The premise
What counts as a global metro depends on what you are measuring. A finance-only ranking puts London, New York, Hong Kong, and Singapore at the top. A tourism-only ranking promotes Paris, Tokyo, Bangkok, and Istanbul. A sports-only ranking produces a different list again. None of these single-axis views describes the whole picture, which is the gap this project addresses.
I settled on sixteen dimensions because that is where each axis is meaningful without any single one being able to swing the result. A metro can reach the upper tiers through any combination of strengths. New York is rank 1 because it competes on every dimension. Boston sits in the top thirteen largely on the Harvard and MIT cluster. Houston earns top 25 mostly on energy-sector market cap. All three routes count.
The rankings are not a popularity contest and not a forecast. They are an inventory of what each metro currently has, expressed on a continuous scale.
The sixteen dimensions
Every dimension below has a weight, a shape, and a source. Linear means the contribution scales directly with the input. Capped means the contribution stops growing past a ceiling, so an outsized input cannot dominate. Log-scaled means the contribution grows fast at low values and slows at high values, which fits counts where the hundredth station is worth much less than the tenth.
The Source line on each dimension names the upstream third- party dataset, ranking, or organisation behind it. Where a recognised ranking is the truth (CWUR, GaWC, CTBUH, ACI, WFE, BIS, Lloyd's List, UFI, Newsweek, ICOLD, the Michelin Guide, Forbes Travel Guide), it is named so any number on the page can be traced back to its origin.
Population
Weight: pop / 3,000,000 · linear
Sets a baseline floor. A metro with thirty million people gets credit for that scale before any other dimension is counted. The 3M divisor keeps the contribution moderate so population alone never decides the ranking.
Source: citypopulation.de for both municipality-level data (approximately 133,000 entries across 33 countries with full municipality coverage: US, Canada, UK, Germany, France, Spain, Italy, Japan, South Korea, Australia, Brazil, Mexico, China, India, Russia, Turkey, Poland, Netherlands, Belgium, Austria, Switzerland, Sweden, Norway, Denmark, Finland, Czech Republic, Portugal, Ireland, New Zealand, South Africa, Argentina, Colombia, Chile) and county/district-level data (approximately 49,000 entries across 237 countries) used as the secondary lookup for everywhere else. Country-level totals reconciled against Wikipedia's 'List of countries and dependencies by population' (en.wikipedia.org/wiki/List_of_countries_and_dependencies_by_population).
Major league teams
Weight: raw count, capped at 10 · capped
Top-tier franchises signal civic identity, broadcast economy, and cultural reach. The cap stops a city with eight or more major-league teams from sweeping the score on this dimension alone.
Source: Official league websites for NFL, NBA, NHL, MLB, and MLS, plus Wikipedia season articles for every other tracked competition: European top-flight football (Premier League, La Liga, Serie A, Bundesliga, Ligue 1, Eredivisie, Primeira Liga, and equivalents), EuroLeague basketball, AFL, CFL, top-flight rugby union (Premiership, Top 14, URC, Super Rugby Pacific), top-flight rugby league (NRL, Super League), and IPL plus other major T20 cricket leagues.
Other professional teams
Weight: (total teams − major league) × 0.25, capped at 10 · capped, low weight
Second-tier and minor-league teams count, but at a fraction of the marquee franchises. The 0.25 multiplier and the 40-team pre-cap reflect that.
Source: Wikipedia season articles for second-division and minor-league rosters across every sport listed above, plus Wikipedia NCAA program lists by conference (FBS, FCS, basketball).
Market capitalization
Weight: USD / 700,000,000,000 · linear
Total enterprise value of public and large private companies headquartered in the metro. The $700B divisor is calibrated so a single trillion-dollar company adds roughly 1.4 points to the score, not 14.
Source: Three upstream sources by company type. Public companies: companiesmarketcap.com. Private companies: Wikipedia's 'List of largest private non-governmental companies by revenue' and the Forbes Largest Private Companies list (forbes.com). Unicorns: CB Insights' Research Unicorn Companies (cbinsights.com). Approximately 12,475 entries combined, refreshed roughly weekly.
Number of major companies
Weight: implicit, via market cap accumulation · linear
Every company above the inclusion threshold is in the corpus. Count is not weighted separately because count and total market cap are tightly correlated, and double-counting would bias the result toward US tech-heavy metros.
Source: Same three upstream sources as Market capitalization above (companiesmarketcap.com, Wikipedia / Forbes for private companies, CB Insights for unicorns). Count is informational only; the score uses cumulative market cap, not the count, to avoid double-rewarding metros with many small public companies.
