10+ corners v2
← lab · AUC 0.505 (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? |
|---|
| home__team.conversion_l5 | 0.0044 | ↑ +0.007 | no |
| home__team.dominance_l5 | 0.0039 | ↑ +0.007 | no |
| home__team.shot_diff_avg_l5 | 0.0038 | ↑ +0.007 | no |
| away__team.corners_against_avg_l5 | 0.0033 | ↑ +0.044 | ⏳ yes |
| home__team.corners_against_avg_l5 | 0.0033 | ↑ +0.018 | no |
| away__team.elo_momentum_l5 | 0.0032 | ↑ +0.021 | no |
| match.drought_gap | 0.0030 | ↓ -0.010 | no |
| home__team.corners_volatility_l10 | 0.0023 | ↑ +0.011 | no |
| home__team.matches_since_blank | 0.0022 | ↓ -0.009 | no |
| home__team.cards_avg_l5 | 0.0021 | ↓ -0.008 | no |
| away__team.defensive_leak_l5 | 0.0011 | ↓ -0.013 | no |
| home__team.congestion_21d | 0.0010 | ↑ +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": "10+ corners v2",
"sport": "football",
"target": {
"op": ">=",
"value": 10,
"metric": "match.total_corners"
},
"features": "all"
}