Football Swarm Plots: Distributions, Density and Outliers

A ranking tells you who is first, tenth or fiftieth. It does not show whether those positions are separated by a large gap or packed into a dense cluster.

A swarm plot keeps individual observations visible along a metric scale. That makes it useful for reading distributions, highlighting subjects and questioning apparent outliers.

What Position Means

Each point’s position corresponds to its value on the metric scale. Points may be displaced slightly to avoid overlap, but the quantitative reading comes from their axis position—not from the vertical jitter used to separate them.

Before interpreting the shape, identify:

• metric and definition;
• total, per-90, percentage or other unit;
• period;
• comparison population;
• minutes or matches rule;
• highlighted players or teams.

Read Density Before Rank

Dense sections show that many observations occupy a similar part of the scale. Sparse sections and gaps show greater separation.

This changes the language of a report. “Ranked fifth” can sound important, but if positions two through fifteen are tightly clustered, the available data may not support a strong distinction.

The percentile guide connects rank to population. A swarm adds the missing view of spacing.

Treat Outliers as Questions

NIST describes an outlier as an observation unusually distant from others in a sample. Distance alone does not explain why it exists.

In football data, investigate:

• sample size;
• role and position;
• team style;
• set-piece or penalty responsibility;
• competition and period;
• data quality;
• whether the metric definition fits the player.

An extreme point can be informative, erroneous, structurally advantaged or genuinely unusual. The chart cannot decide which explanation is correct.

Choose a Coherent Population

A swarm is only as meaningful as its comparison set. Mixing unrelated positions can create a broad distribution that mostly reflects role differences.

Use a population relevant to the question and make exclusions visible. The fair comparison guide provides the full contract, while the small-sample guide covers exposure rules.

Highlight Without Erasing Context

Highlights should make the subject findable while leaving the population readable.

Use labels selectively:

• the player under discussion;
• a small number of relevant comparators;
• an observation being investigated;
• a threshold or benchmark explained in the text.

Avoid labelling only famous players if the article claims to analyse the full population. The unlabelled points are still evidence.

Use Multiple Metric Rows Carefully

Multiple swarm rows can compare where the same population sits across different metrics. Keep the ordering, units and role rationale explicit.

Do not treat left-to-right position across separate rows as a combined score. Each row has its own metric distribution. If the article needs a multidimensional profile, a player bar or radar may be more appropriate.

A Swarm Reading Checklist

1. Confirm metric and unit.
2. Confirm period and cutoff.
3. Confirm population and eligibility.
4. Inspect density and gaps.
5. Locate highlighted observations.
6. Check their minutes and matches.
7. Investigate outliers before explaining them.
8. State what the distribution cannot establish.

The Moment a Swarm Becomes More Useful Than a Ranking

Picture a recruitment meeting with twenty full-backs sorted by progressive carries per 90. A ranking gives the order, but the order may exaggerate tiny gaps. If the third, fourth and fifth players sit almost on top of one another, calling one “third best” adds more certainty than the data contains. A swarm exposes that cluster immediately.

The reading should begin in the middle of the distribution. Is there one dense band, two distinct groups or a long tail? Only then should the eye move to the highlighted candidate. A dot at the edge of a dense group is a different story from a dot separated by a large empty interval, even if both occupy the same formal rank in different samples.

Now add minutes. A highlighted outlier with a small sample is not a discovery to announce; it is a case to investigate. Raise the minutes threshold and observe whether the dot remains isolated. Switch from totals to per 90 and ask whether availability was driving the first view. Change the role population and watch whether the comparison still makes football sense. This sequence turns the swarm from a decorative strip of points into a sensitivity tool.

The final caption should say what the chart cannot. It can describe the metric, unit, season, population, eligibility rule and highlighted player. It cannot explain why the player is distant. That explanation belongs to role analysis, match context and video—not to the spacing between dots.

Multiple rows can then tell a richer story without becoming a scorecard. A full-back might sit in a dense cluster for progressive carries, near an edge for chance creation and close to the middle for defensive actions. The value is the contrast between distributions, not the number of rows in which the dot appears far to the right.

Keep the player highlight identical across rows and preserve the full population. If the surrounding dots fade so much that they disappear, the visual has stopped explaining a distribution and become a disguised player profile. The swarm earns its place precisely because the neighbours remain visible.

Build a Distribution View

Open the FBPlot Swarm Chart Creator, choose one metric and one coherent population, then highlight only the subjects needed for the argument. Before export, write a caption that includes the unit, period, population and exposure rule. A good swarm plot makes the surrounding evidence harder—not easier—to ignore.