tennis_atp on GitHub returns 404: where Jeff Sackmann's ATP and WTA data is now

The options

• JeffSackmann/tennis_atp and tennis_wta (the original repositories)

  ATP and WTA match results, rankings and player files as CSV, with serve statistics where available, under CC BY-NC-SA 4.0.

  HTTP 404 from github.com and from api.github.com on 2026-10-14, for both repositories. The Internet Archive lists a snapshot of the tennis_atp page from 2026-03-14 and of the tennis_wta page from 2025-11-16. We did not check what those snapshots contain.

• JeffSackmann/tennis_MatchChartingProject

  Crowdsourced shot-by-shot data for professional matches, as point-level and aggregate CSVs.

  Public. Last pushed 2026-09-18; 446 stars. Its README licenses the data under CC BY-NC-SA 4.0 and says: attribution is required, non-commercial use only. GitHub's licence detector shows none for the repository. It holds a sample of charted matches, not every match.

• Tennis Abstract

  Jeff Sackmann's statistics site, which links to his GitHub profile.

  HTTP 200 on 2026-10-14. The GitHub profile it links to listed one public repository, the Match Charting Project.

• Kadantte/tennis_atp (fork)

  A fork of the ATP repository: atp_matches files from 1968 to 2026, the players file and the ranking files, with the original README and its CC BY-NC-SA notice.

  8 stars. Last push 2026-06-08, with a commit message reading "thru 8 jun 2026". A third-party copy, not maintained by Sackmann.

• Aneeshers/tennis-sackmann-archive (mirror)

  A self-described archival mirror of the ATP, WTA and Grand Slam point-by-point datasets, data only, with each original README kept. Also published as a Hugging Face dataset.

  15 stars; created 2026-06-25. Its README says the ATP and WTA snapshots come from upstream commits made in June 2026 and the point-by-point snapshot from October 2024. ATP and WTA match files run 1968 to 2026. The Hugging Face copy showed 21,352 downloads and a cc-by-nc-sa-4.0 tag on 2026-10-14.

• WTA forks with 2026 results

  VictorSquidWei/tennis_wta and mikecristancho/tennis_wta, forks of the WTA repository that include wta_matches_2026.csv.

  0 stars each. Last push 2026-05-06 for both. Of the 100 forks named tennis_wta that GitHub search returned, these two were the only ones with a 2026 match file.

• Older forks

  GitHub search lists at least 99 forks named tennis_atp and 100 named tennis_wta (the search stops at 100). Most froze on the day they were forked. Of the 99 ATP forks, five had an atp_matches_2026.csv: Kadantte (8 stars, pushed 2026-06-08) and four with no stars, William1515 (pushed 2026-05-25), NicolaiSchmid, srcde and seanalamojr (all 2026-06-08).

  sleepomeno/tennis_atp has 22 stars and ends at 2015. tastethewhip/tennis_atp and vishaan2010-dotcom/tennis_atp end at 2024. Look at the newest atp_matches file in a fork before you trust it.

• Kaggle

  Datasets that name his data: SeraphicFrost's GrandSlams(1968-2024) -- JeffSackMann.

  Last updated 2025-08-07; licence listed as Unknown. Older Kaggle ATP and WTA match sets that list CC BY-NC-SA 4.0, such as GMAdevs' (last updated 2017), do not name a source in Kaggle's listing, so we cannot say they are copies.

Side by side

Example call

One tournament-final result, with the winner, the score and every set. This is the shape of a row that tennis_atp kept as winner_name, loser_name, score, surface and round. Here winner_id says who won, and the sets array carries the games and tiebreaks.

curl -X POST https://mcp.acesapi.com/v1/get_matches \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
  "season": 2025,
  "round": "Final",
  "limit": 1
}'

