peggy44/PremierLeague25-26

English Premier League — 2025/26 (SkillCorner)

Broadcast-tracking and event data for English Premier League, season 2025/26, produced by SkillCorner. Distributed here in the raw layout used by the Deep Learning & AI in Sport course (repo).

Directory layout

PremierLeague25-26/
├── metadata/                # league-level metadata
│   ├── matches.json                         # full fixtures list
│   └── available_dynamic_event_match_ids.csv
├── matches/                 # per-match summary (.json, one per match_id)
├── dynamic_events/          # per-match dynamic events (.json, one per match_id)
├── tracking/                # per-match tracking frames (.json, one per match_id)
└── physical/                # aggregated physical KPIs
    └── physical.json

Loading

from huggingface_hub import snapshot_download
from pathlib import Path
import pandas as pd, json

# Download the whole dataset (resumable):
local_dir = snapshot_download(
    repo_id="peggy44/PremierLeague25-26",
    repo_type="dataset",
    local_dir="data/PremierLeague25-26",
)
local_dir = Path(local_dir)

# Example — read one match's tracking + events:
match_id = next(local_dir.glob("tracking/*.json")).stem
with open(local_dir / f"tracking/{match_id}.json") as f:
    tracking = json.load(f)

To download only a subset (e.g. metadata + one match) pass allow_patterns= to snapshot_download.

Data types

Folder / file	Format	One row / file per
matches.parquet	parquet	match
players.parquet	parquet	player
teams.parquet	parquet	team
metadata/matches.json	json	competition (all fixtures)
dynamic/ or dynamic_events/	parquet / json	match (event stream)
tracking/	json	match (10 Hz tracking)
meta/	json	match (lineup, pitch, coords)
physical/	json	match / competition
freeze/	parquet	match (freeze frames)

Only folders present in the layout above ship with this dataset.

Licensing & use

• Distributed under CC BY-NC 4.0 for research and educational use.
• Original data © SkillCorner. Commercial use requires a separate SkillCorner licence.
• Personally identifying data is limited to publicly known player/team identifiers already published by the competition organiser.

Citation

If you use this dataset in academic work, please cite:

@misc{rahimian_dlaia_PremierLeague25_26_2026,
  author       = {Rahimian, Pegah},
  title        = {English Premier League — 2025/26 (SkillCorner)},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/peggy44/PremierLeague25-26}},
}

Contact

Curated by Pegah Rahimian (rahimian.pegah@gmail.com) for the Deep Learning & AI in Sport course cohort (September 2026).