sweep: match.second_half_goals >= 1
← lab · AUC 0.523 (no real pattern — honest result) · ran 7/5/2026
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? |
|---|
| home__team.corners_for_avg_l5 | 0.0002 | ↑ +0.019 | no |
| market.p_draw | 0.0001 | ↓ -0.064 | ⏳ yes |
| ratio__team.season_ppg | 0.0000 | ↑ +0.015 | no |
| ratio__team.rest_days | 0.0000 | ↑ +0.004 | no |
| away__team.conversion_l5 | 0.0000 | ↓ -0.007 | no |
| away__team.blown_lead_rate_l20 | 0.0000 | ↓ -0.010 | no |
| away__team.comeback_rate_l20 | 0.0000 | ↑ +0.018 | no |
| away__team.cards_avg_l5 | 0.0000 | ↓ -0.001 | no |
| away__team.corners_volatility_l10 | 0.0000 | ↑ +0.013 | no |
| away__team.defensive_leak_l5 | 0.0000 | ↓ -0.007 | no |
| away__team.dominance_l5 | 0.0000 | ↓ -0.002 | no |
| away__team.elo | 0.0000 | ↑ +0.005 | 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.second_half_goals >= 1",
"sport": "football",
"target": {
"op": ">=",
"value": 1,
"metric": "match.second_half_goals"
},
"features": "all"
}