Nightly self-improvement · slate 2026-08-28
Generated 2026-08-29T02:00:00.006918-04:00 ET from official finals (mlb-api). Every learning stage is listed with what it did. "Blocked" means it wanted to act and a safety rule stopped it; "nothing to learn" means it had no eligible evidence.
Slate
2026-08-28
status completed
Winners
7/15
Brier 0.2330
Learned or changed
0
of 5 stages
Blocked
1
wanted to act and could not
Release
v4.8.2
consistent
What each learning stage did
| Stage | Outcome | Detail |
|---|---|---|
| KPI settlement | failed | did not run |
| Bullpen usage model | nothing to learn | No pregame bullpen forecasts to settle. Refit: not attempted. |
| Adaptive parameter proposal (Luna) | blocked | quarantined_invalid_artifact: slate quarantined from new parameter proposals until critical/error artifact findings are corrected |
| Auto-apply | nothing to learn | no_promotable_candidate; failed gates ['minimum_paired_games', 'minimum_distinct_slates'] |
| Code agent | nothing to learn | disabled_by_config: code changes are made on the development host, never in the container. |
Games
| Game | Predicted | Actual | Home win prob | Result | Main driver | Starters |
|---|---|---|---|---|---|---|
| ARI@SF | 5-6 | 10-6 | 0.534 | miss | Starting-pitcher projection was wrong | Kohl Drake: 13 proj / 12 real outs; Blade Tidwell: 15 proj / 15 real outs |
| BAL@OAK | 6-5 | 4-3 | 0.452 | hit | Brandon Young: 16 proj / 21 real outs; Jacob Lopez: 15 proj / 15 real outs | |
| BOS@NYY | 4-6 | 0-1 | 0.530 | hit | Patrick Sandoval: 15 proj / 16 real outs; Cam Schlittler: 18 proj / 17 real outs | |
| CHW@MIN | 5-6 | 7-4 | 0.527 | miss | Small edge on the losing side | Luis Castillo: 15 proj / 12 real outs; Dean Kremer: 15 proj / 12 real outs |
| CIN@CHC | 5-7 | 10-8 | 0.560 | miss | Run environment far higher than projected (under-dispersed) | Rhett Lowder: 15 proj / 15 real outs; David Peterson: 14 proj / 15 real outs |
| COL@ATL | 3-8 | 4-6 | 0.848 | hit | Tomoyuki Sugano: 16 proj / 12 real outs; Grant Holmes: 15 proj / 15 real outs | |
| HOU@NYM | 3-4 | 3-1 | 0.549 | miss | Several errors of similar size | Hunter Brown: 16 proj / 20 real outs; Christian Scott: 17 proj / 16 real outs |
| KC@CLE | 4-5 | 3-0 | 0.574 | miss | Run environment far lower than projected (over-dispersed) | Michael Wacha: 18 proj / 24 real outs; Tanner Bibee: 18 proj / 18 real outs |
| LAD@DET | 4-1 | 2-1 | 0.335 | hit | Tarik Skubal: 18 proj / 18 real outs; Drew Anderson: 13 proj / 14 real outs | |
| MIA@WAS | 5-4 | 2-9 | 0.445 | miss | Starting-pitcher projection was wrong | Eury Pérez: 17 proj / 9 real outs; Jackson Kent: 11 proj / 21 real outs |
| PHI@LAA | 4-5 | 5-3 | 0.566 | miss | Backed the wrong side with real conviction | Andrew Painter: 15 proj / 18 real outs; Reid Detmers: 17 proj / 18 real outs |
| PIT@STL | 5-3 | 1-4 | 0.400 | miss | Backed the wrong side with real conviction | Jared Jones: 15 proj / 14 real outs; Quinn Mathews: 13 proj / 17 real outs |
| SD@TB | 4-5 | 4-9 | 0.557 | hit | Casey Mize: 15 proj / 15 real outs; Shane McClanahan: 16 proj / 17 real outs | |
| SEA@TOR | 2-4 | 1-8 | 0.530 | hit | Emerson Hancock: 17 proj / 15 real outs; Dylan Cease: 18 proj / 18 real outs | |
| TEX@MIL | 2-6 | 1-6 | 0.848 | hit | Cody Bradford: 15 proj / 17 real outs; Logan Henderson: 18 proj / 21 real outs |
Day over day
Rolling evidence window against the prior night. Associations, not causal proof.
