Nightly self-improvement · slate 2026-09-01
Generated 2026-09-02T17:45:50.200398-04:00 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
2026-09-01
status completed
Winners
7/15
Brier 0.2975
Learned or changed
0
of 5 stages
Blocked
0
wanted to act and could not
Release
v4.11.3
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) | nothing to learn | unavailable: |
| Auto-apply | nothing to learn | no_promotable_candidate |
| Code agent | nothing to learn | disabled_by_flag: code changes are made on the development host, never in the container. |
Games
| Game | Predicted | Actual | Home win prob | Result | Main driver | Starters |
|---|---|---|---|---|---|---|
| ATL@WAS | 7-5 | 5-9 | 0.468 | miss | Small edge on the losing side | AJ Smith-Shawver: 13 proj / 12 real outs; Jake Irvin: 15 proj / 12 real outs |
| BAL@COL | 7-4 | 2-4 | 0.254 | miss | Starting-pitcher projection was wrong | Kyle Bradish: 16 proj / 15 real outs; Gabriel Hughes: 13 proj / 14 real outs |
| CHW@HOU | 6-5 | 5-1 | 0.470 | hit | Sean Burke: 15 proj / 17 real outs; Ronel Blanco: 13 proj / 12 real outs | |
| DET@MIN | 4-3 | 2-15 | 0.470 | miss | Run environment far higher than projected (under-dispersed) | Troy Melton: 18 proj / 16 real outs; Andrew Morris: 14 proj / 3 real outs |
| MIA@KC | 3-5 | 6-3 | 0.646 | miss | Starting-pitcher projection was wrong | Tyler Phillips: 15 proj / 15 real outs; Randy Dobnak: 17 proj / 18 real outs |
| MIL@CHC | 6-8 | 9-4 | 0.699 | miss | Backed the wrong side with real conviction | Robert Gasser: 15 proj / 13 real outs; Matthew Boyd: 18 proj / 16 real outs |
| NYM@TB | 5-6 | 2-6 | 0.527 | hit | Sean Manaea: 18 proj / 17 real outs; Freddy Peralta: 16 proj / 18 real outs | |
| NYY@LAA | 6-4 | 7-3 | 0.460 | hit | Gerrit Cole: 18 proj / 16 real outs; Grayson Rodriguez: 15 proj / 18 real outs | |
| OAK@TEX | 4-5 | 5-8 | 0.549 | hit | Brady Basso: 13 proj / 7 real outs; MacKenzie Gore: 16 proj / 16 real outs | |
| PHI@ARI | 4-3 | 7-1 | 0.440 | hit | Jesús Luzardo: 20 proj / 20 real outs; Eduardo Rodriguez: 18 proj / 15 real outs | |
| SD@CIN | 7-4 | 3-4 | 0.400 | miss | Starting-pitcher projection was wrong | Randy Vásquez: 15 proj / 14 real outs; Nick Lodolo: 12 proj / 18 real outs |
| SEA@BOS | 3-5 | 9-6 | 0.530 | miss | Starting-pitcher projection was wrong | Bryan Woo: 17 proj / 18 real outs; Jedixson Paez: 9 proj / 11 real outs |
| SF@PIT | 5-6 | 12-13 | 0.545 | hit | Logan Webb: 18 proj / 6 real outs; Paul Skenes: 16 proj / 14 real outs | |
| STL@LAD | 5-7 | 13-8 | 0.580 | miss | Run environment far higher than projected (under-dispersed) | Michael McGreevy: 17 proj / 15 real outs; Eric Lauer: 16 proj / 9 real outs |
| TOR@CLE | 4-5 | 1-6 | 0.555 | hit | Spencer Miles: 15 proj / 3 real outs; Gavin Williams: 16 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 | 0.2408 | 0.2519 | 0.0110 | regressed |
| accuracy | 0.6167 | 0.5667 | -0.0500 | regressed |
| team_run_mae | 2.6500 | 2.7458 | 0.0958 | regressed |
| total_run_mae | 3.7333 | 3.8083 | 0.0750 | regressed |
| calibration_error | 0.1321 | 0.1128 | -0.0193 | improved |
| starter_er_mae | 1.8147 | 1.9483 | 0.1336 | regressed |
| starter_outs_mae | 3.0345 | 2.9267 | -0.1078 | 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
Luna (gpt-5.6-luna) post-mortem, status received. Full analysis.
# 1. Evidence receipt and gaps ## Sources received All seven manifest entries were checked against the supplied evidence: | Source | Path | Size | Status | Use | |---|---|---:|---|---| | Prediction review | `/app/output/reports/prediction_review_20260901.json` | 373,178 bytes | Loaded | Official outcomes and scoring errors | | Forensic deep audit | `/app/output/reports/PREDICTION_SYSTEM_DEEP_AUDIT_20260901.json` | 240,806 bytes | Loaded | Probability, provenance, starter, and decision contracts | | Continuous learning | `/app/output/learning/continuous_learning_daily_20260901.json` | 1,939,278 bytes | Loaded | Stage quality, calibration, lineage, score distribution, starter comparisons | | Detailed report | `/app/output/reports/detailed_20260901.html` | 355,722 bytes | Loaded | Full game and simulation rendering | | Ultra-detailed player audit | `/app/output/reports/20260901_ultra_detailed_player_audit.html` | 6,288,497 bytes | Loaded | Player-level prediction trace | | Structural audit | `/app/output/reports/audit_report_20260901.html` | 383,197 bytes | Loaded | Invariant findings | | Integrity revision audit | `/app/output/reports/audit_report_20260901_integrity_revision.html` | Not supplied | **Missing** | Post-remediation proof for immutable revision artifacts | The missing integrity-revision audit is the principal evidence gap. It does not, by itself, prove a prediction error, but it prevents a fully closed artifact-revision audit. ## Slate coverage The evidence re