Nightly self-improvement · slate 2026-09-18
Generated 2026-09-19T05:00:00.006822-04:00 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
2026-09-18
status completed
Winners
10/15
Brier 0.2141
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 |
|---|---|---|---|---|---|---|
| ATL@HOU | 4-3 | 6-2 | 0.438 | hit | Tyler Mahle: 18 proj / 18 real outs; Peter Lambert: 16 proj / 16 real outs | |
| BOS@TB | 4-5 | 4-2 | 0.572 | miss | Backed the wrong side with real conviction | Ranger Suarez: 17 proj / 17 real outs; Ian Seymour: 15 proj / 15 real outs |
| CHC@CIN | 6-4 | 4-6 | 0.470 | miss | Small edge on the losing side | Clay Holmes: 17 proj / 16 real outs; Chase Burns: 15 proj / 9 real outs |
| DET@CHW | 6-4 | 11-8 | 0.457 | hit | Andrew Sears: 14 proj / 2 real outs; David Sandlin: 16 proj / 12 real outs | |
| KC@PIT | 3-4 | 5-8 | 0.586 | hit | Randy Dobnak: 18 proj / 11 real outs; Paul Skenes: 16 proj / 18 real outs | |
| MIA@SD | 4-6 | 2-8 | 0.580 | hit | Tyler Phillips: 15 proj / 11 real outs; Nick Pivetta: 15 proj / 15 real outs | |
| MIL@BAL | 7-4 | 6-5 | 0.334 | hit | Dustin May: 14 proj / 7 real outs; Cade Povich: 13 proj / 11 real outs | |
| MIN@LAA | 4-5 | 3-0 | 0.534 | miss | Starting-pitcher projection was wrong | Connor Prielipp: 15 proj / 21 real outs; Grayson Rodriguez: 15 proj / 20 real outs |
| NYY@ARI | 5-4 | 9-2 | 0.495 | hit | Gerrit Cole: 17 proj / 18 real outs; Eduardo Rodriguez: 18 proj / 16 real outs | |
| OAK@CLE | 3-7 | 3-5 | 0.580 | hit | Mason Barnett: 11 proj / 10 real outs; Daniel Espino: 9 proj / 3 real outs | |
| PHI@NYM | 5-4 | 3-6 | 0.488 | miss | Genuine toss-up that flipped | Tim Mayza: 15 proj / 1 real outs; Zac Thornton: 17 proj / 18 real outs |
| SEA@COL | 9-5 | 5-4 | 0.240 | hit | Bryan Woo: 18 proj / 16 real outs; Gabriel Hughes: 12 proj / 14 real outs | |
| SF@LAD | 2-4 | 2-8 | 0.538 | hit | Cesar Perdomo: 15 proj / 14 real outs; Tyler Glasnow: 17 proj / 18 real outs | |
| TOR@TEX | 4-5 | 1-7 | 0.505 | hit | Dylan Cease: 18 proj / 8 real outs; Kumar Rocker: 15 proj / 13 real outs | |
| WAS@STL | 3-4 | 9-1 | 0.509 | miss | Starting-pitcher projection was wrong | Cade Cavalli: 17 proj / 15 real outs; Kyle Leahy: 15 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 | 0.2425 | 0.2413 | -0.0012 | improved |
| accuracy | 0.5917 | 0.5833 | -0.0084 | regressed |
| team_run_mae | 2.5333 | 2.4333 | -0.1000 | improved |
| total_run_mae | 3.5333 | 3.2333 | -0.3000 | improved |
| calibration_error | 0.0728 | 0.0572 | -0.0156 | improved |
| starter_er_mae | 1.7326 | 1.6860 | -0.0466 | improved |
| starter_outs_mae | 3.2984 | 3.3721 | 0.0737 | 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
No Luna post-mortem this night.