Deep slate diagnosis — 20260926

• Games: 13 · hits 7 · misses 6 · winner accuracy 0.5385
• Mean total-run bias on losses: 0.167 (negative = under-projected)
• Mean team-run error on losses: 6.167 · mean Brier on losses: 0.323

Loss primary drivers

• Run environment far higher than projected (under-dispersed): 1
• Starting-pitcher projection was wrong: 4
• Backed the wrong side with real conviction: 1

High-level causal issue register

[CRITICAL] Published probability differs materially without a recorded transformation

• Domain: probability_lineage · affected: COL@CHW
• Evidence: 1 game(s) have an unexplained ≥2pp calibrated-to-published delta
• Action: Block any promotion until the missing probability transformation is recorded and reconciled.
• Do not infer: Do not assume the probability governor was bypassed until you have traced the pick-side math.

[HIGH] An intermediate stage materially degraded winner direction

• Domain: pipeline_stages · affected: CLE@KC, HOU@OAK, NYM@WAS
• Evidence: historical_ml → variance_applied changed winner accuracy by -0.2308 (-3 hits)
• Action: List the exact games whose winner flipped at this stage, and keep both the before-stage and after-stage picks in the trace.
• Do not infer: Even if a later stage recovers the pick, this unstable step is still a risk; do not treat it as harmless.

[HIGH] Published margins are compressed relative to realized margins

• Domain: score_distribution · affected: ATL@MIA, CLE@KC, HOU@OAK, LAA@SEA, NYM@WAS, TB@PHI, TEX@MIN
• Evidence: 12/13 predictions were ≤2 runs; 8/13 actual margins were ≥4; mean predicted margin 1.62 vs actual 4.62
• Action: Check whether predicted margins are too narrow across several recent out-of-sample slates (measuring tail coverage and margin MAE) before widening the score spread.
• Do not infer: A single blowout game is not a reason to widen every score prediction.

[HIGH] Published confidence buckets were poorly aligned on this slate

• Domain: calibration · affected: ARI@SD, ATL@MIA, CIN@TOR, CLE@KC, COL@CHW, HOU@OAK, LAA@SEA, LAD@SF, NYM@WAS, PIT@DET, STL@MIL, TB@PHI, TEX@MIN
• Evidence: weighted absolute calibration error 0.1524 across 13 games
• Action: Watch the same confidence buckets over the full rolling window before changing any calibration weight.
• Do not infer: A single slate has far too few games to establish a stable calibration curve.

[HIGH] Validated run models explain very little holdout variance

• Domain: model_validation · affected: ARI@SD, ATL@MIA, CIN@TOR, CLE@KC, COL@CHW, HOU@OAK, LAA@SEA, LAD@SF, NYM@WAS, PIT@DET, STL@MIL, TB@PHI, TEX@MIN
• Evidence: home/away/total R² = 0.0141/0.0320/0.0038
• Action: Treat exact run scores as low-confidence estimates, and look for stronger features or a better target before tuning the score tails.
• Do not infer: A small positive R² only means the model beats guessing the average; it does not make the exact scores reliable.

[INFO] Every settled game received structural QA

• Domain: coverage · affected: ARI@SD, ATL@MIA, CIN@TOR, CLE@KC, COL@CHW, HOU@OAK, LAA@SEA, LAD@SF, NYM@WAS, PIT@DET, STL@MIL, TB@PHI, TEX@MIN
• Evidence: 14/13 games structurally audited
• Action: Keep the per-game audit receipt in every report.
• Do not infer: Full audit coverage does not cancel the findings raised inside the audited games.

[INFO] Simulation produced no winner-direction intervention on this slate

• Domain: simulation · affected: ARI@SD, ATL@MIA, CIN@TOR, CLE@KC, COL@CHW, HOU@OAK, LAA@SEA, LAD@SF, NYM@WAS, PIT@DET, STL@MIL, TB@PHI, TEX@MIN
• Evidence: flip candidates 0; published flips 0
• Action: Measure the simulation's contribution across rolling slates, separating score realization from winner flipping.
• Do not infer: Zero winner flips is not a defect and does not mean the simulation did no score-reconciliation work.