Major cultural events
Weight: × 0.65 (combined with museums and infrastructure) · linear
World expos, NATO summits, G7 and G20 hostings, World's Fairs, royal weddings, papal events. One-off cultural moments that put a city at the center of global attention for a period of days.
Source: Manually compiled from Wikipedia lists for World Expos, World's Fairs, NATO summits, G7 and G20 hostings, plus Vatican press records for papal events, royal-wedding press records, the World Marathon Majors organizers, the Cannes Film Festival, Oktoberfest, the Tour de France, the Masters Tournament, the Tennis Grand Slam organizers, and individual festival/biennial organizers. Historical Events sub-category (added 2026-05-01) is sourced manually from Wikipedia and primary historical references.
Museums and landmarks
Weight: × 0.65 (combined with cultural events and infrastructure) · linear
Museums, opera houses, concert halls, parks, religious sites, theme parks, world heritage sites, plus iconic bridges, tunnels, dams, and canals. The durable cultural and engineering assets that draw visitors and signal civic priority.
Source: museumworldranking.net (Top 426 museums globally), TEA/AECOM Global Experience Index 2024 (theme parks), TripAdvisor plus TEA plus AZA/EAZA accreditation (zoos and aquariums), Wikipedia 'List of contemporary amphitheatres', Wikipedia 'List of concert halls', Wikipedia 'List of opera houses', Wikipedia 'List of the world's largest libraries' (10M+ items), and US National Park Service 2024 visitor data for US national parks. Bridges, tunnels, dams, and canals from Wikipedia (longest bridges, longest tunnels, largest dams, ship canals), ASCE Monuments of the Millennium (asce.org), and ICOLD for major dams. UNESCO World Heritage Sites integration is in progress.
Ports, exchanges, and other infrastructure
Weight: × 0.65 (combined with cultural events and museums) · linear
Container ports, stock exchanges, internet exchanges, military bases, central banks, data center hubs, agricultural and extraction hubs, and trade venues. The plumbing of a globally relevant city.
Source: Container ports: Lloyd's List Top 100, World Shipping Council Top 50, and AJOT Top 100. Passenger ports: CLIA State of the Cruise Industry Report 2025 plus AAA Cruise Forecast 2025. Trade venues: UFI World Map of Exhibition Venues 2025 (80,000 square metre minimum). Stock exchanges: World Federation of Exchanges (WFE) member list (world-exchanges.org) plus Wikipedia commodity-exchange lists. Internet exchanges: DE-CIX (de-cix.net), AMS-IX (ams-ix.net), IX.br, PeeringDB (peeringdb.com), Internet Society Pulse (pulse.internetsociety.org), and the Newby Ventures IXP directory. Data center hubs: Cushman & Wakefield Global Data Center Market Comparison (cushmanwakefield.com) plus Cloudscene (cloudscene.com). Central banks: BIS member directory (bis.org) plus Wikipedia 'List of central banks'. Military bases: DMDC (dmdc.osd.mil), Pentagon Base Structure Report 2024, Congressional Research Service reports, plus Wikipedia country-specific lists. Agriculture and extraction: FAO GIAHS Programme (fao.org/giahs), OriGIn geographic indications (origin-gi.com), CGIAR research centres (cgiar.org), MINING.com, USGS, plus company filings.
Airport score
Weight: × 0.25 · linear, low weight
Tier-1 mega-hubs (JFK, Heathrow, Hong Kong) score higher than tier-2 international gateways, which score higher than tier-3 regional airports. The weight is low because international connectivity is already partly captured by other dimensions.
Source: ACI World Airport Traffic Dataset 2024 (final, released July 2025), supplemented by Wikipedia 'List of busiest airports by passenger traffic' and FAA classifications. Each entry is tier-classified 1 (mega-hub) through 5 (regional) using a composite of passenger volume, freight volume, and international destination count.
Top-50 universities
Weight: × 3.5 per qualifying institution · linear, high weight
A top-50 global university is a generational asset. Boston ranks far higher than its population would predict because of the Harvard and MIT cluster, and the formula reflects that.
Source: Center for World University Rankings 2024 (cwur.org), filtered to global rank ≤ 50.
Other top-500 universities, hospitals, research
Weight: × 2.2 per institution (top-500 minus top-50) · linear
Universities ranked 51 to 500, top-250 hospitals, and major research institutions. Hospitals and research institutes carry half the weight of a university because they cluster less around a single peak ranking.
Source: Center for World University Rankings 2024 (cwur.org) for ranks 51 to 500, Newsweek World's Best Hospitals 2024 (Top 250), and a manually-curated set of research institutions (47 entries across biomedical research institutes, applied research labs, national laboratories, independent think tanks, and policy research institutes) verified against each institution's own publicly-published profile.
