home wins v2 (feature factory)
← lab · AUC 0.677 (real signal) · 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? |
|---|
| match.elo_gap | 0.0505 | ↑ +0.356 | ⏳ yes |
| home__team.shots_for_avg_l5 | 0.0046 | ↑ +0.179 | ⏳ yes |
| match.temp_c | 0.0030 | ↓ -0.009 | no |
| away__team.elo | 0.0027 | ↓ -0.241 | ⏳ yes |
| away__team.corners_for_avg_l5 | 0.0017 | ↓ -0.123 | ⏳ yes |
| away__team.venue_ppg_l5 | 0.0016 | ↓ -0.160 | ⏳ yes |
| home__team.corners_for_avg_l5 | 0.0007 | ↑ +0.140 | ⏳ yes |
| away__team.matches_since_win | 0.0002 | ↑ +0.098 | ⏳ yes |
| away__team.corner_diff_avg_l5 | 0.0001 | ↓ -0.158 | ⏳ yes |
| away__team.defensive_leak_l5 | -0.0001 | ↑ +0.035 | ⏳ yes |
| match.rained | -0.0002 | ↓ -0.001 | no |
| home__team.elo | -0.0002 | ↑ +0.260 | ⏳ 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": "home wins v2 (feature factory)",
"sport": "football",
"target": {
"equals": "H",
"metric": "match.result"
},
"features": "all"
}