Cross-slate trends (persistent-pattern memory)

• Window: 20260916, 20260917, 20260918, 20260919, 20260920, 20260921, 20260922, 20260923, 20260924, 20260925, 20260926 (10 prior slates loaded)
• Highest-confidence pick loss streak: 1 slate(s)
• Actionable pattern detected: True

Escalated persistent patterns

• simulation_inert_on_slate — 11 consecutive slates (11 in window)
• Target lever(s): simulation.allow_score
• Direction: confirm the simulation module is running, not silently no-oping
• Guardrail: simulation_inert_on_slate has persisted 11 consecutive slates; this is no longer single-slate noise. Treat it as an actionable pattern and raise the code/diagnostic action named; do not force a parameter move.
• weak_run_model_signal — 11 consecutive slates (11 in window)
• Target lever(s): training.xgb.n_estimators, training.xgb.max_depth, training.xgb.learning_rate
• Direction: investigate features/target via isolated training backtest (manual)
• Guardrail: weak_run_model_signal has persisted 11 consecutive slates; this is no longer single-slate noise. Treat it as an actionable pattern and raise the code/diagnostic action named; do not force a parameter move.

Metric direction across the window

• winner_accuracy: flat · series [0.6667, 0.4444, 0.6667, 0.7333, 0.6, 0.3333, 0.4, 0.5, 0.6667, 0.6471, 0.5385]
• weighted_calibration_error: improving · series [0.1661, 0.1016, 0.1247, 0.3458, 0.3184, 0.2028, 0.1881, 0.1108, 0.1701, 0.1302, 0.1524]
• margin_bias: regressing · series [-1.133, -2.111, -2.0, -0.667, -1.533, -2.667, -1.867, -1.188, -1.083, -1.706, -3.0]
• mean_team_run_error_on_losses: regressing · series [6.4, 3.6, 5.2, 6.5, 6.5, 3.0, 6.222, 5.75, 3.5, 5.333, 6.167]

Pipeline stage quality

• historical_ml: winner 9/13 (0.6923), team-run MAE 2.7122, margin MAE 4.6277
• variance_applied: winner 6/13 (0.4615), team-run MAE 2.7308, margin MAE 4.6923
• enhanced_prediction: winner 7/13 (0.5385), team-run MAE 2.6923, margin MAE 4.7692
• embedded_savant_run_environment: winner 7/13 (0.5385), team-run MAE 2.7692, margin MAE 4.9231
• fundamental_edge_target_review: winner 7/13 (0.5385), team-run MAE 2.8077, margin MAE 5.0
• capped_target: winner 7/13 (0.5385), team-run MAE 2.8077, margin MAE 5.0
• simulation_result: winner 7/13 (0.5385), team-run MAE 2.8077, margin MAE 5.0
• published_final: winner 7/13 (0.5385), team-run MAE 2.8077, margin MAE 5.0

Probability lineage

• ARI@SD: raw 0.5822 → calibrated 0.5474 → published 0.5342; score coherence followed by matchup-recency guardrail
• ATL@MIA: raw 0.6259 → calibrated 0.5703 → published 0.53; score coherence followed by matchup-recency guardrail
• CIN@TOR: raw 0.6356 → calibrated 0.6195 → published 0.6195; published probability matches the calibrated/governed value
• CLE@KC: raw 0.5778 → calibrated 0.6014 → published 0.56; score coherence followed by matchup-recency guardrail
• COL@CHW: raw 0.7429 → calibrated 0.7218 → published 0.6978; no recorded transformation explains the publication delta
• HOU@OAK: raw 0.5595 → calibrated 0.5355 → published 0.5229; score coherence followed by matchup-recency guardrail
• LAA@SEA: raw 0.5622 → calibrated 0.537 → published 0.518; score coherence followed by matchup-recency guardrail
• LAD@SF: raw 0.6376 → calibrated 0.5985 → published 0.5535; score coherence followed by matchup-recency guardrail
• NYM@WAS: raw 0.5 → calibrated 0.5132 → published 0.5064; score coherence followed by matchup-recency guardrail
• PIT@DET: raw 0.5657 → calibrated 0.5388 → published 0.5133; score coherence followed by matchup-recency guardrail
• STL@MIL: raw 0.5672 → calibrated 0.5626 → published 0.5354; score coherence followed by matchup-recency guardrail
• TB@PHI: raw 0.5834 → calibrated 0.6076 → published 0.6076; published probability matches the calibrated/governed value
• TEX@MIN: raw 0.5 → calibrated 0.5117 → published 0.51; published probability matches the calibrated/governed value

