confirm biases, many soccer clubs are still lagging behind. Just look at Tottenham Hotspur, the club reportedly contemplating how to tell fans it had “redefined what a modern football club can be.”

Understanding Soccer Analytics

At the heart of soccer analytics lies a fundamental truth: the best team doesn’t always win.

This principle is epitomized by the concept of expected goals (xG). A team’s xG differential at any point in the season is a better predictor of future performance than traditional metrics like shots, goals, or points. If the best team always emerged victorious, then earlier wins would naturally indicate the strongest contenders and predict outcomes accordingly.

Instead, it seems the leading teams accumulate a higher ratio of expected goals during matches. Simply put, the top teams are those that create more valuable scoring chances than their opponents.

This might be intuitively understood by anyone who has watched or played the game over time—whether they want to accept it or not. Acknowledging this fact reveals the significant randomness inherent in soccer, as there’s considerable unpredictability in trying to score with a moving ball and a restricted goalkeeper.

The Premier League season may be relatively short, consisting of around 20 team-level experiments. Over a decade, that translates to about 200 experiments. Throughout these seasons, it is expected to witness instances where randomness significantly impacts a team’s success or failure during the year.

The data backs this up. When we analyze every Premier League season since 2010 based on xG differentials, we find notable examples of extreme over- or underperformance.

The far-right team represents Tottenham in the 2016-17 season, while Tottenham in 2025-26 would fittingly occupy the far-left spot. For a team among the top 10 richest in the world to be fighting relegation with mere games left in the season, it’s tempting to attribute this to “historical bad luck,” right?

Not so. That honor goes to Sheffield United in 2023-24.

This season, Tottenham isn’t an outlier. Their goal differential (+11) is marginally better than their xG differential (+15.13).

So how does a club with one of the world’s most valuable rosters find itself among the league’s worst? One explanation: they prioritize metrics that don’t truly matter.

Tottenham’s Primary Challenge: Passing Ability

Normally, soccer is a complex game where individual player skills are intertwined with team dynamics, managerial guidance, and on-field collaboration. Yet in the case of Tottenham, the diagnosis appears straightforward: the team struggles with passing.

At Gradient Sports, analysts evaluate every Premier League match, grading each pass from -2 to +2. Their methodology focuses on gauging performance rather than merely outcomes. For instance, a simple pass to an open teammate unchallenged would score a ‘0’, whereas a precise pass under pressure would gain positive marks.

Evaluating Tottenham’s top five passers for this season reveals concerning rankings:

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 cornerstone of soccer play. Each Premier League team attempts about 450 passes per game—far more than any other statistical category. Without effective passing, every other skill is rendered moot; it is the force that underpins all else in the game.

So how does a wealthy club—one that claims to embody the modern concept of a soccer organization—end up featuring only two players among the top 150 passers in their own league?

The Misguided Focus on New Analytics

In recent years, a wave of new metrics has emerged in the soccer realm. Rather than measuring elements that contribute to wins, these metrics analyze attributes traditionally valued by scouts and coaches: size, speed, and appearance.

Companies like Gradient and SkillCorner now provide numerous physical metrics that track players’ running frequencies—both with and without the ball, at various speeds. While there’s nothing wrong with this data, it’s crucial to recognize that capturing what players do off the ball has long been absent in soccer analytics. Most players spend only a few minutes in possession during a match, and available data often fails to depict the bigger picture.

When appropriately leveraged, off-ball physical metrics can enhance our understanding of player value significantly. A team that successfully integrates this data with winning-related metrics can establish a much more comprehensive evaluation compared to those utilizing only traditional stats like passing and shots. However, achieving this integration is challenging, and many teams are not doing it effectively.

A source active within several Champions League clubs articulated that, instead of leading to insights, physical metrics often end up affirming existing biases. This is reminiscent of the long-standing tension between scouts and statistics since “Moneyball” was penned—only now there are new stats reinforcing the scouts’ viewpoints.

This raises questions about Spurs’ situation.

Tottenham’s roster is packed with athletes capable of running quickly. Gradient’s “athleticism” score—a blend of endurance, explosiveness, and speed—ranks players. Tottenham boasts seven players scoring 90 or above, several of whom were signed under the club’s current technical director, Johan Lange.

The issues arise when a team is constructed without recognizing vital factors, like passing accuracy. Given that Romero, their top passer, was signed in 2021, and Maddison—another standout passer—joined in summer 2023, the neglect of critical areas becomes glaringly evident.

In “Moneyball,” a memorable scene showcases Billy Beane’s frustration as scouts fixate on irrelevant attributes—like a player’s physique or attractiveness of their partner—instead of performance. Beane continually emphasizes that crucially, “we’re not selling jeans here.”

There’s a persuasive argument for having someone who understands data and advocates for evidence-based decision-making within a club. Such an individual would prevent unwise choices. Yet, at Spurs, it feels as if the organization’s embrace of a new class of metrics may have blinded them to the essentials. What they ultimately need—and what might have saved them from relegation—is someone who continually poses the fundamental question:

But can he pass?