Metro and subway stations
Weight: log(stations) · log-scaled
Log-scaled because adding the 200th station matters much less than adding the 20th. Tokyo at over 1,000 stations should not overwhelm London at 270.
Source: Wikipedia 'List of metro systems' (en.wikipedia.org/wiki/List_of_metro_systems). Subway, light rail, and tram lines all counted; per-line station counts preserved.
Commuter rail stations
Weight: log(stations) × 0.5 · log-scaled, low weight
Suburban rail extends a metro's effective catchment. The 0.5 multiplier is half the metro-stations weight because commuter rail is functionally less central to daily urban life than the subway.
Source: Wikipedia 'List of suburban and commuter rail systems' (en.wikipedia.org/wiki/List_of_suburban_and_commuter_rail_systems).
Intercity train hubs
Weight: log(hubs) × 2 · log-scaled, high weight
Major intercity rail hubs (London King's Cross, Tokyo Station, Gare du Nord) require both a dense national rail network and a city worth converging on, which is why they sit at high weight.
Source: Wikipedia 'List of busiest railway stations' (en.wikipedia.org/wiki/List_of_busiest_railway_stations). Threshold is roughly 30 million or more passengers per year.
Skyscrapers and towers
Weight: log(150m+ count) × 5.7 · log-scaled, high weight
Buildings over 150m are a near-universal indicator of capital concentration and land-value pressure. The high multiplier reflects how hard this metric is to fake. Building tall requires sustained economic and political consensus.
Source: Council on Tall Buildings and Urban Habitat (CTBUH) via skyscrapercenter.com. 150m+ buildings only; tier counts (150m+ / 200m+ / 300m+) preserved per metro.
Luxury hospitality (Michelin and Forbes Travel Guide)
Weight: log(stars) × 3 · log-scaled
Michelin 2- and 3-star restaurants plus Forbes Travel Guide 4- and 5-star hotels. A density measure of high-end consumption that tracks discretionary income, tourism, and cultural prestige.
Source: Michelin Guide 2025 (442 two- and three-star restaurants across 105 metros) plus Forbes Travel Guide 2026 Star Awards (forbestravelguide.com): 343 Five-Star and 708 Four-Star hotels. Star count weighted 3-star × 3 + 2-star × 2 for restaurants; Five-Star × 3 + Four-Star × 2 for hotels.
Major sporting events
Weight: × 0.2, capped at 4 · capped, low weight
Olympic and multi-sport hosting plus championship finals. Capped at 4 so a city that hosted three Olympics last century cannot outrank a finance capital on this dimension alone.
Source: Wikipedia season and event articles for Olympic Games, FIFA World Cup, tennis Grand Slams, Formula 1 Grands Prix, and NASCAR Cup Series, plus continental finals (UEFA Champions League, Copa Libertadores, AFC Champions League, CAF Champions League, and equivalents).
Annual recurring events
Weight: raw count · linear
F1 Grands Prix, marathons, tennis Grand Slams, the Cannes Film Festival. Each annual draw is one point, with no cap, because a city hosting five different marquee fixtures should get the cumulative credit.
Source: Per-fixture official sources: the Formula 1 official calendar, the Wikipedia NASCAR Cup Series season articles, the World Marathon Majors organizers, the Tennis Grand Slam organizers, the BWF World Tour calendar (badminton), WTT (table tennis), Riot Games (esports), evo.gg (fighting games), the Cannes Film Festival, and the European Broadcasting Union for Eurovision.
GDP tier bonus
Weight: +3 / +2 / +1 / +0.5 / 0 · tiered bonus
Economic mass that the rest of the formula does not directly capture. Above $500B gets +3, $200B to $500B gets +2, $50B to $200B gets +1, $10B to $50B gets +0.5. Tiered rather than linear because US and Japanese metro-GDP estimates run unusually high and a linear bonus would skew the result.
Source: Wikipedia 'List of cities by GDP' (en.wikipedia.org/wiki/List_of_cities_by_GDP). Reference years vary 2019-2025 by metro; each figure is in billions USD.
GaWC class adjustment
Weight: 12 × (1 / GaWC class) · inverse, capped
The Globalization and World Cities Research Network's tier (1, 2, 3, 4, 5, 6) is a 25-year academic standard for ranking cities by global integration. Class 1 metros get +12, class 2 get +6, and so on. It is the only externally-sourced rank that feeds the composite, and it acts as a check on the dimension-level math.
Source: Globalization and World Cities Research Network 2024 World Cities Index (lboro.ac.uk/microsites/geography/gawc). Class 1 = Alpha++ down to Class 12 = Sufficiency. Approximately 280 metros classified.
