Run Outcome Regression (per-fight)
Generated: 2026-07-10 16:27 UTC
Rows: 1506 fights from 720 runs
Filter: post-shop only (F2+boss)
Pooled fight win rate: 57.2%
McFadden pseudo-R²: 0.3669
Accuracy @0.5: 0.7935
Top multivariate coefficients (standardized)
| Feature |
Coef |
Odds ratio |
p-value |
| enemy_Mom_&_Pop_Owners |
-1.381 |
0.251 |
0.040866 |
| enemy_Mr._&_Mrs._Watts_Owners |
-1.221 |
0.295 |
0.007557 |
| hp_pct |
+1.134 |
3.108 |
0.0 |
| enemy_Mom_&_Pop_Founders |
-1.105 |
0.331 |
0.043256 |
| archetype_bleed |
-0.877 |
0.416 |
0.041377 |
| archetype_balanced_sustain |
-0.737 |
0.478 |
0.022425 |
| enemy_Nuclear_Physicist |
-0.721 |
0.486 |
0.182179 |
| sector_energy |
+0.652 |
1.919 |
0.311914 |
| enemy_other |
+0.642 |
1.900 |
0.061847 |
| archetype_aggro_surge |
+0.603 |
1.827 |
0.268808 |
| archetype_hype_aggro |
-0.592 |
0.553 |
0.255782 |
| archetype_balanced_scaling |
+0.558 |
1.748 |
0.278724 |
| enemy_Pipeline_Foreman |
-0.557 |
0.573 |
0.316094 |
| enemy_hp |
+0.472 |
1.603 |
0.016195 |
| deck_size |
-0.417 |
0.659 |
0.004417 |
| enemy_Nurse_Sister |
+0.381 |
1.464 |
0.547211 |
| archetype_control_defensive |
-0.327 |
0.721 |
0.523123 |
Univariate screen (lowest p-values)
| Feature |
Direction |
Odds ratio |
p-value |
| tier_index |
hurts |
0.7734 |
1e-06 |
| enemy_hp |
hurts |
0.7572 |
1e-06 |
| sector_ai |
helps |
1.2836 |
2e-06 |
| shop_totems_bought |
helps |
1.2405 |
3.6e-05 |
| sector_energy |
helps |
1.2401 |
4.5e-05 |
| totem_count |
hurts |
0.9116 |
0.074849 |
| shop_removals |
hurts |
0.9423 |
0.249987 |
| shop_spend_frac |
helps |
1.0543 |
0.308775 |
| tier_mega |
hurts |
0.9887 |
0.826296 |
| hp_pct |
helps |
1.4373 |
0.0 |
| deck_size |
helps |
1.3421 |
0.0 |
| log_capital |
helps |
1.3611 |
0.0 |
| battles_won |
helps |
1.3395 |
0.0 |
| stage_index |
helps |
1.351 |
0.0 |
| hp_lost_prev_fight |
helps |
1.3568 |
0.0 |
| shop_cards_bought |
helps |
1.3594 |
0.0 |
| stage_regular_2 |
helps |
1.351 |
0.0 |
| stage_boss |
helps |
1.351 |
0.0 |
| tier_mid |
helps |
1.3388 |
0.0 |
Marginal win rates (sanity check)
hp_pct_quartile
| Bucket |
n |
Win% |
| 2 |
540 |
68.2% |
| 3 |
398 |
87.2% |
| 1 |
385 |
35.1% |
| 0 |
183 |
6.0% |
deck_size
| Bucket |
n |
Win% |
| 14 |
353 |
72.2% |
| 15 |
310 |
58.4% |
| 13 |
230 |
68.7% |
| 16 |
209 |
37.3% |
| 17 |
87 |
37.9% |
| 12 |
68 |
76.5% |
| 18 |
65 |
46.2% |
| 19 |
63 |
38.1% |
| 20 |
53 |
35.9% |
| 21 |
29 |
41.4% |
| 11 |
10 |
90.0% |
| 22 |
9 |
33.3% |
totem_count
| Bucket |
n |
Win% |
| 2 |
946 |
62.5% |
| 1 |
231 |
56.3% |
| 4 |
172 |
44.2% |
| 3 |
86 |
40.7% |
| 0 |
46 |
32.6% |
| 5 |
13 |
46.2% |
| 6 |
12 |
66.7% |
stage
| Bucket |
n |
Win% |
| regular_2 |
906 |
66.2% |
| boss |
600 |
43.5% |
tier
| Bucket |
n |
Win% |
| Family |
1192 |
60.8% |
| MidTier |
285 |
42.1% |
| MegaCorp |
29 |
55.2% |
shop_cards_bought
| Bucket |
n |
Win% |
| 3 |
493 |
67.8% |
| 4 |
404 |
57.9% |
| 5 |
256 |
37.5% |
| 2 |
216 |
65.3% |
| 6 |
88 |
23.9% |
| 1 |
47 |
74.5% |
| 7 |
2 |
0.0% |
Calibration hints
- Weight/feature
enemy_Mom_&_Pop_Owners: decrease (OR=0.2513, p=0.040866)
- Weight/feature
enemy_Mr._&_Mrs._Watts_Owners: decrease (OR=0.295, p=0.007557)
- Weight/feature
enemy_Mom_&_Pop_Founders: decrease (OR=0.3312, p=0.043256)
- Weight/feature
archetype_bleed: decrease (OR=0.4161, p=0.041377)
- Weight/feature
archetype_balanced_sustain: decrease (OR=0.4784, p=0.022425)
- Weight/feature
enemy_hp: increase (OR=1.6031, p=0.016195)
- Univariate: tier_index hurts win rate (OR=0.7734, p=1e-06)
- Univariate: enemy_hp hurts win rate (OR=0.7572, p=1e-06)
- Univariate: sector_ai helps win rate (OR=1.2836, p=2e-06)
- Univariate: shop_totems_bought helps win rate (OR=1.2405, p=3.6e-05)
- Univariate: sector_energy helps win rate (OR=1.2401, p=4.5e-05)
- Marginal deck size: best win% at size 12 (76.5%), worst at 20 (35.9%)
Reproducing
py -3 scripts/regress_run_outcomes.py