Nightly self-improvement · slate 2026-09-06

Generated 2026-09-07T02:00:00.005227-04:00 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

2026-09-06

status completed

Winners

9/15

Brier 0.2455

Learned or changed

of 5 stages

Blocked

wanted to act and could not

Release

v4.25.0

consistent

Audit Detailed Deep Slate Review Luna Post-mortem Daily Learning Adaptive Status

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	candidate_registered:
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@HOU	4-3	3-2	0.463	hit		Eduardo Rodriguez: 18 proj / 18 real outs; Peter Lambert: 16 proj / 21 real outs
ATL@PHI	3-4	5-4	0.551	miss	Backed the wrong side with real conviction	Tyler Mahle: 17 proj / 18 real outs; Aaron Nola: 16 proj / 21 real outs
BOS@BAL	4-3	3-1	0.451	hit		Payton Tolle: 18 proj / 18 real outs; Kyle Bradish: 16 proj / 12 real outs
CHC@MIA	6-4	3-10	0.420	miss	Starting-pitcher projection was wrong	Clay Holmes: 18 proj / 13 real outs; Tyler Phillips: 15 proj / 15 real outs
DET@CLE	3-4	2-3	0.520	hit		Jackson Jobe: 15 proj / 15 real outs; Gavin Williams: 16 proj / 18 real outs
LAA@PIT	3-5	0-1	0.572	hit		Walbert Ureña: 17 proj / 24 real outs; Paul Skenes: 17 proj / 16 real outs
MIL@CIN	7-5	8-12	0.470	miss	Run environment far higher than projected (under-dispersed)	Kyle Harrison: 15 proj / 11 real outs; Brady Singer: 15 proj / 15 real outs
MIN@CHW	4-6	1-10	0.530	hit		Bailey Ober: 16 proj / 13 real outs; Bryan Hudson: 15 proj / 5 real outs
NYY@SD	4-3	3-4	0.464	miss	Several errors of similar size	Gerrit Cole: 19 proj / 21 real outs; Michael King: 18 proj / 18 real outs
OAK@SEA	6-7	0-2	0.544	hit		Gage Jump: 14 proj / 12 real outs; Bryan Woo: 17 proj / 24 real outs
SF@NYM	4-6	2-4	0.516	hit		Cesar Perdomo: 12 proj / 18 real outs; Christian Scott: 15 proj / 14 real outs
STL@COL	5-4	10-8	0.441	hit		Kyle Leahy: 16 proj / 8 real outs; Tanner Gordon: 15 proj / 14 real outs
TB@TEX	5-4	6-8	0.473	miss	Run environment far higher than projected (under-dispersed)	Ian Seymour: 16 proj / 9 real outs; MacKenzie Gore: 16 proj / 9 real outs
TOR@KC	5-4	1-6	0.463	miss	Small edge on the losing side	Spencer Arrighetti: 15 proj / 14 real outs; Randy Dobnak: 18 proj / 18 real outs
WAS@LAD	4-5	5-7	0.530	hit		Andrew Alvarez: 15 proj / 13 real outs; Justin Wrobleski: 16 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	0.2459	0.2509	0.0050	regressed
accuracy	0.5833	0.5583	-0.0250	regressed
team_run_mae	2.5917	2.6083	0.0166	regressed
total_run_mae	3.6167	3.6500	0.0333	regressed
calibration_error	0.0868	0.0640	-0.0228	improved
starter_er_mae	1.9426	1.9795	0.0369	regressed
starter_outs_mae	2.8811	3.0123	0.1312	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
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

Six of seven manifest entries were loaded:

1. **Prediction review** — `/app/output/reports/prediction_review_20260906.json`, 377,236 bytes; official outcomes and scoring errors.
2. **Forensic deep audit** — `/app/output/reports/PREDICTION_SYSTEM_DEEP_AUDIT_20260906.json`, 240,848 bytes; probability, provenance, starter, and decision contracts.
3. **Continuous learning** — `/app/output/learning/continuous_learning_daily_20260906.json`, 2,489,405 bytes; pipeline quality, calibration, lineage, score distribution, and starter comparisons.
4. **Detailed report** — `/app/output/reports/detailed_20260906.html`, 349,460 bytes.
5. **Ultra-detailed player audit** — `/app/output/reports/20260906_ultra_detailed_player_audit.html`, 30,202,030 bytes.
6. **Structural audit** — `/app/output/reports/audit_report_20260906.html`, 1,864,642 bytes.
7. **Missing:** `/app/output/reports/audit_report_20260906_integrity_revision.html`, purpose “post-Luna remediation proof for immutable revision artifacts.” Its status is explicitly `missing`.

## Coverage

All 15 expected games were reviewed:

ARI@HOU, ATL@PHI, BOS@BAL, CHC@MIA, DET@CLE, LAA@PIT, MIL@CIN, MIN@CHW, NYY@SD, OAK@SEA, SF@NYM, STL@COL, TB@TEX, TOR@KC, and WAS@LAD.

Coverage counts are complete for the supplied artifacts:

- Canonical artifacts loaded: **15/15**
- Continuous-learning games: **15/15**
- Deep-audit games: **15/15**
- Starter comparisons: **15/15**
- Structural audits: **15