MLB Prediction System

Metric Evolution & Audit

Every number that moved, the stage that moved it, the input it used, and the arithmetic that proves it

Slate 20261005Games 2Checks 313Schema metric_evolution_report_v1Built 2026-10-06 17:19Z

Verdict across the slate

313 invariants checked across 2 games: NO DEFECT FOUND

How to read this, and what a defect looks like

Every figure below is shown with the arithmetic that produced it and a verdict beside it, so a disagreement can be pointed at a specific line rather than at the system as a whole.

Stage chain. Each game lists every stage that touched the score, the score after it, and whether it changed the target. Most stages move nothing on most games; a chain that showed only movement would misrepresent it. A stage reporting a delta while changed_target is false computed an adjustment and did not apply it, which is normal and is stated.
Blend arithmetic. The KPI blend must satisfy blended = baseline + weight x (kpi_only - baseline) for all three targets. A FAIL here means the layer reported a number it did not compute.
Rounding. Published integers must be the rounding of the blended runs under the documented tie rule. A FAIL means the published score and the continuous line describe different games.
Log5 additivity. For every player metric, own skill plus opponent plus the Log5 remainder must equal the total against the league, exactly. A FAIL means the decomposition is not a decomposition.
Direction. A green delta helps the subject, a red one hurts him. Strikeouts invert for a hitter. If an elite arm's opponent shows green strikeouts, the orientation is wrong and the whole player section is suspect.
Measured versus published. The KPI layer moves team runs and the win probability. The per-player metric adjustments here are a measurement of the same catalog at player level, not what the pipeline published; the one club-wide ratio the pipeline actually applied to hitter lines is printed beside them.
Integer versus continuous. The integer player line is a simulation draw reconciled to the box score. The continuous expected_* beside it is the model's own view and is the rankable quantity. Judging the model by the integer is how a working model looks broken.

Slate overview

Game	Published	p(home)	Stages	Stages that moved it	KPI applied	Δ away runs	Δ home runs	Simulation draw chosen on	0-out starter
20261005_CHW@CLE	4 - 5	0.5185	8	1	yes	-0.2841	+0.2723	-	no
20261005_NYY@TB	4 - 2	0.4604	8	3	yes	+0.3329	+0.0510	-	no

KPI model in force. candidate 20261005T125039Z_f5114dc0, 48 features built from 40 registered KPIs, blend weights {"home_runs": 0.44, "away_runs": 0.78, "home_win": 0.0}, frozen evaluation ends 2026-09-26. A target whose weight is zero contributes nothing by construction: the winner probability carries weight zero while its walk-forward verdict is inconclusive, so every win-probability movement below is expected to be exactly zero.

Per game

20261005_CHW@CLE — published 4-5, 156 checks, all pass

Artifact. 20261005_CHW@CLE__20261006T023851Z.json, phase post_start_rerun, generated 2026-10-06T02:38:51, first pitch 2026-10-05T21:00:00.
Published 4 - 5 with p(home) 0.518498. Simulation draw chosen on not recorded; ? of ? draws realized the target exactly.

Stage chain: every stage that touched the score

Stage	Score after	Δ away	Δ home	Cum. away	Cum. home	Changed target	Own adjustment and description
1. `historical_ml`	4.052787417582998 - 4.657712402044706	0	0	0	0	no	Base ML historical model prediction
2. `variance_applied`	4 - 5	-0.0528	+0.3423	-0.0528	+0.3423	yes	Deterministic baseline target (variance factor 1.0)
3. `enhanced_prediction`	4 - 5	0	0	-0.0528	+0.3423	no	Enhanced ML with recent form, H2H, low-score analysis (confidence shrunk by projected_margin_is_one_run, matchup_recency_guardrail)
4. `kpi_layer`	4 - 5	0	0	-0.0528	+0.3423	no	away -0.2841; home +0.2723 Savant KPI blend applied to the model target
5. `embedded_savant_run_environment`	4 - 5	0	0	-0.0528	+0.3423	no	away -0.1691; home +0.2405 Target score adjusted by temporal Savant lineup quality versus opposing starter quality
6. `fundamental_edge_target_review`	4 - 5	0	0	-0.0528	+0.3423	no	Refreshed score target and win probability after Savant/pitcher/team-strength adjustments
7. `capped_target`	4 - 5	0	0	-0.0528	+0.3423	no	Capped at max 12 realistic runs
8. `simulation_result`	4 - 5	-	-	-	-	no	Final inning-by-inning simulation result

A stage reporting its own adjustment while Changed target reads no computed a movement that a later reconciliation absorbed, or that rounded away at the integer score. That is recorded rather than hidden.

KPI team layer: the blend, with its arithmetic

Target	Baseline	KPI only	Weight	Blended	Computation	Verdict
`away_runs`	4.0000	3.6357	0.78	3.7159	4.0000 + 0.78 × (3.6357 − 4.0000) = 3.7159	PASS
`home_runs`	5.0000	5.6188	0.44	5.2723	5.0000 + 0.44 × (5.6188 − 5.0000) = 5.2723	PASS
`home_win`	0.5256	0.5791	0.00	0.5256	0.5256 + 0.00 × (0.5791 − 0.5256) = 0.5256	PASS

Published by the layer: away 4, home 5, p(home) 0.525603, continuous away 3.7159 / home 5.2723

Inputs and gates. candidate 20261005T125039Z_f5114dc0, value age 8 days, lineup coverage away 0.8947 / home 1.0000, movement cap 2.5 runs.
values_present ok values_fresh ok one_source_bundle ok lineup_coverage ok target_eligible ok movement_within_limit ok published_movement_within_limit ok

What drove the KPI forecast for `away_runs`

intercept 4.2573 + sum of contributions = 3.6357; prediction 3.6357; contributions outside the top listed 1.1556 runs

Feature	Raw	Standardised	Runs contributed	Which KPIs it is built from
`away_x_fastball`	-0.9471	-5.759	-1.2283	batter x pitcher interaction: pitcher_whiff_rate_fastball, batter_whiff_rate_vs_fastball
`home_off_power_spread`	1.0906	1.365	+0.8901	offence group (dispersion across the lineup, not its mean): batter_home_run_rate, batter_bat_speed
`home_platoon_whiff`	0.5948	1.899	-0.8165	platoon term, selected by the opposing starter's throwing hand
`home_off_qualified_share`	1.0000	2.096	-0.7754	share of the lineup carrying a qualifying sample, not a skill measurement
`home_platoon_barrel`	-0.4177	-1.566	+0.6170	platoon term, selected by the opposing starter's throwing hand
`away_sp_whiff`	1.1633	2.168	+0.5548	starter group: pitcher_whiff_rate, pitcher_whiff_rate_fastball, pitcher_whiff_rate_breaking
`away_off_worst3_discipline`	-1.1139	-2.311	-0.5384	mean of the three weakest plate approaches
`away_sp_command`	0.8911	1.479	-0.4717	starter group: pitcher_strikeout_rate, pitcher_walk_rate
`away_off_contact_spread`	0.9624	1.353	-0.4677	offence group (dispersion across the lineup, not its mean): batter_barrel_rate_vs_fastball, batter_barrel_rate_vs_breaking, batter_barrel_rate_vs_offspeed, batter_hard_hit_rate_vs_fastball ...
`home_off_discipline`	0.5366	2.246	+0.4589	offence group: batter_whiff_rate_vs_fastball, batter_whiff_rate_vs_breaking, batter_whiff_rate_vs_offspeed, batter_strikeout_rate ...

What drove the KPI forecast for `home_runs`

intercept 4.2997 + sum of contributions = 5.6188; prediction 5.6188; contributions outside the top listed 2.0013 runs

Feature	Raw	Standardised	Runs contributed	Which KPIs it is built from
`away_x_suppress`	-0.8553	-3.665	+0.8801	batter x pitcher interaction: pitcher_barrel_rate_allowed, batter_barrel_rate_vs_fastball
`home_off_discipline`	0.5366	2.246	-0.5282	offence group: batter_whiff_rate_vs_fastball, batter_whiff_rate_vs_breaking, batter_whiff_rate_vs_offspeed, batter_strikeout_rate ...
`away_off_contact_spread`	0.9624	1.353	-0.4390	offence group (dispersion across the lineup, not its mean): batter_barrel_rate_vs_fastball, batter_barrel_rate_vs_breaking, batter_barrel_rate_vs_offspeed, batter_hard_hit_rate_vs_fastball ...
`away_off_discipline_spread`	0.9369	2.473	-0.3999	offence group (dispersion across the lineup, not its mean): batter_whiff_rate_vs_fastball, batter_whiff_rate_vs_breaking, batter_whiff_rate_vs_offspeed, batter_strikeout_rate ...
`home_off_power`	-0.2953	-1.493	-0.3968	offence group: batter_home_run_rate, batter_bat_speed
`away_off_top3_contact`	1.2780	1.722	-0.3740	mean of the three best contact bats, a tail a club mean cannot express
`home_off_ground_spread`	0.3813	-1.257	-0.3695	offence group (dispersion across the lineup, not its mean): batter_ground_ball_rate_vs_slider, batter_ground_ball_rate_vs_fastball
`away_off_power_spread`	1.1543	1.469	+0.3287	offence group (dispersion across the lineup, not its mean): batter_home_run_rate, batter_bat_speed
`away_off_ground_spread`	1.1795	2.464	+0.3106	offence group (dispersion across the lineup, not its mean): batter_ground_ball_rate_vs_slider, batter_ground_ball_rate_vs_fastball
`away_x_fastball`	-0.9471	-5.759	+0.3057	batter x pitcher interaction: pitcher_whiff_rate_fastball, batter_whiff_rate_vs_fastball

These are Ridge coefficients on standardised inputs, so each contribution is that feature's own push on the run forecast, in runs. They sum exactly to the prediction minus the intercept, which makes the list an accounting of the forecast rather than a ranking of vaguely important things.

Per-player metric movement

CHW against Hagen Smith (LHP)

The pipeline applied one club-wide ×0.929 to every CHW hitter line, identical for the leadoff hitter and the ninth bat.
Opposing staff exposure. These hitters face the starter for 23.1868 of 37.398 projected batters (62.0%) and the 3 projected relievers for the remaining 38.0%. Every metric below is the exposure-weighted blend of both matchups; the starter-only figure is shown beside it. Real 2026 game logs put a starter at 21.75 batters faced on average (median 23, p10-p90 17-27) and a whole staff at 37.9: this starter is inside that band and this staff is inside its. The per-player columns are the measured alternative: each subject moved by his own KPI pairing. Exposure for a hitter is his projected plate appearances; for the starter it is batters faced.

