3+ goals v2
← lab · AUC 0.524 (no real pattern — honest result) · ran 7/3/2026
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.season_ppg | 0.0098 | ↑ +0.013 | no |
| away__team.shots_for_avg_l5 | 0.0065 | ↑ +0.054 | ⏳ yes |
| home__team.season_ppg | 0.0054 | ↑ +0.051 | ⏳ yes |
| home__team.conversion_l5 | 0.0052 | ↓ -0.002 | no |
| home__team.corners_for_avg_l5 | 0.0049 | ↑ +0.044 | ⏳ yes |
| home__team.shots_for_avg_l5 | 0.0047 | ↑ +0.069 | ⏳ yes |
| match.rainfall_mm | 0.0047 | ↑ +0.016 | no |
| away__team.goals_for_avg_l5 | 0.0046 | ↑ +0.026 | ⏳ yes |
| away__team.elo | 0.0042 | ↑ +0.022 | no |
| home__team.goals_volatility_l10 | 0.0040 | ↑ +0.036 | ⏳ yes |
| away__team.goal_diff_avg_l5 | 0.0032 | ↑ +0.014 | no |
| match.elo_gap | 0.0029 | ↑ +0.040 | ⏳ yes |
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": "3+ goals v2",
"sport": "football",
"target": {
"op": ">=",
"value": 3,
"metric": "match.total_goals"
},
"features": "all"
}