National Team Matchup Grid

A Scorigami-inspired heatmap of international football. Every current FIFA member nation is a row and a column of one massive grid; each cell is coloured by how many times those two national teams have ever played each other. The point of the picture is the negative space — the matchups that have never happened.

▶ Live demo: https://dgoodenough.github.io/national-team-grid/

Only 29% of all possible men's matchups have ever been played — and only 13% of the women's. The rest of the grid is empty: pairs of countries that, in 150+ years of international football, have simply never met.

How to read it

• Rows and columns are national teams, grouped by confederation (AFC, CAF, CONCACAF, CONMEBOL, OFC, UEFA) and then ordered by FIFA ranking within each confederation. Teams in the same confederation play each other far more often, so the grid lights up in dense blocks down the diagonal, with sparse cross-confederation regions between them.
• Cell colour = number of times the pair has met (log-scaled, cool → hot). The most-played men's fixture is Argentina–Uruguay (183 meetings since 1902).
• Empty (near-black) cells = the two teams have never played. Flip on Highlight never-played to invert the emphasis and make those cells glow.
• The diagonal is greyed out (a team can't play itself).

Features

• Zoom & pan the full 211×211 grid (mouse wheel or the on-grid −/Fit/+ controls; drag to pan).
• Men's ↔ Women's ↔ Both toggle — two completely separate match archives, each with its own current FIFA ranking. Both overlays them into a four-colour view: pairings that have met in both games (green), the men's only (blue), the women's only (orange), and neither — so you can see where one game has explored a matchup the other hasn't (≈4,000 men's-only vs a few hundred women's-only today, and the gap closes as you scrub forward through time).
• Sort three ways: Confederation, then FIFA rank (default — confederations form blocks); FIFA rank, global (straight down the line, confederations interleave into a colour stripe); or Alphabetical.
• Filter by confederation — show one, several, or all.
• Click any cell — see every meeting between those two teams (year, score, tournament) plus the head-to-head W–D–L record. The per-match detail is loaded on demand on the first click, so it doesn't slow the initial grid.
• Hover any cell for a quick tooltip (meetings + first/last year).
• Manual team checklist — tick one team to see all of its matchups ranked from most- to least-played (with a never-met tally; in Both mode, opponents are grouped by which game(s) they've met in); tick two or more to build a custom sub-grid using the chosen sort.
• Highlight never-played mode for the full Scorigami effect.
• Upcoming FIFAGami — highlight in yellow the pairs that have never met but have a scheduled fixture, plus a scrollable list of them (soonest first; click one for the head-to-head). Fixtures come from martj42's advance listings merged with ESPN's public scoreboard across ~30 senior international competitions (both genders), looking up to two years out. The browser's current date decides what's still "upcoming" vs already played, so it stays accurate between data refreshes.
• Timeline scrubber — drag through the years and watch the grid fill in; empty cells are matchups that hadn't happened yet. (First pass: a team that didn't exist in a given year just shows as not-yet-met, rather than being greyed out.)
• Include defunct teams (advanced) — adds historical sides with no single modern successor (Yugoslavia, Czechoslovakia, East Germany / German DR, the Saar, South Vietnam, South Yemen…).
• A live headline counts how many of the possible matchups in the current view have never happened.

Data

Dataset	Source	Used for
Men's internationals (1872–present, ~49k matches)	martj42/international_results	match counts, first/last meeting
Women's internationals (1969–present, ~12k matches)	martj42/womens-international-results	match counts, first/last meeting
Historical team renames	former_names.csv (martj42)	folding old names into current teams
Current FIFA ranking, men's + women's	FotMob (mirrors the official ranking)	rank ordering within / across confederations
Scheduled internationals (men's + women's)	ESPN public scoreboard API	upcoming first-ever meetings, up to ~2 years out
Confederation membership	cnc8/fifa-world-ranking	which confederation each team belongs to

