Methodology & data sources

Where the numbers come from

Standings and official qualification data are read from the English Wikipedia article 2026 ATP Finals through Wikipedia's public API, once every six hours. The article's wikitext is parsed rather than its rendered page, because the markup is more stable. Source: https://en.wikipedia.org/wiki/2026_ATP_Finals, parser wikipedia-wikitext-1.

ATP's own site refuses automated clients, and this project does not disguise itself or bypass any access control to get around that. Wikipedia is used instead because its content is openly licensed for reuse, its API is intended for programmatic access, and it carries a per-tournament points ledger plus a citation for every announced qualification.

Freshness — this page is not live

Wikipedia publishes these standings weekly and states the date they are current to. That stated date is what the page shows as the source date, and the page never claims to be more current than its source. Once the data is older than the configured threshold the page labels itself stale — last known good rather than implying freshness it does not have.

In-tournament projections (points after one more win, maximum reachable this week) are published only by the ATP. Rather than show columns of blanks, this site omits them.

What is computed here

• Provisional selection — ruleset atp-singles-2026.1 applied to the parsed table: the top seven by race rank, then up to two current-year Grand Slam champions ranked 8–20 in ranking order, then everyone else in rank order. The first eight unique players are the provisional selections and the ninth is the first alternate. Labelled “if the season ended now.”
• Margins — signed points distance to the qualification line, suppressed when the Grand Slam champion provision is active because no single contiguous line exists then.
• Movement — derived by comparing this snapshot with our own previous stored snapshot, not taken from any source.

Official versus provisional

Officially announced qualification is never inferred from points. Each announcement is ingested with the URL that announced it and stored alongside the snapshot; provisional selection is always computed and always labelled as such.

Integrity

Every fetch produces a candidate snapshot that must pass structural validation — the table must be present, ranks contiguous and unique, points non-increasing, and any announced qualifier must appear in the standings. A candidate that fails is rejected wholesale and can never replace the stored snapshot. Snapshots are append-only and immutable; corrections are new versions.

Licensing

Standings data originates from Wikipedia and is available under CC BY-SA 4.0; our reproduction of it is offered under the same licence. The site's source code is licensed separately (MIT) — code licensing and data licensing are not the same thing.

Season: 2026. Standings and qualification data come from Wikipedia, which publishes them weekly with a stated as-of date.