Player	Pos	Metric	Published → KPI	Δ	League	Own rate	Opponent rate	Log5	Own skill	Opponent	Remainder
Sam Antonacci	LF	Hitswhole staff	0.841 → 0.775	-0.066	0.2169	0.2300n=553 Good	0.1978n=721 Good	0.2101	+0.055	-0.080	-0.004 ok
		split	starter x0.9134 at 0.620 share (log5 0.2101 vs his 0.1978) + bullpen x0.9344 at 0.380 share (log5 0.2149 vs pen 0.2024) = x0.9214 \| starter alone would have been -0.073
		Strikeoutswhole staff	0.842 → 1.167	+0.325	0.2212	0.1525n=553 Good	0.3179n=721 Very good	0.2280	-0.289	+0.407	-0.089 ok
		split	starter x1.495 at 0.620 share (log5 0.2280 vs his 0.3179) + bullpen x1.2074 at 0.380 share (log5 0.1842 vs pen 0.2627) = x1.3857 \| starter alone would have been +0.417
		Walkswhole staff	0.346 → 0.351	+0.004	0.0889	0.0797n=553 Average	0.0854n=721 Average	0.0765	-0.039	-0.015	+0.001 ok
		split	starter x0.9598 at 0.620 share (log5 0.0765 vs his 0.0854) + bullpen x1.0988 at 0.380 share (log5 0.0876 vs pen 0.0976) = x1.0126 \| starter alone would have been -0.014
		Home runswhole staff	0.068 → 0.069	+0.001	0.0304	0.0196n=553 Poor	0.0330n=721 Below average	0.0213	-0.045	+0.011	-0.004 ok
		split	starter x1.0881 at 0.620 share (log5 0.0213 vs his 0.0330) + bullpen x0.909 at 0.380 share (log5 0.0178 vs pen 0.0277) = x1.02 \| starter alone would have been +0.006
Neutralising Sam Antonacci to the qualified-population mean moves the club's KPI run forecast by 0.1940 runs, and the weighted stage by 0.1514. Interactions make these non-additive across players, so they do not sum to the published movement.
Kyle Teel	C	Hitswhole staff	0.707 → 0.650	-0.056	0.2169	0.2169n=0 averagebelow floor	0.1978n=721 Good	0.1978	0	-0.076	0 ok
		split	starter x0.9121 at 0.620 share (log5 0.1978 vs his 0.1978) + bullpen x0.9333 at 0.380 share (log5 0.2024 vs pen 0.2024) = x0.9202 \| starter alone would have been -0.062
		Strikeoutswhole staff	1.730 → 2.323	+0.592	0.2212	0.2212n=0 averagebelow floor	0.3179n=721 Very good	0.3179	0	+0.387	0 ok
		split	starter x1.4373 at 0.620 share (log5 0.3179 vs his 0.3179) + bullpen x1.1875 at 0.380 share (log5 0.2627 vs pen 0.2627) = x1.3424 \| starter alone would have been +0.757
		Walkswhole staff	0.417 → 0.422	+0.005	0.0889	0.0889n=0 averagebelow floor	0.0854n=721 Average	0.0854	0	-0.014	0 ok
		split	starter x0.9601 at 0.620 share (log5 0.0854 vs his 0.0854) + bullpen x1.0977 at 0.380 share (log5 0.0976 vs pen 0.0976) = x1.0124 \| starter alone would have been -0.017
		Home runswhole staff	0.102 → 0.104	+0.002	0.0304	0.0304n=0 averagebelow floor	0.0330n=721 Below average	0.0330	0	+0.011	0 ok
		split	starter x1.0871 at 0.620 share (log5 0.0330 vs his 0.0330) + bullpen x0.9099 at 0.380 share (log5 0.0277 vs pen 0.0277) = x1.0198 \| starter alone would have been +0.009
Neutralising Kyle Teel to the qualified-population mean moves the club's KPI run forecast by -0.4272 runs, and the weighted stage by -0.3332. Interactions make these non-additive across players, so they do not sum to the published movement.
Miguel Vargas	3B	Hitswhole staff	0.831 → 0.764	-0.067	0.2169	0.2136n=695 Average	0.1978n=721 Good	0.1947	-0.014	-0.081	+0.001 ok
		split	starter x0.9117 at 0.620 share (log5 0.1947 vs his 0.1978) + bullpen x0.933 at 0.380 share (log5 0.1993 vs pen 0.2024) = x0.9198 \| starter alone would have been -0.073
		Strikeoutswhole staff	0.992 → 1.355	+0.363	0.2212	0.1826n=695 Good	0.3179n=721 Very good	0.2682	-0.164	+0.411	-0.047 ok
		split	starter x1.4692 at 0.620 share (log5 0.2682 vs his 0.3179) + bullpen x1.1986 at 0.380 share (log5 0.2188 vs pen 0.2627) = x1.3664 \| starter alone would have been +0.465
		Walkswhole staff	0.528 → 0.535	+0.006	0.0889	0.1346n=695 Very good	0.0854n=721 Average	0.1295	+0.194	-0.015	-0.007 ok
		split	starter x0.9621 at 0.620 share (log5 0.1295 vs his 0.0854) + bullpen x1.0924 at 0.380 share (log5 0.1470 vs pen 0.0976) = x1.0116 \| starter alone would have been -0.020
		Home runswhole staff	0.154 → 0.157	+0.003	0.0304	0.0428n=695 Very good	0.0330n=721 Below average	0.0464	+0.053	+0.011	+0.004 ok
		split	starter x1.0859 at 0.620 share (log5 0.0464 vs his 0.0330) + bullpen x0.911 at 0.380 share (log5 0.0390 vs pen 0.0277) = x1.0194 \| starter alone would have been +0.013
Neutralising Miguel Vargas to the qualified-population mean moves the club's KPI run forecast by 0.5724 runs, and the weighted stage by 0.4465. Interactions make these non-additive across players, so they do not sum to the published movement.
Munetaka Murakami	1B	Hitswhole staff	0.724 → 0.665	-0.059	0.2169	0.1915n=544 Poor	0.1978n=721 Good	0.1742	-0.111	-0.083	+0.008 ok
		split	starter x0.9095 at 0.620 share (log5 0.1742 vs his 0.1978) + bullpen x0.9313 at 0.380 share (log5 0.1783 vs pen 0.2024) = x0.9178 \| starter alone would have been -0.066
		Strikeoutswhole staff	1.873 → 2.383	+0.509	0.2212	0.3438n=544 Poor	0.3179n=721 Very good	0.4623	+0.536	+0.423	+0.095 ok
		split	starter x1.3447 at 0.620 share (log5 0.4623 vs his 0.3179) + bullpen x1.1534 at 0.380 share (log5 0.3966 vs pen 0.2627) = x1.272 \| starter alone would have been +0.646
		Walkswhole staff	0.632 → 0.640	+0.007	0.0889	0.1532n=544 Very good	0.0854n=721 Average	0.1475	+0.281	-0.015	-0.009 ok
		split	starter x0.9629 at 0.620 share (log5 0.1475 vs his 0.0854) + bullpen x1.0902 at 0.380 share (log5 0.1670 vs pen 0.0976) = x1.0113 \| starter alone would have been -0.024
		Home runswhole staff	0.211 → 0.215	+0.004	0.0304	0.0532n=544 Very good	0.0330n=721 Below average	0.0578	+0.100	+0.012	+0.008 ok
		split	starter x1.0848 at 0.620 share (log5 0.0578 vs his 0.0330) + bullpen x0.9118 at 0.380 share (log5 0.0485 vs pen 0.0277) = x1.0191 \| starter alone would have been +0.018
Neutralising Munetaka Murakami to the qualified-population mean moves the club's KPI run forecast by -0.4628 runs, and the weighted stage by -0.3610. Interactions make these non-additive across players, so they do not sum to the published movement.
Andrew Benintendi	DH	Hitswhole staff	0.706 → 0.650	-0.056	0.2169	0.2160n=460 Average	0.1978n=721 Good	0.1970	-0.003	-0.066	+0.000 ok
		split	starter x0.912 at 0.620 share (log5 0.1970 vs his 0.1978) + bullpen x0.9332 at 0.380 share (log5 0.2016 vs pen 0.2024) = x0.9201 \| starter alone would have been -0.062
		Strikeoutswhole staff	1.021 → 1.361	+0.341	0.2212	0.2354n=460 Average	0.3179n=721 Very good	0.3357	+0.049	+0.334	+0.012 ok
		split	starter x1.4259 at 0.620 share (log5 0.3357 vs his 0.3179) + bullpen x1.1834 at 0.380 share (log5 0.2786 vs pen 0.2627) = x1.3337 \| starter alone would have been +0.435
		Walkswhole staff	0.308 → 0.312	+0.004	0.0889	0.0856n=460 Average	0.0854n=721 Average	0.0822	-0.011	-0.012	+0.000 ok
		split	starter x0.96 at 0.620 share (log5 0.0822 vs his 0.0854) + bullpen x1.0981 at 0.380 share (log5 0.0940 vs pen 0.0976) = x1.0125 \| starter alone would have been -0.012
		Home runswhole staff	0.088 → 0.090	+0.002	0.0304	0.0318n=460 Average	0.0330n=721 Below average	0.0346	+0.005	+0.009	+0.000 ok
		split	starter x1.0869 at 0.620 share (log5 0.0346 vs his 0.0330) + bullpen x0.91 at 0.380 share (log5 0.0289 vs pen 0.0277) = x1.0197 \| starter alone would have been +0.008
Neutralising Andrew Benintendi to the qualified-population mean moves the club's KPI run forecast by 0.2292 runs, and the weighted stage by 0.1788. Interactions make these non-additive across players, so they do not sum to the published movement.
Tristan Peters	CF	Hitswhole staff	0.756 → 0.697	-0.060	0.2169	0.2266n=484 Good	0.1978n=721 Good	0.2069	+0.036	-0.071	-0.002 ok
		split	starter x0.9131 at 0.620 share (log5 0.2069 vs his 0.1978) + bullpen x0.9341 at 0.380 share (log5 0.2117 vs pen 0.2024) = x0.9211 \| starter alone would have been -0.066
		Strikeoutswhole staff	0.924 → 1.262	+0.338	0.2212	0.1835n=484 Good	0.3179n=721 Very good	0.2695	-0.140	+0.360	-0.040 ok
		split	starter x1.4684 at 0.620 share (log5 0.2695 vs his 0.3179) + bullpen x1.1983 at 0.380 share (log5 0.2199 vs pen 0.2627) = x1.3658 \| starter alone would have been +0.433
		Walkswhole staff	0.298 → 0.302	+0.004	0.0889	0.0789n=484 Average	0.0854n=721 Average	0.0757	-0.037	-0.013	+0.001 ok
		split	starter x0.9597 at 0.620 share (log5 0.0757 vs his 0.0854) + bullpen x1.0989 at 0.380 share (log5 0.0867 vs pen 0.0976) = x1.0126 \| starter alone would have been -0.012
		Home runswhole staff	0.083 → 0.085	+0.002	0.0304	0.0268n=484 Average	0.0330n=721 Below average	0.0291	-0.013	+0.010	-0.001 ok
		split	starter x1.0874 at 0.620 share (log5 0.0291 vs his 0.0330) + bullpen x0.9096 at 0.380 share (log5 0.0243 vs pen 0.0277) = x1.0198 \| starter alone would have been +0.007