Methodology notes

• Current FIFA members only (211 teams) in the default view. Confederation membership comes from the cnc8 snapshot, plus a tiny supplement (data/members_extra.csv) for any member missing from it (e.g. Cook Islands).
• Name reconciliation. The three datasets spell some teams differently (Côte d'Ivoire↔Ivory Coast, Korea Republic↔South Korea, FotMob's USA/Turkiye/Czechia, …). build.py reconciles them all, and former_names.csv folds historical names into their modern team (Zaïre→DR Congo, Upper Volta→Burkina Faso; USSR→Russia and Serbia & Montenegro→Serbia, per the source's lineage). Any team that resolves to neither a current member nor a curated defunct side (mostly non-FIFA territories like Martinique or Jersey) is excluded, and the build prints a full report of them.
• Rankings are current and per-gender. build.py pulls the latest men's and women's FIFA rankings from FotMob at build time (FIFA's own ranking API is gated; FotMob mirrors it). The exact publication date used is recorded in members.json and shown in the app footer. cnc8's 2020 men's rank is kept only as an offline fallback. ~14 FIFA members have never been given a women's ranking; they sort last within their confederation.

Build it yourself

build.py uses only the Python standard library (no dependencies).

python build.py            # download sources, reconcile, aggregate, write docs/data/*.json
python build.py --refresh  # force re-download of every source (latest matches + latest ranking)

It downloads the sources into data/raw/ (gitignored), then writes the artifacts the site loads:

• docs/data/members.json — ordered members with confederation + men's/women's rank + data vintage
• docs/data/matrix_men.json, matrix_women.json — sparse [i, j, meetings, firstYear, lastYear]
• docs/data/matches_men.json, matches_women.json — per-pair meeting detail (year, score, tournament), lazy-loaded by the site only when a cell is clicked
• docs/data/defunct.json — the advanced defunct-teams layer
• docs/data/upcoming.json — scheduled first-ever meetings (upcoming FIFAGami)

Then serve the static site from docs/:

python -m http.server -d docs 8000   # open http://localhost:8000

Data vintage & updating

The match archives are the live edge of the project — martj42 updates them within a day or two of most internationals. The current cut-off date for each archive is recorded in members.json (data_through) and shown in the app footer. Everything is easy to refresh or extend:

• More recent matches — just re-run python build.py --refresh. It re-pulls the latest martj42 results and the latest FotMob ranking, so the grid moves forward with no code changes.
• Override a ranking — drop a data/ranking_men.csv or data/ranking_women.csv (name,rank header, names matching the canonical member names) and the build uses it instead of FotMob. Handy for pinning a specific publication or hand-correcting.
• Add a missing member — append to data/members_extra.csv (name,code,confed).
• Append your own matches — the aggregation reads plain CSVs with martj42's columns (date,home_team,away_team,home_score,away_score,tournament,city,country,neutral); extra rows in data/raw/results_*.csv flow straight through.

Roadmap / ideas

• Cluster analysis. The grid already hints at tight clusters — the British Isles (England / Scotland / Wales / Northern Ireland / Ireland) light up as a dense little block. A proper community-detection pass over the "have-played" graph could surface these automatically.
• "Closest to never having met." Rank the pairs that have met the fewest times (the rarest one-off fixtures), and the teams that have faced the fewest distinct opponents — the loneliest nodes in the graph.

Tech

Pure static site — vanilla JavaScript + HTML5 Canvas, no build step and no runtime dependencies. The ~44,000-cell grid is drawn directly to a canvas with view-culling for smooth zoom/pan. Pointer Events drive one interaction path for both mouse (hover, drag, click, wheel) and touch (drag to pan, pinch to zoom, tap for detail), and the layout reflows for phones. Hosted on GitHub Pages from docs/.

Styled with Ledger (vendored tokens.css) — greyscale carries structure, color carries meaning: the green ramp is "how often they've played", red is "never", yellow is "upcoming".

Credits

Match data © the martj42 datasets; current FIFA rankings via FotMob; upcoming fixtures via ESPN's public scoreboard API; confederation mapping from cnc8/fifa-world-ranking. Concept inspired by Scorigami (Jon Bois). Code under the MIT License.