Where sports meets architecture: 10,553 stadiums across 228 countries, coordinates on 96 percent, capacity on 6,578, and the capacity column tells the sport's whole economic story: the median venue holds 10,000 while the 100,000-plus club has exactly 19 members, Narendra Modi Stadium's 132,000 at the top. Opened years prefer P1619, the date of official opening, over P571 inception, because inception for buildings often means the first shovel: Wembley's inception says 2002, but the doors opened in 2007 and the dataset says so (both read best-rank with time and precision paired, the lesson the newspaper build taught). Capacity is the MAX across best-rank statements, answering the stable question "how many could it ever hold", with a physics guard: nothing seats more than 150,000, so attendance records and typos masquerading as capacities went NA, all 12 disclosed. The opening decades trace the booms: steady mid-century growth, then the 2000s as the busiest building decade in history (759 venues), stadium diplomacy and World Cup cycles visible in one histogram. Sport labels on 5,354 rows span football to cricket to speedway; tenants on 6,406 let you join venues to the clubs that fill them; 404 recorded closures with closed never before opened by test. Anchors are checked by QID, a practice the probe re-justified immediately: guessed stadium QIDs resolved to a rugby tour, an asteroid and Big Ben before the real Wembley, Camp Nou, Maracana and Melbourne Cricket Ground were located by label. Built with a 12-test acceptance suite.
10,553 rows in one CSV file, with 13 columns in total: Stadium, Stadium_ID, Country, City, Sport, Tenant, Capacity, Opened_Year, Opened_Decade, Closed_Year, Latitude, Longitude, Stadium_URL.
It is built by programmatically scraping wikidata.org, last pulled on 2026-10-10. Datasets are versioned; older versions stay downloadable.
Yes. Every CodeSights dataset is completely free as a CSV download; a free account is all it takes. Anyone can preview the data without signing in.
Yes. The exact Python scraper that built it is viewable on the dataset page by any signed-in member, so every number is reproducible.
Automated scraping leaves room for error and the underlying sources change over time, so no version is guaranteed accurate or complete. If a number matters, verify it against the original source.