Score tiers
Raw scores are continuous, but readers need categorical labels to talk about. The seven tiers below carve the distribution into bands that align with the way urban economists already group cities. The boundaries echo the GaWC alpha, beta, and gamma convention without requiring readers to know the academic shorthand.
Adjustments and design choices
Three families of adjustment apply across the formula. Each is worth surfacing so the choices are visible rather than hidden inside the spreadsheet.
Caps on counted dimensions
Major league teams are capped at ten and major sporting events at four. The caps stop a single dimension from running away with the score in the rare metros where one input is multiplied. New York with thirty-plus teams could otherwise clear everyone on team count alone. The cap forces its score to come from breadth across dimensions.
Log scaling on count metrics
Metro stations, commuter rail stations, intercity hubs, skyscrapers, and luxury stars are all log-scaled. The claim is simple: the two-hundredth station or fortieth Michelin star is worth much less than the tenth. Linear scaling would over-reward Tokyo and Hong Kong, two metros where count metrics are unusually high.
Tier bonuses for GDP and GaWC class
GDP is tiered (+3 / +2 / +1 / +0.5 / 0) rather than linear so that the rest of the formula does most of the work and GDP acts as a sanity check. The GaWC adjustment uses the 2024 GaWC tier as an external benchmark: class 1 metros get +12, class 2 get +6, and so on. GaWC is the only outside ranking that feeds the composite, and it is there to test the dimension math against an independent yardstick.
Declared editorial decisions
The formula is deterministic. The input data is not. Below are the editorial choices behind the corpus, each one documented so you can decide whether you agree.
Metro corridor consolidation
Several entries in the rankings are multi-city corridors rather than single-municipality metros: Washington-Baltimore, Rhine-Ruhr (Cologne, Düsseldorf, Essen), Saxon Triangle (Leipzig, Dresden, Chemnitz), San Francisco-San Jose (the Bay Area), Osaka-Kyoto-Kobe, Padua-Venice, Boston-Cambridge, Hannover-Brunswick. Each was consolidated because the labor market and infrastructure are genuinely shared. Treating them as separate metros would misrepresent reality.
Continent assignment overrides
A handful of countries are assigned to a continent that differs from strict geography so the regional aggregates make economic and political sense. Australia and New Zealand are grouped with Asia. Turkey, Israel, Russia, Ukraine, Belarus, Kazakhstan, Georgia, Armenia, and Azerbaijan are all grouped with Europe. For the purposes of regional rankings, economic gravity and political alignment carry more signal than tectonic geography.
Naming conventions and disambiguation
Where multiple metros share a name, the rankings use a disambiguating suffix or an ASCII variant: Cordova (Argentina) versus Córdoba (Spain), Toledo (Spain) versus Toledo (Ohio), Naples (Italy) versus Naples (Florida). Sao Paulo appears without the diacritic for keying consistency. These are deliberate.
Bloated municipality counts in some Spanish metros
Some Spanish metros aggregate a large number of small municipalities (Burgos at 371, Salamanca at 362, Zaragoza at 293). The structure follows the Spanish national statistics agency (INE) definitions. I reviewed each one and left them intact rather than impose a US-style core-cluster approach.
Known limitations
Three places where the rankings are weakest, in order of how much it bothers me.
Developing-world coverage is uneven. Sub-Saharan Africa, parts of Central Asia, and rural China and India have municipality and county data far sparser than the OECD comparison set. Some metros in these regions are likely under-counted on dimensions that need granular sub-metro data. I am working to close the gap, but the source data does not yet exist at parity.
Cultural events skew toward Western recognition. The Cultural Events corpus reflects the events I know about, which biases toward NATO summits, world expos, papal visits, and English-language coverage. A South American or African counterpart of equal scale may be missing simply because I have not sourced it yet.
The formula is opinionated. The weights are mine. A reader who thinks luxury hospitality should not be a dimension at all, or that GDP should weigh three times more, will produce a different ranking. This page exists so the disagreement can land on a specific choice rather than on the project as a whole.
Data vintage and refresh cadence
Most dimensions refresh when their source dataset is updated, and that timestamp drives the "Updated" chip in the site navigation and the lastUpdate field in the dataset metadata. Market cap data refreshes on its own cadence, and its age is shown on every metro page's Top Companies block as "Source data as of YYYY-MM-DD."
The composite is licensed CC-BY for reuse. If you cite the rankings, use "Global Metro Power Rankings, Citizen of Nowhere, [date]" with a link to the relevant metro page or to this methodology page. Third-party source data keeps its original license; the CC-BY scope covers the composite score and its derived rankings only.
This methodology version
This page documents the methodology as of the last refresh. Material changes to the formula are logged on the release notes page, and prior versions of the methodology are kept when material changes occur. Minor weight calibration is not a material change. Adding or removing a dimension is.