---
title: "What we hold, per sport and league (the discovery catalogue)"
method: GET
path: "/v1/coverage"
tags: ["coverage"]
---

# What we hold, per sport and league (the discovery catalogue)

`GET /v1/coverage`

## Request

- Base URL: `https://api.bigballsdata.com`
- URL: `https://api.bigballsdata.com/v1/coverage`
- Auth: one of:
  - API key in header `x-api-key`
  - HTTP bearer

The catalogue an agent reads to decide whether a dataset exists here, precomputed on a schedule and served from a snapshot. Every league we hold appears with its match counts, seasons, freshness and per-datatype coverage.

VOCABULARY. `data_types` is the stable default and does not change. `?v=2` adds `data_types_v2` beside it, which answers questions about a GRAIN rather than a datatype: "do you have xG" is a v1 question, "do you have PER-PLAYER xG" is a v2 one. In v2 each measure carries its grains (subject x period, e.g. `team.match` or `player.season`), every grain declares the denominator its figures count, and `not_applicable` names measures that CANNOT apply to a league as distinct from ones we simply lack. `data.vocabulary` echoes which shape was served.

v2 is opt-in because the summary answers most questions and the nested shape is only needed for grain questions. An unrecognised `?v=` serves v1 rather than failing: a catalogue call is discovery, and a query-string typo should not 400.

## Query parameters

- `sport` 'football' | 'basketball' | 'baseball' | 'ice_hockey' | 'cricket' | 'mma' | 'boxing' | 'formula1' | 'american_football'
- `v` '2'

## Response `200`

ok. `{ data, meta }`. Declared no `content` until 2026-09-25, on the one endpoint whose entire job is discovery.

- object: The standard response wrapper. Only `data` is guaranteed. `error` is absent (not null) on 16 of 40 live 200 responses measured 2026-09-18, so test for the KEY before reading it; this block declared it required until then. On a non-2xx, `error` is present and `data` is absent.
  - `data` object, required: Shape read off the live v1 snapshot on 2026-09-25 (9 sports). `data_types_v2` is present only with `?v=2`, which was NOT exercised by that read, so it is declared open rather than described.
    - `generated_at` string, date-time: When the snapshot was precomputed, NOT when you asked.
    - `snapshot_age_seconds` number
    - `vocabulary` string: Echoes which shape was served, so a `?v=` typo is visible rather than silent.
    - `totals` object
      - `sports` integer
      - `leagues` integer
      - `matches` integer
      - `counts` object
      - `leagues_by_depth` object
      - `leagues_by_sport` object
    - `sports` object[]
      - `name` string
      - `slug` string
      - `seo_url` string
      - `aliases` string[]
      - `leagues` object[]
  - `meta` ResponseMeta: Response metadata. NOTHING here is guaranteed: measured over 38 live envelopes on 2026-09-18, `cached` and `request_id` appeared in 29, `source` in 28, `cache_age_ms` in 9 and `confidence` in 6. This block previously declared all five REQUIRED, which no response has ever satisfied. Treat every field as optional and branch on its absence. Routes add their own keys here (`count`, `limit`, `offset`, `note`, `total`, and per-endpoint ones), so do not assume a closed set.
    - `date_window` object: Present when ?date= was applied. The exact window the filter used, so an omission at a day boundary is visible. start_utc is inclusive, end_utc exclusive. timezone is UTC, or the zone named by ?tz=, in which case the window is that local day. Boxing card dates are published event dates, so timezone and bounds are null there.
      - `date` string
      - `timezone` string, nullable
      - `start_utc` string, date-time, nullable
      - `end_utc` string, date-time, nullable
      - `note` string
    - `source` 'official-league' | 'aggregator-paid' | 'aggregator-free' | 'community-scraper' | 'forecast' | 'mixed' | 'cache' | 'none' | 'unknown-source' | 'stored' | 'catalogue' | 'gateway' | 'coverage-snapshot': Opaque source-tier label. Customers never see the vendor identity. `forecast` marks a weather forecast (GET /v1/matches/{id}/weather).
    - `confidence` number, float
    - `cached` boolean
    - `cache_age_ms` integer
    - `request_id` string
    - `fields_missing` string[]
    - `note` string: Short factual disclosure about something this response deliberately withheld, present only when there is something to disclose. On live scores it appears when `match_id` could not be resolved to a stored match and was therefore omitted: it gives the counts and points at `/v1/stored/matches`, whose ids you can call back into. Absent when every served id resolves. Distinct from `message`, which explains `coverage: false` or a stale response. This one is attached to a response that DID serve data and names the part of it that is deliberately missing.
  - `error` ApiError
    - `code` string, required
    - `message` string, required
    - `details` unknown

## Other responses

- `503`: The snapshot is still being built. Carries Retry-After; retry in about a minute.

---

[API](https://skmtc.dev/bigballsdata/apis/big-ball-sports-api.md) · [All operations](https://skmtc.dev/bigballsdata/apis/big-ball-sports-api/llms.txt) · [OpenAPI document](https://skmtc.dev/bigballsdata/apis/big-ball-sports-api/revisions/31fb54ac414e?raw)