Nightly self-improvement · slate 2026-09-24
Generated 2026-09-25T05:00:00.007509-04:00 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
2026-09-24
status completed
Winners
8/12
Brier 0.2403
Learned or changed
0
of 5 stages
Blocked
0
wanted to act and could not
Release
v4.31.0
DRIFT
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 | skipped_open_rollback_monitor: |
| Auto-apply | nothing to learn | no_promotable_candidate |
| 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@COL | 6-5 | 12-8 | 0.440 | hit | Eduardo Rodriguez: 17 proj / 18 real outs; Tanner Gordon: 13 proj / 10 real outs | |
| CHW@KC | 7-5 | 9-1 | 0.455 | hit | David Sandlin: 15 proj / 18 real outs; Randy Dobnak: 17 proj / 4 real outs | |
| CIN@ATL | 5-7 | 7-6 | 0.560 | miss | Backed the wrong side with real conviction | Brady Singer: 12 proj / 9 real outs; Tyler Mahle: 18 proj / 15 real outs |
| CLE@BOS | 4-2 | 1-0 | 0.494 | hit | Daniel Espino: 9 proj / 3 real outs; Ranger Suarez: 17 proj / 15 real outs | |
| HOU@OAK | 7-5 | 7-5 | 0.440 | hit | Peter Lambert: 17 proj / 8 real outs; Mason Barnett: 11 proj / 18 real outs | |
| LAA@SEA | 4-5 | 6-4 | 0.560 | miss | Backed the wrong side with real conviction | Grayson Rodriguez: 15 proj / 15 real outs; Bryan Woo: 17 proj / 16 real outs |
| MIA@CHC | 4-5 | 1-2 | 0.519 | hit | Tyler Phillips: 15 proj / 12 real outs; Matthew Boyd: 17 proj / 21 real outs | |
| MIL@PHI | 5-4 | 5-1 | 0.461 | hit | Shane Drohan: 14 proj / 8 real outs; Andrew Painter: 15 proj / 18 real outs | |
| NYM@TEX | 5-4 | 1-3 | 0.479 | miss | Run environment far lower than projected (over-dispersed) | Zac Thornton: 17 proj / 15 real outs; Kumar Rocker: 14 proj / 15 real outs |
| SD@LAD | 4-5 | 4-2 | 0.510 | miss | Genuine toss-up that flipped | Nick Pivetta: 15 proj / 14 real outs; Tyler Glasnow: 18 proj / 19 real outs |
| STL@PIT | 4-5 | 1-2 | 0.528 | hit | Kyle Leahy: 15 proj / 6 real outs; Paul Skenes: 17 proj / 19 real outs | |
| TB@NYY | 2-4 | 4-6 | 0.531 | hit | Ian Seymour: 15 proj / 15 real outs; Cam Schlittler: 19 proj / 9 real outs |
Day over day
Rolling evidence window against the prior night. Associations, not causal proof.
| Metric | Prior | Current | Change | Result |
|---|---|---|---|---|
| brier_score | 0.2442 | 0.2454 | 0.0011 | regressed |
| accuracy | 0.5546 | 0.5630 | 0.0084 | improved |
| team_run_mae | 2.5168 | 2.4832 | -0.0336 | improved |
| total_run_mae | 3.3697 | 3.3529 | -0.0168 | improved |
| calibration_error | 0.1010 | 0.0535 | -0.0475 | improved |
| starter_er_mae | 1.7560 | 1.7323 | -0.0237 | improved |
| starter_outs_mae | 3.5720 | 3.5984 | 0.0264 | regressed |
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 | Status | Purpose | |---|---:|---| | `prediction review` — `/app/output/reports/prediction_review_20260924.json` | Loaded, 301,958 bytes | Official outcomes and scoring errors | | `forensic deep audit` — `/app/output/reports/PREDICTION_SYSTEM_DEEP_AUDIT_20260924.json` | Loaded, 196,012 bytes | Probability, provenance, starter, and decision contracts | | `continuous learning` — `/app/output/learning/continuous_learning_daily_20260924.json` | Loaded, 2,023,270 bytes | Stage quality, calibration, lineage, score distribution, starter comparisons | | `detailed report` — `/app/output/reports/detailed_20260924.html` | Loaded, 285,081 bytes | Full game and simulation rendering | | `ultra detailed player audit` — `/app/output/reports/20260924_ultra_detailed_player_audit.html` | Loaded, 25,971,500 bytes | Player-level prediction trace | | `structural audit` — `/app/output/reports/audit_report_20260924.html` | Loaded, 1,530,211 bytes | Invariant findings | | `integrity revision audit` — `/app/output/reports/audit_report_20260924_integrity_revision.html` | **Missing** | Post-Luna remediation proof for immutable revision artifacts | The missing integrity-revision audit is a provenance gap, but it did not prevent the supplied per-game structural checks from running. It does mean immutable revision/remediation claims cannot be independently confirmed from t