Every game, in detail

ARI@SD — ❌ miss

• Predicted ARI 5-4 (conf -0.07, lean) · actual SD 7-10
• Root cause: Run environment far higher than projected (under-dispersed)
• Errors: winner WRONG · Brier 0.285418 · team-run err 8.0 · total bias -8
• Contributing factors:
• total runs under-projected by 8 (9 vs 17)
• SD offense under-projected by 6

ATL@MIA — ✅ hit

• Predicted ATL 5-4 (conf -0.06, lean) · actual ATL 8-3
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.2209 · team-run err 4.0 · total bias -2
• Contributing factors:
• ATL offense under-projected by 3
• Brent Suter (ATL) allowed 4 fewer ER than projected (0 vs 4)

CIN@TOR — ❌ miss

• Predicted TOR 3-5 (conf +0.24, strong) · actual CIN 5-1
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Backed the wrong side with real conviction
• Errors: winner WRONG · Brier 0.383809 · team-run err 6.0 · total bias +2
• Contributing factors:
• TOR offense over-projected by 4
• Rhett Lowder (CIN) allowed 4 fewer ER than projected (0 vs 4)
• committed to TOR at 0.24 confidence and it lost

CLE@KC — ✅ hit

• Predicted CLE 6-5 (conf -0.12, moderate) · actual CLE 11-5
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.1936 · team-run err 5.0 · total bias -5
• Contributing factors:
• total runs under-projected by 5 (11 vs 16)
• CLE offense under-projected by 5
• Tanner Bibee (CLE) allowed 3 more ER than projected (3 vs 0)

COL@CHW — ❌ miss

• Predicted CHW 4-9 (conf +0.40, strong) · actual COL 9-6
• Root cause: Backed the wrong side with real conviction
• Secondary cause: Small edge on the losing side
• Errors: winner WRONG · Brier 0.486882 · team-run err 8.0 · total bias -2
• Contributing factors:
• COL offense under-projected by 5
• CHW offense over-projected by 3
• committed to CHW at 0.40 confidence and it lost

HOU@OAK — ✅ hit

• Predicted HOU 6-5 (conf -0.05, lean) · actual HOU 13-2
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.227629 · team-run err 10.0 · total bias -4
• Contributing factors:
• total runs under-projected by 4 (11 vs 15)
• HOU offense under-projected by 7
• OAK offense over-projected by 3
• Jack Perkins (OAK) allowed 3 more ER than projected (6 vs 3)

LAA@SEA — ✅ hit

• Predicted SEA 4-5 (conf +0.04, toss_up) · actual SEA 4-8
• Root cause: toss_up_went_our_way
• Errors: winner correct · Brier 0.232304 · team-run err 3.0 · total bias -3
• Contributing factors:
• total runs under-projected by 3 (9 vs 12)
• SEA offense under-projected by 3
• Ryan Johnson (LAA) allowed 3 more ER than projected (4 vs 1)

LAD@SF — ✅ hit

• Predicted LAD 4-2 (conf -0.11, moderate) · actual LAD 4-3
• Root cause: strong_correct_call
• Errors: winner correct · Brier 0.199385 · team-run err 1.0 · total bias -1