Neutralising Tristan Peters to the qualified-population mean moves the club's KPI run forecast by 0.0078 runs, and the weighted stage by 0.0061. Interactions make these non-additive across players, so they do not sum to the published movement.
Braden Montgomery	RF	Hitswhole staff	0.720 → 0.662	-0.058	0.2169	0.2100n=362 Below average	0.1978n=721 Good	0.1914	-0.026	-0.072	+0.002 ok
		split	starter x0.9114 at 0.620 share (log5 0.1914 vs his 0.1978) + bullpen x0.9328 at 0.380 share (log5 0.1959 vs pen 0.2024) = x0.9195 \| starter alone would have been -0.064
		Strikeoutswhole staff	0.995 → 1.349	+0.354	0.2212	0.1991n=362 Average	0.3179n=721 Very good	0.2898	-0.084	+0.366	-0.023 ok
		split	starter x1.4554 at 0.620 share (log5 0.2898 vs his 0.3179) + bullpen x1.1938 at 0.380 share (log5 0.2377 vs pen 0.2627) = x1.356 \| starter alone would have been +0.453
		Walkswhole staff	0.374 → 0.379	+0.005	0.0889	0.1010n=362 Good	0.0854n=721 Average	0.0970	+0.046	-0.013	-0.002 ok
		split	starter x0.9607 at 0.620 share (log5 0.0970 vs his 0.0854) + bullpen x1.0963 at 0.380 share (log5 0.1107 vs pen 0.0976) = x1.0122 \| starter alone would have been -0.015
		Home runswhole staff	0.069 → 0.070	+0.001	0.0304	0.0224n=362 Below average	0.0330n=721 Below average	0.0244	-0.030	+0.010	-0.003 ok
		split	starter x1.0879 at 0.620 share (log5 0.0244 vs his 0.0330) + bullpen x0.9092 at 0.380 share (log5 0.0204 vs pen 0.0277) = x1.02 \| starter alone would have been +0.006
Neutralising Braden Montgomery to the qualified-population mean moves the club's KPI run forecast by 0.1445 runs, and the weighted stage by 0.1127. Interactions make these non-additive across players, so they do not sum to the published movement.
Chase Meidroth	2B	Hitswhole staff	0.781 → 0.720	-0.061	0.2169	0.2330n=635 Good	0.1978n=721 Good	0.2129	+0.059	-0.070	-0.004 ok
		split	starter x0.9137 at 0.620 share (log5 0.2129 vs his 0.1978) + bullpen x0.9346 at 0.380 share (log5 0.2177 vs pen 0.2024) = x0.9216 \| starter alone would have been -0.067
		Strikeoutswhole staff	1.054 → 1.417	+0.363	0.2212	0.2176n=635 Average	0.3179n=721 Very good	0.3134	-0.013	+0.356	-0.004 ok
		split	starter x1.4403 at 0.620 share (log5 0.3134 vs his 0.3179) + bullpen x1.1885 at 0.380 share (log5 0.2586 vs pen 0.2627) = x1.3446 \| starter alone would have been +0.464
		Walkswhole staff	0.327 → 0.331	+0.004	0.0889	0.0936n=635 Average	0.0854n=721 Average	0.0899	+0.017	-0.013	-0.001 ok
		split	starter x0.9603 at 0.620 share (log5 0.0899 vs his 0.0854) + bullpen x1.0972 at 0.380 share (log5 0.1027 vs pen 0.0976) = x1.0123 \| starter alone would have been -0.013
		Home runswhole staff	0.083 → 0.085	+0.002	0.0304	0.0267n=635 Average	0.0330n=721 Below average	0.0291	-0.013	+0.010	-0.001 ok
		split	starter x1.0874 at 0.620 share (log5 0.0291 vs his 0.0330) + bullpen x0.9096 at 0.380 share (log5 0.0243 vs pen 0.0277) = x1.0198 \| starter alone would have been +0.007
Neutralising Chase Meidroth to the qualified-population mean moves the club's KPI run forecast by 0.1770 runs, and the weighted stage by 0.1381. Interactions make these non-additive across players, so they do not sum to the published movement.
Colson Montgomery	SS	Hitswhole staff	0.592 → 0.544	-0.049	0.2169	0.1929n=630 Poor	0.1978n=721 Good	0.1755	-0.084	-0.067	+0.006 ok
		split	starter x0.9096 at 0.620 share (log5 0.1755 vs his 0.1978) + bullpen x0.9314 at 0.380 share (log5 0.1797 vs pen 0.2024) = x0.9179 \| starter alone would have been -0.053
		Strikeoutswhole staff	1.683 → 2.139	+0.456	0.2212	0.3459n=630 Poor	0.3179n=721 Very good	0.4647	+0.440	+0.341	+0.077 ok
		split	starter x1.3433 at 0.620 share (log5 0.4647 vs his 0.3179) + bullpen x1.1529 at 0.380 share (log5 0.3988 vs pen 0.2627) = x1.2709 \| starter alone would have been +0.578
		Walkswhole staff	0.309 → 0.312	+0.004	0.0889	0.0836n=630 Average	0.0854n=721 Average	0.0802	-0.019	-0.013	+0.001 ok
		split	starter x0.9599 at 0.620 share (log5 0.0802 vs his 0.0854) + bullpen x1.0984 at 0.380 share (log5 0.0918 vs pen 0.0976) = x1.0125 \| starter alone would have been -0.012
		Home runswhole staff	0.129 → 0.132	+0.003	0.0304	0.0436n=630 Very good	0.0330n=721 Below average	0.0474	+0.047	+0.009	+0.004 ok
		split	starter x1.0858 at 0.620 share (log5 0.0474 vs his 0.0330) + bullpen x0.911 at 0.380 share (log5 0.0398 vs pen 0.0277) = x1.0194 \| starter alone would have been +0.011
Neutralising Colson Montgomery to the qualified-population mean moves the club's KPI run forecast by 0.0168 runs, and the weighted stage by 0.0131. Interactions make these non-additive across players, so they do not sum to the published movement.
STARTER Anthony Kay20.6793 BF	Starter	Hits	5.084 → 5.173	+0.089	0.2169	0.2242n=686 Below average	0.22079/9 qualified lineup mean	0.2281	+0.152	+0.079	+0.002 ok
		Strikeouts	3.531 → 2.960	-0.572	0.2212	0.1837n=686 Below average	0.18689/9 qualified lineup mean	0.1540	-0.776	-0.712	+0.097 ok
		Walks	1.555 → 1.549	-0.006	0.0889	0.0818n=686 Average	0.08869/9 qualified lineup mean	0.0815	-0.147	-0.007	+0.001 ok
		Home runs	0.514 → 0.485	-0.029	0.0304	0.0305n=686 Average	0.02879/9 qualified lineup mean	0.0288	+0.003	-0.035	-0.000 ok
RELIEVER Sean Newcomb3.6911 BF	Middle Reliever (High Leverage)	Hits	1.889 → 1.923	+0.034	0.2169	0.2048n=369 Good	0.22079/9 qualified lineup mean	0.2084	-0.045	+0.014	-0.001 ok
		Strikeouts	1.485 → 1.255	-0.230	0.2212	0.2265n=369 Average	0.18689/9 qualified lineup mean	0.1915	+0.019	-0.127	-0.002 ok
		Walks	0.562 → 0.560	-0.002	0.0889	0.0912n=369 Average	0.08869/9 qualified lineup mean	0.0908	+0.008	-0.001	0 ok
		Home runs	0.033 → 0.031	-0.002	0.0304	0.0215n=369 Very good	0.02879/9 qualified lineup mean	0.0203	-0.033	-0.006	+0.002 ok
RELIEVER Erick Fedde3.5859 BF	Reliever	Hits	0.808 → 0.822	+0.014	0.2169	0.2253n=606 Below average	0.22079/9 qualified lineup mean	0.2292	+0.030	+0.014	+0.000 ok
		Strikeouts	0.610 → 0.510	-0.100	0.2212	0.1701n=606 Poor	0.18689/9 qualified lineup mean	0.1422	-0.183	-0.123	+0.023 ok
		Walks	0.310 → 0.309	-0.001	0.0889	0.0865n=606 Average	0.08869/9 qualified lineup mean	0.0861	-0.009	-0.001	0 ok
		Home runs	0.125 → 0.118	-0.007	0.0304	0.0347n=606 Below average	0.02879/9 qualified lineup mean	0.0328	+0.016	-0.006	-0.001 ok
RELIEVER Jordan Hicks2.3948 BF	Setup Man	Hits	0.912 → 0.928	+0.016	0.2169	0.2054n=216 Good	0.22079/9 qualified lineup mean	0.2091	-0.027	+0.009	-0.000 ok
		Strikeouts	1.234 → 1.053	-0.180	0.2212	0.2760n=216 Good	0.18689/9 qualified lineup mean	0.2356	+0.131	-0.082	-0.014 ok
		Walks	0.599 → 0.596	-0.002	0.0889	0.1258n=216 Poor	0.08869/9 qualified lineup mean	0.1254	+0.088	-0.001	-0.000 ok
		Home runs	0.098 → 0.092	-0.005	0.0304	0.0265n=216 Good	0.02879/9 qualified lineup mean	0.0250	-0.009	-0.004	+0.001 ok
RELIEVER Bryan Hudson2.1591 BF	Reliever	Hits	0.494 → 0.502	+0.009	0.2169	0.2287n=282 Below average	0.22079/9 qualified lineup mean	0.2327	+0.026	+0.008	+0.000 ok
		Strikeouts	0.458 → 0.386	-0.072	0.2212	0.2119n=282 Average	0.18689/9 qualified lineup mean	0.1786	-0.020	-0.074	+0.002 ok
		Walks	0.157 → 0.157	-0.001	0.0889	0.0728n=282 Good	0.08869/9 qualified lineup mean	0.0725	-0.035	-0.001	+0.000 ok
		Home runs	0.055 → 0.052	-0.003	0.0304	0.0255n=282 Good	0.02879/9 qualified lineup mean	0.0240	-0.011	-0.004	+0.001 ok
RELIEVER Grant Taylor1.6248 BF	Reliever	Hits	0.321 → 0.327	+0.006	0.2169	0.1974n=333 Good	0.22079/9 qualified lineup mean	0.2010	-0.032	+0.006	-0.000 ok
		Strikeouts	0.483 → 0.414	-0.069	0.2212	0.2970n=333 Very good	0.18689/9 qualified lineup mean	0.2547	+0.123	-0.056	-0.013 ok
		Walks	0.146 → 0.145	-0.001	0.0889	0.0897n=333 Average	0.08869/9 qualified lineup mean	0.0894	+0.001	-0.001	0 ok
		Home runs	0.043 → 0.041	-0.002	0.0304	0.0264n=333 Good	0.02879/9 qualified lineup mean	0.0249	-0.006	-0.003	+0.000 ok
RELIEVER David Sandlin1.1927 BF	Reliever	Hits	0.258 → 0.263	+0.004	0.2169	0.2164n=167 Average	0.22079/9 qualified lineup mean	0.2202	-0.001	+0.005	0 ok
		Strikeouts	0.279 → 0.236	-0.043	0.2212	0.2336n=167 Average	0.18689/9 qualified lineup mean	0.1977	+0.015	-0.041	-0.002 ok
		Walks	0.099 → 0.099	-0.000	0.0889	0.0832n=167 Average	0.08869/9 qualified lineup mean	0.0829	-0.007	-0.000	0 ok
		Home runs	0.040 → 0.038	-0.002	0.0304	0.0335n=167 Below average	0.02879/9 qualified lineup mean	0.0316	+0.004	-0.002	-0.000 ok
RELIEVER Hagen Smith1.0654 BF	Reliever	Hits	0.231 → 0.235	+0.004	0.2169	0.2169n=0 averagebelow floor	0.22079/9 qualified lineup mean	0.2207	0	+0.004	0 ok
		Strikeouts	0.236 → 0.199	-0.037	0.2212	0.2212n=0 averagebelow floor	0.18689/9 qualified lineup mean	0.1868	0	-0.037	0 ok
		Walks	0.095 → 0.094	-0.000	0.0889	0.0889n=0 averagebelow floor	0.08869/9 qualified lineup mean	0.0886	0	-0.000	0 ok
		Home runs	0.032 → 0.031	-0.002	0.0304	0.0304n=0 averagebelow floor	0.02879/9 qualified lineup mean	0.0287	0	-0.002	0 ok

