Case Study 1: How FBref and StatsBomb Changed Public Soccer Discourse

Background

For decades, public discussion of soccer performance was limited to the statistics available in match reports: goals, assists, clean sheets, and possession. Fans and pundits debated player quality using anecdote, reputation, and small handfuls of numbers that --- as Chapter 5 argues --- were poorly suited to measuring actual contribution.

This changed dramatically between 2017 and 2022. Two developments reshaped how the public engaged with soccer data:

1. StatsBomb (founded by Ted Knutson in 2017) began offering free, open-access event data for select competitions, including the FIFA Women's World Cup, La Liga, and several other leagues. Their data included detailed event-level information --- shot coordinates, pass end-points, pressure events --- that had previously been available only to professional clubs.

2. FBref (part of Sports Reference, powered by StatsBomb data from 2021 onward) made advanced per-90 statistics, percentile rankings, and scouting reports freely available for players across Europe's top leagues. Suddenly, anyone with an internet connection could look up a player's progressive passes per 90, pressures per 90, or expected assists.

The combined effect was transformative. Twitter (now X) accounts dedicated to soccer analytics exploded in popularity. Fans began debating xG, progressive carries, and pressing metrics in the same breath as goals and assists. Scouting, once the exclusive domain of professionals, became a participatory activity.

But this democratization also introduced new problems: misuse of metrics, over-reliance on per-90 stats without considering sample size, and the tendency to treat numbers as definitive verdicts rather than uncertain estimates.

Analytical Questions

This case study asks you to examine both the benefits and pitfalls of the public metrics revolution, using the concepts from Chapter 5.

Question 1: The Limitations of Pre-Revolution Statistics

Before StatsBomb and FBref, a typical fan debate about whether Player A or Player B was a better midfielder might cite:

• Goals scored
• Assists
• Pass completion rate
• "Big-game performances" (anecdotal)

Using the framework from Section 5.1, identify three specific analytical weaknesses of this approach. For each weakness, explain how the availability of advanced metrics (e.g., xG, xA, progressive passes) addresses it.

Question 2: Per-90 Misuse in Public Discourse

A common pattern on social media is the following argument:

"Player X has 0.85 progressive carries per 90 and Player Y has only 0.62. Player X is clearly the better ball-carrier."

The following data is available:

(a) Verify the per-90 rates.

(b) Explain why comparing these two per-90 numbers directly is problematic. Reference the minimum sample size guidelines from Section 5.3.2.

(c) Propose a more responsible way to present this comparison.

Question 3: The Percentile Trap

FBref's player pages display percentile rankings relative to positional peers. This feature is extremely popular but can be misused.

Consider a centre-back with the following percentile profile:

(a) A fan concludes: "This is an elite defender --- look at those defensive numbers!" Using concepts from Sections 5.1 and 5.4, explain why this conclusion may be flawed.

(b) What context adjustments would you want to apply before evaluating this player's defensive numbers?

(c) How might the low passing percentiles change the overall assessment?

Question 4: Building a Responsible Public Metric Dashboard

Imagine you are designing a free, public-facing player comparison tool (similar to FBref but with improvements based on Chapter 5's lessons). Design the tool by answering the following:

(a) Which five metrics would you display for attacking players, and why?

(b) What minimum minutes threshold would you enforce, and how would you communicate this to users?

(c) How would you display uncertainty so that casual users understand the limitations of small-sample statistics?

(d) What context adjustments would you apply behind the scenes, and which would you make visible to users?

Question 5: Simulation Exercise

Using the provided code (see code/case-study-code.py), simulate the following scenario:

Two players have identical true talent (0.35 goals per 90). Player A plays 2,800 minutes; Player B plays 600 minutes.

(a) Run 10,000 simulations and plot the distributions of observed goals-per-90 for each player.

(b) What percentage of simulations show Player B with a higher per-90 rate than Player A?

(c) What percentage of simulations show Player B with a per-90 rate more than double that of Player A?

(d) Write a paragraph interpreting these results for a non-technical audience (e.g., a blog post about why per-90 stats for low-minute players are unreliable).

Discussion Points

1. Democratization vs. Quality Control: The open-data movement has empowered millions of fans with analytical tools. But it has also led to widespread misuse of metrics. Is the net effect positive? What responsibilities do data providers have?

2. The Dunning-Kruger Effect in Analytics: As advanced metrics became public, many fans gained enough knowledge to be dangerous but not enough to be careful. How can the analytics community educate casual users without gatekeeping?

3. Metric Monoculture: When everyone uses the same FBref statistics, clubs lose their informational advantage. How might professional analytics departments maintain an edge in a world of open data?

4. The xG Backlash: Expected goals has faced periodic backlash from fans, pundits, and even some coaches who argue it is overused or misapplied. Using concepts from Section 5.6, how should the analytics community respond to this criticism?

Key Takeaways

• The public availability of advanced soccer metrics since 2017--2021 has transformed fan discourse, enabling more sophisticated analysis but also introducing new forms of misuse.
• Per-90 statistics require minimum sample size thresholds to be meaningful. Public platforms should communicate uncertainty and enforce or recommend minimums.
• Percentile rankings are intuitive but can mislead when context (opponent quality, team style, possession share) is not considered.
• Responsible metric communication requires transparency about limitations, not just presentation of impressive numbers.
• The best public-facing tools combine advanced metrics with clear guidance on interpretation, helping users avoid the most common pitfalls identified in Chapter 5.