sweep: match.late_goal (bool)
← lab · AUC 0.501 (no real pattern — honest result) · ran 7/4/2026
What this is: Asks which pre-match factors drive one specific outcome, using a walk-forward model and permutation importance.
| Factor | Importance | Direction | Survives all eras? |
|---|
| away__team.congestion_21d | 0.0000 | ↑ +0.014 | no |
| away__team.elo_momentum_l5 | 0.0000 | — | no |
| away__team.form_points_l5 | 0.0000 | — | no |
| away__team.goal_diff_avg_l5 | 0.0000 | — | no |
| away__team.goals_for_avg_l5 | 0.0000 | — | no |
| away__team.goals_against_avg_l5 | 0.0000 | — | no |
| away__team.matches_since_blank | 0.0000 | ↑ +0.183 | no |
| away__team.matches_since_clean_sheet | 0.0000 | ↑ +0.110 | no |
| home__team.elo | 0.0000 | ↑ +0.055 | no |
| away__team.matches_since_win | 0.0000 | ↓ -0.042 | no |
| away__team.rest_days | 0.0000 | ↓ -0.015 | no |
| away__team.state_index | 0.0000 | ↑ +0.045 | no |
Reading the columnswhat each number actually means
| AUC | predictability: 0.50 = coin flip, ~0.70 = ceiling for sports |
| Importance | how much the model leans on this factor (permutation importance) |
| Direction | sign of the raw correlation with the outcome |
| Survives all eras | effect points the same way in every historical era |
Spec · the reproducible recipe
{
"name": "sweep: match.late_goal (bool)",
"sport": "football",
"target": {
"metric": "match.late_goal"
},
"features": "all"
}