CLE against Matt Festa (RHP)

The pipeline applied one club-wide ×1.0545 to every CLE hitter line, identical for the leadoff hitter and the ninth bat.
Opposing staff exposure. These hitters face the starter for 20.6793 of 36.393 projected batters (56.8%) and the 2 projected relievers for the remaining 43.2%. Every metric below is the exposure-weighted blend of both matchups; the starter-only figure is shown beside it. Real 2026 game logs put a starter at 21.75 batters faced on average (median 23, p10-p90 17-27) and a whole staff at 37.9: this starter is inside that band and this staff is inside its. The per-player columns are the measured alternative: each subject moved by his own KPI pairing. Exposure for a hitter is his projected plate appearances; for the starter it is batters faced.

Player	Pos	Metric	Published → KPI	Δ	League	Own rate	Opponent rate	Log5	Own skill	Opponent	Remainder
Steven Kwan	CF	Hitswhole staff	1.063 → 1.076	+0.014	0.2169	0.2296n=620 Good	0.2242n=686 Below average	0.2373	+0.056	+0.032	+0.001 ok
		split	starter x1.0334 at 0.568 share (log5 0.2373 vs his 0.2242) + bullpen x0.986 at 0.432 share (log5 0.2264 vs pen 0.2138) = x1.0129 \| starter alone would have been +0.035
		Strikeoutswhole staff	0.441 → 0.398	-0.042	0.2212	0.1013n=620 Very good	0.1837n=686 Below average	0.0820	-0.527	-0.165	+0.080 ok
		split	starter x0.8093 at 0.568 share (log5 0.0820 vs his 0.1837) + bullpen x1.0284 at 0.432 share (log5 0.1042 vs pen 0.2266) = x0.9039 \| starter alone would have been -0.084
		Walkswhole staff	0.517 → 0.501	-0.015	0.0889	0.1280n=620 Very good	0.0818n=686 Average	0.1181	+0.172	-0.031	-0.012 ok
		split	starter x0.9232 at 0.568 share (log5 0.1181 vs his 0.0818) + bullpen x1.0326 at 0.432 share (log5 0.1321 vs pen 0.0920) = x0.9704 \| starter alone would have been -0.040
		Home runswhole staff	0.047 → 0.046	-0.002	0.0304	0.0125n=620 Poor	0.0305n=686 Average	0.0125	-0.079	+0.001	-0.000 ok
		split	starter x1.0046 at 0.568 share (log5 0.0125 vs his 0.0305) + bullpen x0.9156 at 0.432 share (log5 0.0114 vs pen 0.0279) = x0.9662 \| starter alone would have been +0.000
Neutralising Steven Kwan to the qualified-population mean moves the club's KPI run forecast by -0.0915 runs, and the weighted stage by -0.0403. Interactions make these non-additive across players, so they do not sum to the published movement.
José Ramírez	3B	Hitswhole staff	0.983 → 0.996	+0.013	0.2169	0.2095n=531 Below average	0.2242n=686 Below average	0.2167	-0.034	+0.034	-0.001 ok
		split	starter x1.0343 at 0.568 share (log5 0.2167 vs his 0.2242) + bullpen x0.9856 at 0.432 share (log5 0.2065 vs pen 0.2138) = x1.0133 \| starter alone would have been +0.034
		Strikeoutswhole staff	0.572 → 0.519	-0.053	0.2212	0.1376n=531 Very good	0.1837n=686 Below average	0.1122	-0.380	-0.171	+0.055 ok
		split	starter x0.8155 at 0.568 share (log5 0.1122 vs his 0.1837) + bullpen x1.0272 at 0.432 share (log5 0.1414 vs pen 0.2266) = x0.9069 \| starter alone would have been -0.105
		Walkswhole staff	0.533 → 0.517	-0.016	0.0889	0.1086n=531 Good	0.0818n=686 Average	0.1001	+0.089	-0.032	-0.006 ok
		split	starter x0.9216 at 0.568 share (log5 0.1001 vs his 0.0818) + bullpen x1.0333 at 0.432 share (log5 0.1122 vs pen 0.0920) = x0.9698 \| starter alone would have been -0.042
		Home runswhole staff	0.104 → 0.101	-0.004	0.0304	0.0252n=531 Below average	0.0305n=686 Average	0.0253	-0.024	+0.001	-0.000 ok
		split	starter x1.0045 at 0.568 share (log5 0.0253 vs his 0.0305) + bullpen x0.9166 at 0.432 share (log5 0.0231 vs pen 0.0279) = x0.9665 \| starter alone would have been +0.001
Neutralising José Ramírez to the qualified-population mean moves the club's KPI run forecast by 0.3369 runs, and the weighted stage by 0.1482. Interactions make these non-additive across players, so they do not sum to the published movement.
Chase DeLauter	RF	Hitswhole staff	1.069 → 1.082	+0.014	0.2169	0.2364n=543 Very good	0.2242n=686 Below average	0.2443	+0.083	+0.031	+0.002 ok
		split	starter x1.0331 at 0.568 share (log5 0.2443 vs his 0.2242) + bullpen x0.9861 at 0.432 share (log5 0.2332 vs pen 0.2138) = x1.0128 \| starter alone would have been +0.035
		Strikeoutswhole staff	0.655 → 0.595	-0.060	0.2212	0.1534n=543 Good	0.1837n=686 Below average	0.1255	-0.287	-0.159	+0.041 ok
		split	starter x0.8183 at 0.568 share (log5 0.1255 vs his 0.1837) + bullpen x1.0267 at 0.432 share (log5 0.1575 vs pen 0.2266) = x0.9083 \| starter alone would have been -0.119
		Walkswhole staff	0.407 → 0.395	-0.012	0.0889	0.1051n=543 Good	0.0818n=686 Average	0.0968	+0.068	-0.030	-0.005 ok
		split	starter x0.9213 at 0.568 share (log5 0.0968 vs his 0.0818) + bullpen x1.0334 at 0.432 share (log5 0.1086 vs pen 0.0920) = x0.9697 \| starter alone would have been -0.032
		Home runswhole staff	0.112 → 0.108	-0.004	0.0304	0.0298n=543 Average	0.0305n=686 Average	0.0299	-0.003	+0.001	0 ok
		split	starter x1.0045 at 0.568 share (log5 0.0299 vs his 0.0305) + bullpen x0.917 at 0.432 share (log5 0.0273 vs pen 0.0279) = x0.9667 \| starter alone would have been +0.001
Neutralising Chase DeLauter to the qualified-population mean moves the club's KPI run forecast by 0.0139 runs, and the weighted stage by 0.0061. Interactions make these non-additive across players, so they do not sum to the published movement.
Jo Adell	DH	Hitswhole staff	0.974 → 0.987	+0.013	0.2169	0.2185n=656 Average	0.2242n=686 Below average	0.2259	+0.006	+0.030	+0.000 ok
		split	starter x1.0339 at 0.568 share (log5 0.2259 vs his 0.2242) + bullpen x0.9858 at 0.432 share (log5 0.2154 vs pen 0.2138) = x1.0131 \| starter alone would have been +0.033
		Strikeoutswhole staff	0.957 → 0.877	-0.081	0.2212	0.2367n=656 Average	0.1837n=686 Below average	0.1972	+0.064	-0.154	-0.008 ok
		split	starter x0.8332 at 0.568 share (log5 0.1972 vs his 0.1837) + bullpen x1.024 at 0.432 share (log5 0.2423 vs pen 0.2266) = x0.9156 \| starter alone would have been -0.160
		Walkswhole staff	0.271 → 0.262	-0.009	0.0889	0.0575n=656 Poor	0.0818n=686 Average	0.0528	-0.129	-0.029	+0.010 ok
		split	starter x0.9175 at 0.568 share (log5 0.0528 vs his 0.0818) + bullpen x1.0353 at 0.432 share (log5 0.0596 vs pen 0.0920) = x0.9684 \| starter alone would have been -0.022
		Home runswhole staff	0.134 → 0.129	-0.004	0.0304	0.0337n=656 Average	0.0305n=686 Average	0.0339	+0.014	+0.001	+0.000 ok
		split	starter x1.0045 at 0.568 share (log5 0.0339 vs his 0.0305) + bullpen x0.9173 at 0.432 share (log5 0.0309 vs pen 0.0279) = x0.9668 \| starter alone would have been +0.001
Neutralising Jo Adell to the qualified-population mean moves the club's KPI run forecast by -0.0933 runs, and the weighted stage by -0.0411. Interactions make these non-additive across players, so they do not sum to the published movement.
Nathaniel Lowe	1B	Hitswhole staff	0.923 → 0.935	+0.012	0.2169	0.2187n=417 Average	0.2242n=686 Below average	0.2261	+0.007	+0.028	+0.000 ok
		split	starter x1.0339 at 0.568 share (log5 0.2261 vs his 0.2242) + bullpen x0.9858 at 0.432 share (log5 0.2156 vs pen 0.2138) = x1.0131 \| starter alone would have been +0.031
		Strikeoutswhole staff	0.903 → 0.827	-0.077	0.2212	0.2293n=417 Average	0.1837n=686 Below average	0.1907	+0.031	-0.144	-0.004 ok
		split	starter x0.8318 at 0.568 share (log5 0.1907 vs his 0.1837) + bullpen x1.0243 at 0.432 share (log5 0.2348 vs pen 0.2266) = x0.9149 \| starter alone would have been -0.152
		Walkswhole staff	0.321 → 0.311	-0.010	0.0889	0.0906n=417 Average	0.0818n=686 Average	0.0834	+0.007	-0.027	-0.001 ok
		split	starter x0.9201 at 0.568 share (log5 0.0834 vs his 0.0818) + bullpen x1.034 at 0.432 share (log5 0.0937 vs pen 0.0920) = x0.9693 \| starter alone would have been -0.026
		Home runswhole staff	0.123 → 0.119	-0.004	0.0304	0.0353n=417 Good	0.0305n=686 Average	0.0354	+0.019	+0.001	+0.000 ok
		split	starter x1.0045 at 0.568 share (log5 0.0354 vs his 0.0305) + bullpen x0.9174 at 0.432 share (log5 0.0324 vs pen 0.0279) = x0.9669 \| starter alone would have been +0.001
Neutralising Nathaniel Lowe to the qualified-population mean moves the club's KPI run forecast by 0.2210 runs, and the weighted stage by 0.0972. Interactions make these non-additive across players, so they do not sum to the published movement.
Angel Martínez	LF	Hitswhole staff	0.970 → 0.983	+0.013	0.2169	0.2293n=408 Good	0.2242n=686 Below average	0.2369	+0.048	+0.029	+0.001 ok
		split	starter x1.0334 at 0.568 share (log5 0.2369 vs his 0.2242) + bullpen x0.986 at 0.432 share (log5 0.2261 vs pen 0.2138) = x1.0129 \| starter alone would have been +0.032
		Strikeoutswhole staff	0.854 → 0.780	-0.073	0.2212	0.2185n=408 Average	0.1837n=686 Below average	0.1813	-0.011	-0.145	+0.001 ok
		split	starter x0.8299 at 0.568 share (log5 0.1813 vs his 0.1837) + bullpen x1.0246 at 0.432 share (log5 0.2238 vs pen 0.2266) = x0.914 \| starter alone would have been -0.145
		Walkswhole staff	0.201 → 0.194	-0.006	0.0889	0.0429n=408 Poor	0.0818n=686 Average	0.0393	-0.178	-0.028	+0.014 ok
		split	starter x0.9163 at 0.568 share (log5 0.0393 vs his 0.0818) + bullpen x1.0359 at 0.432 share (log5 0.0444 vs pen 0.0920) = x0.9679 \| starter alone would have been -0.017
