Datasets:
match_id string | season string | minute int64 | second int64 | team_in_possession string | opponent string | home_away string | score_for int64 | score_against int64 | press_intensity_0_1 float64 | defensive_line_height_m int64 | turnover_zone string | turnover_forced int64 | transition_seconds int64 | xg_next_possession float64 | shots_next_120s int64 | goal_next_120s int64 | label_goal_next_120s int64 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
EPL_2025_001 | 2025-26 | 12 | 18 | ARS | NEW | home | 0 | 0 | 0.78 | 46 | mid | 1 | 7 | 0.18 | 1 | 0 | 0 |
EPL_2025_001 | 2025-26 | 23 | 5 | NEW | ARS | away | 0 | 0 | 0.62 | 43 | def | 0 | 11 | 0.05 | 0 | 0 | 0 |
EPL_2025_002 | 2025-26 | 31 | 44 | LIV | BHA | home | 0 | 0 | 0.84 | 49 | mid | 1 | 6 | 0.22 | 1 | 0 | 0 |
EPL_2025_002 | 2025-26 | 33 | 9 | LIV | BHA | home | 0 | 0 | 0.88 | 52 | mid | 1 | 5 | 0.38 | 2 | 1 | 1 |
EPL_2025_003 | 2025-26 | 54 | 27 | MCI | WHU | home | 1 | 0 | 0.71 | 55 | att | 0 | 9 | 0.11 | 1 | 0 | 0 |
EPL_2025_003 | 2025-26 | 61 | 52 | WHU | MCI | away | 0 | 1 | 0.76 | 44 | mid | 1 | 8 | 0.16 | 1 | 0 | 0 |
EPL_2025_004 | 2025-26 | 72 | 33 | TOT | EVE | home | 1 | 1 | 0.83 | 50 | mid | 1 | 6 | 0.29 | 1 | 0 | 0 |
EPL_2025_004 | 2025-26 | 74 | 1 | TOT | EVE | home | 1 | 1 | 0.9 | 54 | mid | 1 | 4 | 0.41 | 2 | 1 | 1 |
EPL_2025_005 | 2025-26 | 88 | 10 | CHE | AVL | home | 0 | 0 | 0.67 | 47 | def | 0 | 12 | 0.06 | 0 | 0 | 0 |
EPL_2025_005 | 2025-26 | 89 | 26 | AVL | CHE | away | 0 | 0 | 0.79 | 48 | mid | 1 | 7 | 0.24 | 1 | 0 | 0 |
EPL In-Play Quad Pre-Goal Collapse Window v0.1
What this dataset is
You test whether a model can detect an in-play collapse window before a goal.
Each row represents a live match-state snapshot.
The label asks
Will a goal occur in the next 120 seconds
Core quad coupling
Press intensity
Defensive line height
Turnover zone
xG per possession
Why this matters
Most football models explain goals after the fact.
This dataset tests pre-goal instability detection.
Intended use
You feed a row.
You output a prediction
0 no goal in next 120s
1 goal in next 120s
Columns
match_id
season
minute
second
team_in_possession
opponent
home_away
score_for
score_against
press_intensity_0_1
defensive_line_height_m
turnover_zone def mid att
turnover_forced 0 1
transition_seconds
xg_next_possession
shots_next_120s
goal_next_120s
label_goal_next_120s
Target label
label_goal_next_120s
Evaluation
Use scorer.py.
Input predictions can be
0 or 1
goal or no_goal
probability float where 0.5 is threshold
Files
data/train.csv
scorer.py
License
MIT
- Downloads last month
- 18