| Metric | Prior | Current | Change | Result |
|---|---|---|---|---|
| brier_score | 0.2422 | 0.2418 | -0.0004 | improved |
| accuracy | 0.5980 | 0.5833 | -0.0147 | regressed |
| team_run_mae | 2.4363 | 2.4306 | -0.0057 | improved |
| total_run_mae | 3.7745 | 3.7130 | -0.0615 | improved |
| calibration_error | 0.1080 | 0.1004 | -0.0076 | improved |
| starter_er_mae | 1.8154 | 1.7854 | -0.0300 | improved |
| starter_outs_mae | 3.2154 | 3.1073 | -0.1081 | improved |
Calibration produced for the next slate
| Metric | Scale used | Measured | Rows | Status |
|---|---|---|---|---|
| No calibration produced. | ||||
Luna KPI candidates and their promotion gate
Promotion needs 400+ prospective rows over 5+ slates after the proposal, with a gain interval above zero. Only verified pregame forecasts count; local replays never do.
| KPI | Status | Target | Proposed | Rows | Slates | Gain interval | Progress |
|---|---|---|---|---|---|---|---|
| luna_pitcher_release_speed_vs_righthanded_starters | shadow | SO | 20260923 | 330/400 | 5/5 | [-0.0007, 0.0007] | |
| luna_pitcher_barrel_rate_vs_rhh_righthanded_starters | shadow | HR | 20260923 | 330/400 | 5/5 | [-0.0004, 0.0049] | |
| luna_batter_contact_rate_vs_breaking_righthanded_starters | shadow | H | 20260923 | 330/400 | 5/5 | [-0.0045, 0.0044] | |
| luna_batter_contact_rate_vs_breaking_left_handed_starters_for_strikeouts | shadow | SO | 20260924 | 122/400 | 4/5 | [-0.0027, 0.0021] | |
| luna_pitcher_contact_xwoba_allowed_vs_righthanded_starters_for_strikeouts | shadow | SO | 20260924 | 122/400 | 4/5 | [-0.0001, 0.0000] | |
| luna_pitcher_whiff_rate_breaking_vs_left_handed_hitters_for_hits | shadow | H | 20260924 | 122/400 | 4/5 | [-0.0056, -0.0011] | |
| luna_pitcher_home_run_rate_vs_righthanded_starters | shadow | HR | 20260925 | 122/400 | 4/5 | [-0.0010, 0.0056] | |
| luna_batter_ground_ball_rate_vs_fastball_vs_lhp_for_home_runs | shadow | HR | 20260926 | 86/400 | 3/5 | [-0.0001, -0.0000] | |
| luna_pitcher_whiff_rate_vs_right_handed_starters_for_batter_strikeouts | shadow | SO | 20260929 | 51/400 | 2/5 | [–, –] | |
| luna_batter_whiff_rate_vs_breaking_right_handed_starters_for_hits | shadow | H | 20260929 | 51/400 | 2/5 | [–, –] | |
| luna_batter_walk_rate_vs_righthanded_pitchers_for_home_runs | retired | HR | 20260924 | 0/400 | 0/5 | [–, –] |
Operator actions the cycle asked for (0)
Changes to code or a model recipe are never made by the nightly cycle. These are the measured findings that need one.
| Priority | Owner | Finding | Action |
|---|---|---|---|
| No operator actions requested. | |||
Luna post-mortem
No Luna post-mortem this night.