		Home runswhole staff	0.146 → 0.142	-0.005	0.0304	0.0387n=408 Good	0.0305n=686 Average	0.0389	+0.032	+0.001	+0.000 ok
		split	starter x1.0044 at 0.568 share (log5 0.0389 vs his 0.0305) + bullpen x0.9177 at 0.432 share (log5 0.0355 vs pen 0.0279) = x0.967 \| starter alone would have been +0.001
Neutralising Angel Martínez to the qualified-population mean moves the club's KPI run forecast by 0.1855 runs, and the weighted stage by 0.0816. Interactions make these non-additive across players, so they do not sum to the published movement.
Travis Bazzana	2B	Hitswhole staff	0.868 → 0.879	+0.011	0.2169	0.2129n=506 Average	0.2242n=686 Below average	0.2202	-0.015	+0.028	-0.000 ok
		split	starter x1.0342 at 0.568 share (log5 0.2202 vs his 0.2242) + bullpen x0.9857 at 0.432 share (log5 0.2099 vs pen 0.2138) = x1.0133 \| starter alone would have been +0.030
		Strikeoutswhole staff	0.789 → 0.720	-0.069	0.2212	0.2087n=506 Average	0.1837n=686 Below average	0.1728	-0.047	-0.141	+0.006 ok
		split	starter x0.8281 at 0.568 share (log5 0.1728 vs his 0.1837) + bullpen x1.0249 at 0.432 share (log5 0.2139 vs pen 0.2266) = x0.9131 \| starter alone would have been -0.136
		Walkswhole staff	0.366 → 0.355	-0.011	0.0889	0.1033n=506 Good	0.0818n=686 Average	0.0952	+0.054	-0.027	-0.004 ok
		split	starter x0.9212 at 0.568 share (log5 0.0952 vs his 0.0818) + bullpen x1.0335 at 0.432 share (log5 0.1068 vs pen 0.0920) = x0.9697 \| starter alone would have been -0.029
		Home runswhole staff	0.100 → 0.097	-0.003	0.0304	0.0299n=506 Average	0.0305n=686 Average	0.0300	-0.002	+0.001	0 ok
		split	starter x1.0045 at 0.568 share (log5 0.0300 vs his 0.0305) + bullpen x0.917 at 0.432 share (log5 0.0274 vs pen 0.0279) = x0.9667 \| starter alone would have been +0.000
Neutralising Travis Bazzana to the qualified-population mean moves the club's KPI run forecast by 0.1652 runs, and the weighted stage by 0.0727. Interactions make these non-additive across players, so they do not sum to the published movement.
Patrick Bailey	C	Hitswhole staff	0.821 → 0.832	+0.011	0.2169	0.2062n=333 Below average	0.2242n=686 Below average	0.2133	-0.039	+0.027	-0.001 ok
		split	starter x1.0345 at 0.568 share (log5 0.2133 vs his 0.2242) + bullpen x0.9856 at 0.432 share (log5 0.2033 vs pen 0.2138) = x1.0134 \| starter alone would have been +0.028
		Strikeoutswhole staff	0.924 → 0.849	-0.075	0.2212	0.2666n=333 Below average	0.1837n=686 Below average	0.2236	+0.167	-0.138	-0.020 ok
		split	starter x0.8387 at 0.568 share (log5 0.2236 vs his 0.1837) + bullpen x1.0231 at 0.432 share (log5 0.2728 vs pen 0.2266) = x0.9183 \| starter alone would have been -0.149
		Walkswhole staff	0.330 → 0.319	-0.010	0.0889	0.0809n=333 Average	0.0818n=686 Average	0.0744	-0.029	-0.026	+0.002 ok
		split	starter x0.9194 at 0.568 share (log5 0.0744 vs his 0.0818) + bullpen x1.0344 at 0.432 share (log5 0.0837 vs pen 0.0920) = x0.9691 \| starter alone would have been -0.027
		Home runswhole staff	0.106 → 0.103	-0.004	0.0304	0.0302n=333 Average	0.0305n=686 Average	0.0303	-0.001	+0.001	0 ok
		split	starter x1.0045 at 0.568 share (log5 0.0303 vs his 0.0305) + bullpen x0.917 at 0.432 share (log5 0.0277 vs pen 0.0279) = x0.9667 \| starter alone would have been +0.001
Neutralising Patrick Bailey to the qualified-population mean moves the club's KPI run forecast by -0.0193 runs, and the weighted stage by -0.0085. Interactions make these non-additive across players, so they do not sum to the published movement.
Brayan Rocchio	SS	Hitswhole staff	0.831 → 0.842	+0.011	0.2169	0.2238n=576 Average	0.2242n=686 Below average	0.2314	+0.024	+0.025	+0.001 ok
		split	starter x1.0337 at 0.568 share (log5 0.2314 vs his 0.2242) + bullpen x0.9859 at 0.432 share (log5 0.2207 vs pen 0.2138) = x1.0131 \| starter alone would have been +0.028
		Strikeoutswhole staff	0.509 → 0.462	-0.047	0.2212	0.1501n=576 Good	0.1837n=686 Below average	0.1227	-0.246	-0.130	+0.035 ok
		split	starter x0.8177 at 0.568 share (log5 0.1227 vs his 0.1837) + bullpen x1.0268 at 0.432 share (log5 0.1541 vs pen 0.2266) = x0.908 \| starter alone would have been -0.093
		Walkswhole staff	0.269 → 0.261	-0.008	0.0889	0.0699n=576 Below average	0.0818n=686 Average	0.0642	-0.066	-0.025	+0.005 ok
		split	starter x0.9185 at 0.568 share (log5 0.0642 vs his 0.0818) + bullpen x1.0348 at 0.432 share (log5 0.0723 vs pen 0.0920) = x0.9687 \| starter alone would have been -0.022
		Home runswhole staff	0.083 → 0.080	-0.003	0.0304	0.0250n=576 Below average	0.0305n=686 Average	0.0251	-0.018	+0.001	-0.000 ok
		split	starter x1.0045 at 0.568 share (log5 0.0251 vs his 0.0305) + bullpen x0.9166 at 0.432 share (log5 0.0229 vs pen 0.0279) = x0.9665 \| starter alone would have been +0.000
Neutralising Brayan Rocchio to the qualified-population mean moves the club's KPI run forecast by 0.1767 runs, and the weighted stage by 0.0778. Interactions make these non-additive across players, so they do not sum to the published movement.
STARTER Gavin Williams23.1868 BF	Starter	Hits	3.926 → 3.880	-0.046	0.2169	0.1978n=721 Good	0.21448/9 qualified lineup mean	0.1955	-0.442	-0.058	+0.004 ok
		Strikeouts	8.606 → 8.931	+0.324	0.2212	0.3179n=721 Very good	0.23088/9 qualified lineup mean	0.3299	+2.243	+0.222	+0.056 ok
		Walks	2.254 → 2.568	+0.314	0.0889	0.0854n=721 Average	0.10138/9 qualified lineup mean	0.0973	-0.082	+0.286	-0.010 ok
		Home runs	0.460 → 0.504	+0.044	0.0304	0.0330n=721 Below average	0.03338/9 qualified lineup mean	0.0362	+0.061	+0.067	+0.006 ok
RELIEVER Tim Herrin2.7015 BF	Middle Reliever	Hits	0.868 → 0.858	-0.010	0.2169	0.2029n=237 Good	0.21448/9 qualified lineup mean	0.2006	-0.038	-0.007	+0.000 ok
		Strikeouts	0.974 → 1.015	+0.042	0.2212	0.2262n=237 Average	0.23088/9 qualified lineup mean	0.2359	+0.013	+0.026	+0.000 ok
		Walks	0.493 → 0.560	+0.067	0.0889	0.1065n=237 Below average	0.10138/9 qualified lineup mean	0.1209	+0.047	+0.033	+0.006 ok
		Home runs	0.070 → 0.077	+0.007	0.0304	0.0272n=237 Average	0.03338/9 qualified lineup mean	0.0298	-0.009	+0.008	-0.001 ok
RELIEVER Erik Sabrowski2.585 BF	Middle Reliever (High Leverage)	Hits	0.472 → 0.466	-0.006	0.2169	0.1819n=194 Very good	0.21448/9 qualified lineup mean	0.1797	-0.090	-0.006	+0.001 ok
		Strikeouts	1.588 → 1.646	+0.058	0.2212	0.3346n=194 Very good	0.23088/9 qualified lineup mean	0.3469	+0.293	+0.025	+0.007 ok
		Walks	0.662 → 0.750	+0.087	0.0889	0.1280n=194 Poor	0.10138/9 qualified lineup mean	0.1449	+0.101	+0.032	+0.012 ok
		Home runs	0.046 → 0.050	+0.004	0.0304	0.0259n=194 Good	0.03338/9 qualified lineup mean	0.0284	-0.012	+0.007	-0.001 ok
RELIEVER Daniel Espino2.5637 BF	Middle Reliever (High Leverage)	Hits	0.978 → 0.967	-0.011	0.2169	0.2169n=0 averagebelow floor	0.21448/9 qualified lineup mean	0.2144	0	-0.006	0 ok
		Strikeouts	1.634 → 1.705	+0.071	0.2212	0.2212n=0 averagebelow floor	0.23088/9 qualified lineup mean	0.2308	0	+0.025	0 ok
		Walks	0.396 → 0.451	+0.055	0.0889	0.0889n=0 averagebelow floor	0.10138/9 qualified lineup mean	0.1013	0	+0.032	0 ok
		Home runs	0.182 → 0.199	+0.017	0.0304	0.0304n=0 averagebelow floor	0.03338/9 qualified lineup mean	0.0333	0	+0.007	0 ok
RELIEVER Cade Smith2.5125 BF	Reliever	Hits	0.480 → 0.474	-0.006	0.2169	0.1909n=281 Very good	0.21448/9 qualified lineup mean	0.1886	-0.065	-0.006	+0.001 ok
		Strikeouts	0.833 → 0.864	+0.031	0.2212	0.3315n=281 Very good	0.23088/9 qualified lineup mean	0.3437	+0.277	+0.024	+0.007 ok
		Walks	0.195 → 0.222	+0.027	0.0889	0.0776n=281 Good	0.10138/9 qualified lineup mean	0.0885	-0.029	+0.031	-0.004 ok
		Home runs	0.064 → 0.070	+0.006	0.0304	0.0255n=281 Good	0.03338/9 qualified lineup mean	0.0280	-0.012	+0.007	-0.001 ok
RELIEVER Hunter Gaddis2.4898 BF	Reliever	Hits	0.532 → 0.526	-0.006	0.2169	0.2138n=252 Average	0.21448/9 qualified lineup mean	0.2114	-0.007	-0.006	+0.000 ok
		Strikeouts	0.553 → 0.577	+0.024	0.2212	0.2220n=252 Average	0.23088/9 qualified lineup mean	0.2316	+0.002	+0.024	+0.000 ok
		Walks	0.219 → 0.249	+0.030	0.0889	0.0879n=252 Average	0.10138/9 qualified lineup mean	0.1001	-0.003	+0.031	-0.000 ok
		Home runs	0.073 → 0.080	+0.007	0.0304	0.0294n=252 Average	0.03338/9 qualified lineup mean	0.0323	-0.002	+0.007	-0.000 ok
RELIEVER Shawn Armstrong0.7248 BF	Reliever	Hits	0.150 → 0.149	-0.002	0.2169	0.2076n=148 Average	0.21448/9 qualified lineup mean	0.2052	-0.007	-0.002	+0.000 ok
		Strikeouts	0.158 → 0.165	+0.007	0.2212	0.2180n=148 Average	0.23088/9 qualified lineup mean	0.2274	-0.002	+0.007	-0.000 ok
		Walks	0.076 → 0.086	+0.010	0.0889	0.1048n=148 Below average	0.10138/9 qualified lineup mean	0.1190	+0.011	+0.009	+0.001 ok
		Home runs	0.020 → 0.022	+0.002	0.0304	0.0279n=148 Average	0.03338/9 qualified lineup mean	0.0305	-0.002	+0.002	-0.000 ok
RELIEVER Matt Festa0.6338 BF	Reliever	Hits	0.139 → 0.138	-0.002	0.2169	0.2199n=278 Average	0.21448/9 qualified lineup mean	0.2173	+0.002	-0.002	0 ok
		Strikeouts	0.146 → 0.153	+0.006	0.2212	0.2310n=278 Average	0.23088/9 qualified lineup mean	0.2408	+0.006	+0.006	+0.000 ok
		Walks	0.051 → 0.058	+0.007	0.0889	0.0804n=278 Average	0.10138/9 qualified lineup mean	0.0917	-0.005	+0.008	-0.001 ok
		Home runs	0.017 → 0.019	+0.002	0.0304	0.0270n=278 Good	0.03338/9 qualified lineup mean	0.0296	-0.002	+0.002	-0.000 ok