NYM@WAS — ❌ miss

• Predicted WAS 6-7 (conf +0.01, toss_up) · actual NYM 7-1
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Run environment far lower than projected (over-dispersed)
• Errors: winner WRONG · Brier 0.256482 · team-run err 7.0 · total bias +5
• Contributing factors:
• total runs over-projected by 5 (13 vs 8)
• WAS offense over-projected by 6
• Connelly Early (WAS) allowed 5 fewer ER than projected (0 vs 5)

PIT@DET — ❌ miss

• Predicted PIT 6-4 (conf -0.03, toss_up) · actual DET 3-4
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Genuine toss-up that flipped
• Errors: winner WRONG · Brier 0.263515 · team-run err 3.0 · total bias +3
• Contributing factors:
• total runs over-projected by 3 (10 vs 7)
• PIT offense over-projected by 3
• Justin Verlander (DET) allowed 4 fewer ER than projected (2 vs 6)

STL@MIL — ✅ hit

• Predicted MIL 4-5 (conf +0.07, lean) · actual MIL 2-3
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.215812 · team-run err 4.0 · total bias +4
• Contributing factors:
• total runs over-projected by 4 (9 vs 5)

TB@PHI — ✅ hit

• Predicted TB 4-2 (conf -0.22, strong) · actual TB 12-1
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.154005 · team-run err 9.0 · total bias -7
• Contributing factors:
• total runs under-projected by 7 (6 vs 13)
• TB offense under-projected by 8

TEX@MIN — ❌ miss

• Predicted MIN 4-5 (conf +0.02, toss_up) · actual TEX 6-2
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Genuine toss-up that flipped
• Errors: winner WRONG · Brier 0.260082 · team-run err 5.0 · total bias +1
• Contributing factors:
• MIN offense over-projected by 3
• Nathan Eovaldi (TEX) allowed 3 fewer ER than projected (1 vs 4)
• Bailey Ober (MIN) allowed 5 more ER than projected (6 vs 1)

Cross-game lever tension

Each candidate fix, with the games it would help and the games it would hurt. A lever that repairs one game and breaks two is visible here before it is proposed.

score.variance_multiplier / score.variance_floor (widen run distributions)

• Targets: brier_score, calibration_error
• Net: 2 helped vs 1 hurt
• Would help:
• CIN@TOR (miss): lost a 0.24-confidence pick; wider tails lower overconfidence
• COL@CHW (miss): lost a 0.40-confidence pick; wider tails lower overconfidence
• Would hurt / risk:
• TB@PHI (hit): strong correct call at 0.22; widening softens a good edge

player.pitcher.v4 form weights / player.pitcher_prior_games

• Targets: starter_er_mae, starter_outs_mae
• Net: 8 helped vs 2 hurt
• Would help:
• ATL@MIA (hit): starter ER off by 4
• CIN@TOR (miss): starter ER off by 4
• CLE@KC (hit): starter ER off by 3
• HOU@OAK (hit): starter ER off by 3
• LAA@SEA (hit): starter ER off by 3
• NYM@WAS (miss): starter ER off by 5
• PIT@DET (miss): starter ER off by 4
• TEX@MIN (miss): starter ER off by 5
• Would hurt / risk:
• LAD@SF (hit): starter projection was accurate (ER off 1); reweight risks disturbing it
• TB@PHI (hit): starter projection was accurate (ER off 1); reweight risks disturbing it

recent. / season. run-mean weighting

• Targets: team_run_mae, total_run_mae, accuracy
• Net: 10 helped vs 1 hurt
• Would help:
• ARI@SD (miss): team runs off by 8
• ATL@MIA (hit): team runs off by 4
• CIN@TOR (miss): team runs off by 6
• CLE@KC (hit): team runs off by 5
• COL@CHW (miss): team runs off by 8
• HOU@OAK (hit): team runs off by 10
• NYM@WAS (miss): team runs off by 7
• STL@MIL (hit): team runs off by 4
• TB@PHI (hit): team runs off by 9
• TEX@MIN (miss): team runs off by 5
• Would hurt / risk:
• LAD@SF (hit): team runs already close (off 1); reweight risks disturbing it