CoolFace
Datasetpublic

Emlembow/march-madness-2026-data

NCAA March Madness Dataset (2011-2026) Player-level basketball statistics for NCAA tournament teams, plus game-level tournament matchups. Player Statistics (training_dataset.parquet) Each row is one player-season for a tournament team. Key columns: Column Description player_name, team, season_year Identity pts_per_game, trb_per_game, ast_per_game Box score stats fg_pct, three_p_pct, efg_pct, ts_pct Shooting percentages per, bpm, ws Advanced… See the full description on the dataset page: https://huggingface.co/datasets/Emlembow/march-madness-2026-data.

sourceHugging Facemitupdated 6mo agoView on Hugging Face
0likes90downloads
Dataset Card

NCAA March Madness Dataset (2011-2026)

Player-level basketball statistics for NCAA tournament teams, plus game-level tournament matchups.

Dataset Description

  • —training_dataset.parquet: 13,308 player-season records across 15 seasons (2011-2026), 55 columns
  • —tournament_matchups.jsonl: 917 tournament game records (881 completed + 36 for 2026 bracket)
  • —bracket_2026.json: 2026 tournament bracket structure with seeds and first-round matchups

Player Statistics (training_dataset.parquet)

Each row is one player-season for a tournament team. Key columns:

ColumnDescription
playername, team, seasonyearIdentity
ptspergame, trbpergame, astpergameBox score stats
fgpct, threeppct, efgpct, ts_pctShooting percentages
per, bpm, wsAdvanced metrics
teamwins, teamlosses, teamsos, teamsrsTeam context
tournament_seed1-16 tournament seed
includeintrainingQuality filter flag

Tournament Matchups (tournament_matchups.jsonl)

One JSON object per game with: season_year, round, region, team slugs/names/seeds, winner, scores.

Data Sources

Usage

python
import pandas as pd
df = pd.read_parquet("training_dataset.parquet")
print(df.shape)  # (13308, 55)
print(df["season_year"].unique())  # 2011-2026

Demo

See the interactive bracket predictor: March Madness 2026 Space