Published pitching, integer draw beside the continuous model line

Pitcher	Role	BF	IP	expected IP	H int/cont	SO int/cont	BB int/cont	ER int/cont	Rate source
Anthony Kay	Starter	20.7	4.2	4.680	3 / 5.084	3 / 3.531	2 / 1.555	1 / 2.568	all_players_stats_recent_season_blend
Sean Newcomb	Middle Reliever (High Leverage)	8.5	2.0	2.006	3 / 1.889	2 / 1.485	2 / 0.562	3 / 0.819	all_players_stats_recent_season_blend
Jordan Hicks	Setup Man	4.5	1.1	1.003	1 / 0.912	2 / 1.234	0 / 0.599	0 / 0.488	all_players_stats_recent_season_blend
Gavin Williams	Starter	23.2	5.2	5.669	6 / 3.926	7 / 8.606	3 / 2.254	4 / 1.922	all_players_stats_recent_season_blend
Tim Herrin	Middle Reliever	4.4	1.0	1.000	0 / 0.868	1 / 0.974	0 / 0.493	0 / 0.312	all_players_stats_recent_season_blend
Daniel Espino	Middle Reliever (High Leverage)	4.4	1.0	1.000	1 / 0.978	1 / 1.634	0 / 0.396	0 / 0.320	all_players_stats_recent_season_blend
Erik Sabrowski	Middle Reliever (High Leverage)	4.1	1.1	1.000	0 / 0.472	1 / 1.588	0 / 0.662	0 / 0.262	all_players_stats_recent_season_blend

Audit: NO DEFECT FOUND

Invariant	Verdict	Evidence
blend arithmetic away_runs	PASS	4.0000 + 0.78 x (3.6357 - 4.0000) = 3.7159, recorded 3.7159
blend arithmetic home_runs	PASS	5.0000 + 0.44 x (5.6188 - 5.0000) = 5.2723, recorded 5.2723
blend arithmetic home_win	PASS	0.5256 + 0.00 x (0.5791 - 0.5256) = 0.5256, recorded 0.5256
published rounding faithful to the blend	PASS	blended 3.7159-5.2723 rounds to 4-5, layer published 4-5
published probability on the score's side	PASS	4-5 with p(home) 0.525603
Log5 additivity136 checks	PASS	136 of 136 pass
away club fields exactly one starter	PASS	1 starters in a staff of 3
away staff totals a regulation workload	PASS	24 outs recorded, 24 expected
away starter carries a continuous line	PASS	expected_IP 4.68
away starter batters faced inside the real band	PASS	Anthony Kay projected 20.7 batters faced; real starters run p01-p99 4-30, median 23, mean 21.75
away reliever batters faced inside the real band2 checks	PASS	2 of 2 pass
home club fields exactly one starter	PASS	1 starters in a staff of 4
home staff totals a regulation workload	PASS	27 outs recorded, 27 expected
home starter carries a continuous line	PASS	expected_IP 5.669
home starter batters faced inside the real band	PASS	Gavin Williams projected 23.2 batters faced; real starters run p01-p99 4-30, median 23, mean 21.75
home reliever batters faced inside the real band3 checks	PASS	3 of 3 pass
away hitters satisfy every coherence rule	PASS	9 hitters checked; H<=AB, HR<=H, 2B+3B+HR<=H, SO<=AB, continuous line present
home hitters satisfy every coherence rule	PASS	9 hitters checked; H<=AB, HR<=H, 2B+3B+HR<=H, SO<=AB, continuous line present

20261005_NYY@TB — published 4-2, 157 checks, all pass

Artifact. 20261005_NYY@TB__20261006T023852Z.json, phase post_start_rerun, generated 2026-10-06T02:38:52, first pitch 2026-10-06T00:00:00.
Published 4 - 2 with p(home) 0.460388. Simulation draw chosen on not recorded; ? of ? draws realized the target exactly.

