corners w/ granular
← lab · AUC 0.524 (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? |
|---|
| match.style_clash_corners | 0.0042 | ↑ +0.070 | ⏳ yes |
| market.p_over25 | 0.0022 | ↑ +0.084 | ⏳ yes |
| x__match.temp_c__match.tempo | 0.0018 | ↓ -0.014 | no |
| ratio__team.matches_since_blank | 0.0018 | ↑ +0.017 | no |
| home__team.corners_for_avg_l5 | 0.0016 | ↑ +0.051 | ⏳ yes |
| diff__team.ht_lead_rate_l20 | 0.0015 | ↑ +0.017 | no |
| match.tempo | 0.0013 | ↑ +0.001 | no |
| ratio__team.matches_since_win | 0.0011 | ↓ -0.005 | no |
| diff__team.defensive_leak_l5 | 0.0011 | ↓ -0.004 | no |
| away__team.defensive_leak_l5 | 0.0011 | ↓ -0.000 | no |
| ratio__team.ht_lead_rate_l20 | 0.0011 | ↑ +0.004 | no |
| diff__team.pass_acc_avg_l5 | 0.0010 | ↑ +0.072 | ⏳ 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": "corners w/ granular",
"sport": "football",
"target": {
"op": ">=",
"value": 10,
"metric": "match.total_corners"
},
"features": "all"
}