{"data": {"tour": "ATP", "matches": [{"match_id": "152433", "tournament_id": "971-2025", "tournament": "Bank of China Hong Kong Tennis Open", "season": 2025, "surface": "hard", "indoor": false, "match_date": "2025-01-05", "draw": "main", "round": "Final", "player1_id": "1035", "player1": "Kei Nishikori", "player1_country": "JPN", "player2_id": "3308", "player2": "Alexandre Muller", "player2_country": "FRA", "winner_id": "3308", "status": "final", "outcome": "completed", "score": "2-6 6-1 6-3", "sets1": 1, "sets2": 2, "games1": 10, "games2": 14, "sets": [{"set_no": 1, "games1": 6, "games2": 2, "tiebreak1": null, "tiebreak2": null, "winner": 1}, {"set_no": 2, "games1": 1, "games2": 6, "tiebreak1": null, "tiebreak2": null, "winner": 2}, {"set_no": 3, "games1": 3, "games2": 6, "tiebreak1": null, "tiebreak2": null, "winner": 2}]}], "count": 1, "next_offset": 1}}

The match is cut to the fields a CSV row would hold (the full row also has seeds, court, start time and round order). Responses are wrapped in a data object. get_matches costs 2 credits per call.

Pull it into your own code

A season of ATP results in pandas. get_matches returns up to 500 matches a call and next_offset says where the next call starts, so a whole season is a short loop. Use league "wta" for the women's tour. The key comes from your environment.

import os
import requests
import pandas as pd

URL = "https://mcp.acesapi.com/v1/get_matches"
HEADERS = {"x-api-key": os.environ["ACES_API_KEY"]}


def season(year, league="tennis"):  # "tennis" is the ATP, "wta" is the WTA
    rows, offset = [], 0
    while offset is not None:
        resp = requests.post(
            URL,
            headers=HEADERS,
            json={"season": year, "league": league, "limit": 500, "offset": offset},
            timeout=60,
        )
        resp.raise_for_status()
        data = resp.json()["data"]
        rows += data["matches"]
        offset = data["next_offset"]
    return pd.DataFrame(rows)


df = season(2025)
df = df[(df["draw"] == "main") & (df["status"] == "final")].copy()
df["winner"] = df["player1"].where(df["winner_id"] == df["player1_id"], df["player2"])
df["loser"] = df["player2"].where(df["winner_id"] == df["player1_id"], df["player1"])

print(len(df), "main-draw matches")
print(df["surface"].value_counts().to_dict())
print(df[["match_date", "tournament", "surface", "round", "winner", "loser", "score"]].head(5).to_string(index=False))

Run on 2026-10-14, this made 8 calls (16 credits) and printed 2,659 main-draw ATP matches for 2025: 1,596 on hard, 766 on clay and 297 on grass. The 8 calls also returned qualifying matches, which the filter drops. A season is the tournament's year, so the Brisbane International's first matches are dated 2024-12-29. At 16 credits a tour-season, the 500 free credits cover about 30. Needs requests and pandas (pip install requests pandas).

Use Jeff Sackmann's tennis_atp and tennis_wta files instead if

• You need serve statistics, minutes played or break points. AcesAPI has none of them.
• You need results before 2007, challenger or futures matches, or rankings before 2007.
• Your work is non-commercial, a fork or mirror covers the years you need, and you pin the copy by commit so the notebook keeps running.
• You need shot-by-shot data on charted matches: the Match Charting Project is still public, for non-commercial use.
• You need doubles.

Questions

Sources

• GitHub API: JeffSackmann/tennis_atp (404 on 2026-10-14)
• GitHub API: JeffSackmann/tennis_wta (404 on 2026-10-14)
• GitHub API: Jeff Sackmann's profile (public_repos: 1)
• Match Charting Project README and licence
• Tennis Abstract
• Internet Archive snapshot of the tennis_atp page, 2026-03-14
• Internet Archive snapshot of the tennis_wta page, 2025-11-16
• Kadantte/tennis_atp (fork, README with the original notice and file description)
• Aneeshers/tennis-sackmann-archive (mirror)
• Hugging Face copy of that mirror
• CC BY-NC-SA 4.0 licence summary

Checked October 14, 2026.

See the data