Stage chain: every stage that touched the score

Stage	Score after	Δ away	Δ home	Cum. away	Cum. home	Changed target	Own adjustment and description
1. `historical_ml`	4.864972693493641 - 2.9968524001566985	0	0	0	0	no	Base ML historical model prediction
2. `variance_applied`	5 - 3	+0.1350	+0.0031	+0.1350	+0.0031	yes	Deterministic baseline target (variance factor 1.0)
3. `enhanced_prediction`	4 - 3	-1.0000	0	-0.8650	+0.0031	yes	Enhanced ML with recent form, H2H, low-score analysis (confidence shrunk by matchup_recency_guardrail)
4. `kpi_layer`	4 - 3	0	0	-0.8650	+0.0031	no	away +0.3329; home +0.0510 Savant KPI blend applied to the model target
5. `embedded_savant_run_environment`	4 - 2	0	-1.0000	-0.8650	-0.9969	yes	away +0.2682; home -0.7020 Target score adjusted by temporal Savant lineup quality versus opposing starter quality
6. `fundamental_edge_target_review`	4 - 2	0	0	-0.8650	-0.9969	no	Refreshed score target and win probability after Savant/pitcher/team-strength adjustments
7. `capped_target`	4 - 2	0	0	-0.8650	-0.9969	no	Capped at max 12 realistic runs
8. `simulation_result`	4 - 2	-	-	-	-	no	Final inning-by-inning simulation result

KPI team layer: the blend, with its arithmetic

Target	Baseline	KPI only	Weight	Blended	Computation	Verdict
`away_runs`	4.0000	4.4268	0.78	4.3329	4.0000 + 0.78 × (4.4268 − 4.0000) = 4.3329	PASS
`home_runs`	3.0000	3.1160	0.44	3.0510	3.0000 + 0.44 × (3.1160 − 3.0000) = 3.0510	PASS
`home_win`	0.4534	0.6851	0.00	0.4534	0.4534 + 0.00 × (0.6851 − 0.4534) = 0.4534	PASS

Published by the layer: away 4, home 3, p(home) 0.453398, continuous away 4.3329 / home 3.0510

Inputs and gates. candidate 20261005T125039Z_f5114dc0, value age 8 days, lineup coverage away 0.9064 / home 0.9942, movement cap 2.5 runs.
values_present ok values_fresh ok one_source_bundle ok lineup_coverage ok target_eligible ok movement_within_limit ok published_movement_within_limit ok

What drove the KPI forecast for `away_runs`

intercept 4.2573 + sum of contributions = 4.4268; prediction 4.4268; contributions outside the top listed -0.5952 runs

Feature	Raw	Standardised	Runs contributed	Which KPIs it is built from
`home_off_power_spread`	1.5797	3.423	+2.2318	offence group (dispersion across the lineup, not its mean): batter_home_run_rate, batter_bat_speed
`home_platoon_whiff`	0.8852	2.933	-1.2609	platoon term, selected by the opposing starter's throwing hand
`home_x_fastball`	0.9223	4.653	-1.1913	batter x pitcher interaction: pitcher_whiff_rate_fastball, batter_whiff_rate_vs_fastball
`away_off_contact`	0.5754	2.335	+0.8940	offence group: batter_barrel_rate_vs_fastball, batter_barrel_rate_vs_breaking, batter_barrel_rate_vs_offspeed, batter_hard_hit_rate_vs_fastball ...
`home_off_qualified_share`	0.9942	2.043	-0.7557	share of the lineup carrying a qualifying sample, not a skill measurement
`home_off_discipline`	0.7576	3.296	+0.6733	offence group: batter_whiff_rate_vs_fastball, batter_whiff_rate_vs_breaking, batter_whiff_rate_vs_offspeed, batter_strikeout_rate ...
`home_x_breaking`	1.0853	6.021	+0.5849	batter x pitcher interaction: pitcher_whiff_rate_breaking, batter_whiff_rate_vs_breaking
`home_off_ground_spread`	1.2847	2.859	+0.5180	offence group (dispersion across the lineup, not its mean): batter_ground_ball_rate_vs_slider, batter_ground_ball_rate_vs_fastball
`away_off_worst3_discipline`	-1.0390	-2.018	-0.4701	mean of the three weakest plate approaches
`home_sp_velocity`	1.5794	2.236	-0.4594	starter group: pitcher_release_speed

What drove the KPI forecast for `home_runs`

intercept 4.2997 + sum of contributions = 3.1160; prediction 3.1160; contributions outside the top listed -0.3202 runs

Feature	Raw	Standardised	Runs contributed	Which KPIs it is built from
`home_off_ground_spread`	1.2847	2.859	+0.8403	offence group (dispersion across the lineup, not its mean): batter_ground_ball_rate_vs_slider, batter_ground_ball_rate_vs_fastball
`home_x_fastball`	0.9223	4.653	-0.7820	batter x pitcher interaction: pitcher_whiff_rate_fastball, batter_whiff_rate_vs_fastball
`home_off_discipline`	0.7576	3.296	-0.7751	offence group: batter_whiff_rate_vs_fastball, batter_whiff_rate_vs_breaking, batter_whiff_rate_vs_offspeed, batter_strikeout_rate ...
`home_sp_velocity`	1.5794	2.236	+0.6970	starter group: pitcher_release_speed
`home_off_worst3_discipline`	0.0206	2.138	+0.6437	mean of the three weakest plate approaches
`home_pen_minus_sp`	-0.4227	-2.914	-0.6244	composite feature
`home_off_contact_spread`	1.1519	2.556	+0.4602	offence group (dispersion across the lineup, not its mean): batter_barrel_rate_vs_fastball, batter_barrel_rate_vs_breaking, batter_barrel_rate_vs_offspeed, batter_hard_hit_rate_vs_fastball ...
`away_off_top3_contact`	1.3995	2.117	-0.4599	mean of the three best contact bats, a tail a club mean cannot express
`home_off_power_spread`	1.5797	3.423	-0.4359	offence group (dispersion across the lineup, not its mean): batter_home_run_rate, batter_bat_speed
`home_sp_command`	1.0229	1.761	-0.4276	starter group: pitcher_strikeout_rate, pitcher_walk_rate

Per-player metric movement

NYY against Ryan Yarbrough (LHP)

The pipeline applied one club-wide ×1.0832 to every NYY hitter line, identical for the leadoff hitter and the ninth bat.
Opposing staff exposure. These hitters face the starter for 21.6754 of 35.458 projected batters (61.1%) and the 3 projected relievers for the remaining 38.9%. Every metric below is the exposure-weighted blend of both matchups; the starter-only figure is shown beside it. Real 2026 game logs put a starter at 21.75 batters faced on average (median 23, p10-p90 17-27) and a whole staff at 37.9: this starter is inside that band and this staff is inside its. The per-player columns are the measured alternative: each subject moved by his own KPI pairing. Exposure for a hitter is his projected plate appearances; for the starter it is batters faced.

Player	Pos	Metric	Published → KPI	Δ	League	Own rate	Opponent rate	Log5	Own skill	Opponent	Remainder
Ben Rice	DH	Hitswhole staff	1.019 → 1.002	-0.017	0.2169	0.2240n=651 Average	0.2195n=730 Average	0.2267	+0.031	+0.011	+0.000 ok
		split	starter x1.012 at 0.611 share (log5 0.2267 vs his 0.2195) + bullpen x0.9379 at 0.389 share (log5 0.2101 vs pen 0.2033) = x0.9832 \| starter alone would have been +0.012
		Strikeoutswhole staff	0.946 → 0.952	+0.006	0.2212	0.2398n=651 Average	0.2144n=730 Average	0.2326	+0.080	-0.029	-0.002 ok
		split	starter x0.9701 at 0.611 share (log5 0.2326 vs his 0.2144) + bullpen x1.0624 at 0.389 share (log5 0.2547 vs pen 0.2354) = x1.006 \| starter alone would have been -0.028
		Walkswhole staff	0.499 → 0.467	-0.032	0.0889	0.1228n=651 Very good	0.0868n=730 Average	0.1200	+0.145	-0.009	-0.003 ok
		split	starter x0.9766 at 0.611 share (log5 0.1200 vs his 0.0868) + bullpen x0.8704 at 0.389 share (log5 0.1069 vs pen 0.0770) = x0.9353 \| starter alone would have been -0.012
		Home runswhole staff	0.275 → 0.279	+0.003	0.0304	0.0536n=651 Very good	0.0311n=730 Average	0.0548	+0.099	+0.003	+0.002 ok
		split	starter x1.0223 at 0.611 share (log5 0.0548 vs his 0.0311) + bullpen x0.9966 at 0.389 share (log5 0.0534 vs pen 0.0303) = x1.0123 \| starter alone would have been +0.006
Neutralising Ben Rice to the qualified-population mean moves the club's KPI run forecast by 0.1070 runs, and the weighted stage by 0.0835. Interactions make these non-additive across players, so they do not sum to the published movement.
Cody Bellinger	LF	Hitswhole staff	1.087 → 1.069	-0.018	0.2169	0.2269n=549 Good	0.2195n=730 Average	0.2296	+0.044	+0.011	+0.000 ok
		split	starter x1.0119 at 0.611 share (log5 0.2296 vs his 0.2195) + bullpen x0.9381 at 0.389 share (log5 0.2129 vs pen 0.2033) = x0.9832 \| starter alone would have been +0.013
		Strikeoutswhole staff	0.704 → 0.709	+0.005	0.2212	0.1619n=549 Good	0.2144n=730 Average	0.1566	-0.260	-0.030	+0.006 ok
		split	starter x0.9672 at 0.611 share (log5 0.1566 vs his 0.2144) + bullpen x1.0693 at 0.389 share (log5 0.1731 vs pen 0.2354) = x1.0069 \| starter alone would have been -0.023
		Walkswhole staff	0.423 → 0.395	-0.028	0.0889	0.1027n=549 Good	0.0868n=730 Average	0.1002	+0.060	-0.009	-0.001 ok
		split	starter x0.9761 at 0.611 share (log5 0.1002 vs his 0.0868) + bullpen x0.8678 at 0.389 share (log5 0.0891 vs pen 0.0770) = x0.934 \| starter alone would have been -0.010
		Home runswhole staff	0.164 → 0.166	+0.002	0.0304	0.0295n=549 Average	0.0311n=730 Average	0.0302	-0.004	+0.003	-0.000 ok
		split	starter x1.0229 at 0.611 share (log5 0.0302 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0294 vs pen 0.0303) = x1.0126 \| starter alone would have been +0.004
Neutralising Cody Bellinger to the qualified-population mean moves the club's KPI run forecast by 0.6268 runs, and the weighted stage by 0.4889. Interactions make these non-additive across players, so they do not sum to the published movement.
Luis García Jr.	1B	Hitswhole staff	1.092 → 1.074	-0.018	0.2169	0.2435n=539 Very good	0.2195n=730 Average	0.2464	+0.109	+0.011	+0.001 ok
		split	starter x1.0117 at 0.611 share (log5 0.2464 vs his 0.2195) + bullpen x0.9393 at 0.389 share (log5 0.2288 vs pen 0.2033) = x0.9836 \| starter alone would have been +0.013
		Strikeoutswhole staff	0.760 → 0.765	+0.005	0.2212	0.1817n=539 Good	0.2144n=730 Average	0.1758	-0.162	-0.028	+0.004 ok
		split	starter x0.9679 at 0.611 share (log5 0.1758 vs his 0.2144) + bullpen x1.0675 at 0.389 share (log5 0.1939 vs pen 0.2354) = x1.0066 \| starter alone would have been -0.024
		Walkswhole staff	0.219 → 0.204	-0.015	0.0889	0.0541n=539 Poor	0.0868n=730 Average	0.0527	-0.142	-0.009	+0.003 ok
		split	starter x0.9749 at 0.611 share (log5 0.0527 vs his 0.0868) + bullpen x0.8616 at 0.389 share (log5 0.0466 vs pen 0.0770) = x0.9309 \| starter alone would have been -0.005
		Home runswhole staff	0.215 → 0.218	+0.003	0.0304	0.0473n=539 Very good	0.0311n=730 Average	0.0484	+0.069	+0.003	+0.002 ok
		split	starter x1.0224 at 0.611 share (log5 0.0484 vs his 0.0311) + bullpen x0.9966 at 0.389 share (log5 0.0472 vs pen 0.0303) = x1.0124 \| starter alone would have been +0.005
Neutralising Luis García Jr. to the qualified-population mean moves the club's KPI run forecast by 0.4351 runs, and the weighted stage by 0.3393. Interactions make these non-additive across players, so they do not sum to the published movement.
Spencer Jones	RF	Hitswhole staff	0.960 → 0.944	-0.016	0.2169	0.2177n=283 Average	0.2195n=730 Average	0.2204	+0.004	+0.011	0 ok
		split	starter x1.0121 at 0.611 share (log5 0.2204 vs his 0.2195) + bullpen x0.9374 at 0.389 share (log5 0.2041 vs pen 0.2033) = x0.9831 \| starter alone would have been +0.012
		Strikeoutswhole staff	1.384 → 1.391	+0.007	0.2212	0.3460n=283 Poor	0.2144n=730 Average	0.3371	+0.512	-0.028	-0.009 ok
		split	starter x0.9742 at 0.611 share (log5 0.3371 vs his 0.2144) + bullpen x1.0532 at 0.389 share (log5 0.3644 vs pen 0.2354) = x1.0049 \| starter alone would have been -0.036
		Walkswhole staff	0.342 → 0.319	-0.023	0.0889	0.0885n=283 Average	0.0868n=730 Average	0.0864	-0.002	-0.009	0 ok
		split	starter x0.9757 at 0.611 share (log5 0.0864 vs his 0.0868) + bullpen x0.866 at 0.389 share (log5 0.0767 vs pen 0.0770) = x0.9331 \| starter alone would have been -0.008
		Home runswhole staff	0.159 → 0.161	+0.002	0.0304	0.0348n=283 Average	0.0311n=730 Average	0.0356	+0.018	+0.003	+0.000 ok
		split	starter x1.0227 at 0.611 share (log5 0.0356 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0347 vs pen 0.0303) = x1.0125 \| starter alone would have been +0.004
Neutralising Spencer Jones to the qualified-population mean moves the club's KPI run forecast by -0.2804 runs, and the weighted stage by -0.2187. Interactions make these non-additive across players, so they do not sum to the published movement.
George Lombard Jr.	SS	Hitswhole staff	0.973 → 0.957	-0.017	0.2169	0.2169n=0 averagebelow floor	0.2195n=730 Average	0.2195	0	+0.011	0 ok
		split	starter x1.0121 at 0.611 share (log5 0.2195 vs his 0.2195) + bullpen x0.9373 at 0.389 share (log5 0.2033 vs pen 0.2033) = x0.983 \| starter alone would have been +0.012
		Strikeoutswhole staff	1.028 → 1.034	+0.006	0.2212	0.2212n=0 averagebelow floor	0.2144n=730 Average	0.2144	0	-0.027	0 ok
		split	starter x0.9694 at 0.611 share (log5 0.2144 vs his 0.2144) + bullpen x1.0641 at 0.389 share (log5 0.2354 vs pen 0.2354) = x1.0062 \| starter alone would have been -0.031
		Walkswhole staff	0.262 → 0.245	-0.018	0.0889	0.0889n=0 averagebelow floor	0.0868n=730 Average	0.0868	0	-0.009	0 ok
		split	starter x0.9758 at 0.611 share (log5 0.0868 vs his 0.0868) + bullpen x0.866 at 0.389 share (log5 0.0770 vs pen 0.0770) = x0.9331 \| starter alone would have been -0.006
		Home runswhole staff	0.158 → 0.160	+0.002	0.0304	0.0304n=0 averagebelow floor	0.0311n=730 Average	0.0311	0	+0.003	0 ok
		split	starter x1.0228 at 0.611 share (log5 0.0311 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0303 vs pen 0.0303) = x1.0126 \| starter alone would have been +0.004
Neutralising George Lombard Jr. to the qualified-population mean moves the club's KPI run forecast by -0.0767 runs, and the weighted stage by -0.0598. Interactions make these non-additive across players, so they do not sum to the published movement.
Jazz Chisholm Jr.	2B	Hitswhole staff	0.817 → 0.802	-0.014	0.2169	0.2095n=526 Below average	0.2195n=730 Average	0.2121	-0.027	+0.010	-0.000 ok
		split	starter x1.0122 at 0.611 share (log5 0.2121 vs his 0.2195) + bullpen x0.9368 at 0.389 share (log5 0.1963 vs pen 0.2033) = x0.9829 \| starter alone would have been +0.010
		Strikeoutswhole staff	1.039 → 1.044	+0.006	0.2212	0.2902n=526 Below average	0.2144n=730 Average	0.2821	+0.258	-0.025	-0.005 ok
		split	starter x0.9721 at 0.611 share (log5 0.2821 vs his 0.2144) + bullpen x1.0581 at 0.389 share (log5 0.3070 vs pen 0.2354) = x1.0055 \| starter alone would have been -0.029
		Walkswhole staff	0.363 → 0.339	-0.024	0.0889	0.1032n=526 Good	0.0868n=730 Average	0.1008	+0.053	-0.008	-0.001 ok
		split	starter x0.9761 at 0.611 share (log5 0.1008 vs his 0.0868) + bullpen x0.8679 at 0.389 share (log5 0.0896 vs pen 0.0770) = x0.934 \| starter alone would have been -0.009
		Home runswhole staff	0.153 → 0.155	+0.002	0.0304	0.0342n=526 Average	0.0311n=730 Average	0.0350	+0.014	+0.003	+0.000 ok
		split	starter x1.0227 at 0.611 share (log5 0.0350 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0341 vs pen 0.0303) = x1.0125 \| starter alone would have been +0.004
Neutralising Jazz Chisholm Jr. to the qualified-population mean moves the club's KPI run forecast by 0.0937 runs, and the weighted stage by 0.0731. Interactions make these non-additive across players, so they do not sum to the published movement.
Trent Grisham	CF	Hitswhole staff	0.791 → 0.778	-0.014	0.2169	0.1996n=474 Poor	0.2195n=730 Average	0.2021	-0.066	+0.010	-0.001 ok
		split	starter x1.0124 at 0.611 share (log5 0.2021 vs his 0.2195) + bullpen x0.936 at 0.389 share (log5 0.1868 vs pen 0.2033) = x0.9827 \| starter alone would have been +0.010
		Strikeoutswhole staff	0.722 → 0.726	+0.005	0.2212	0.1871n=474 Good	0.2144n=730 Average	0.1811	-0.131	-0.026	+0.003 ok
		split	starter x0.9681 at 0.611 share (log5 0.1811 vs his 0.2144) + bullpen x1.0671 at 0.389 share (log5 0.1996 vs pen 0.2354) = x1.0066 \| starter alone would have been -0.023
		Walkswhole staff	0.416 → 0.389	-0.027	0.0889	0.1190n=474 Very good	0.0868n=730 Average	0.1162	+0.115	-0.008	-0.002 ok
		split	starter x0.9765 at 0.611 share (log5 0.1162 vs his 0.0868) + bullpen x0.8699 at 0.389 share (log5 0.1035 vs pen 0.0770) = x0.9351 \| starter alone would have been -0.010
		Home runswhole staff	0.136 → 0.137	+0.002	0.0304	0.0339n=474 Average	0.0311n=730 Average	0.0347	+0.013	+0.003	+0.000 ok
		split	starter x1.0228 at 0.611 share (log5 0.0347 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0338 vs pen 0.0303) = x1.0126 \| starter alone would have been +0.003
Neutralising Trent Grisham to the qualified-population mean moves the club's KPI run forecast by 0.5213 runs, and the weighted stage by 0.4066. Interactions make these non-additive across players, so they do not sum to the published movement.
Austin Wells	C	Hitswhole staff	0.674 → 0.662	-0.012	0.2169	0.1904n=387 Poor	0.2195n=730 Average	0.1927	-0.087	+0.009	-0.001 ok
		split	starter x1.0125 at 0.611 share (log5 0.1927 vs his 0.2195) + bullpen x0.9353 at 0.389 share (log5 0.1780 vs pen 0.2033) = x0.9825 \| starter alone would have been +0.008
		Strikeoutswhole staff	0.790 → 0.794	+0.005	0.2212	0.2588n=387 Below average	0.2144n=730 Average	0.2513	+0.123	-0.022	-0.003 ok
		split	starter x0.9709 at 0.611 share (log5 0.2513 vs his 0.2144) + bullpen x1.0608 at 0.389 share (log5 0.2746 vs pen 0.2354) = x1.0058 \| starter alone would have been -0.023
		Walkswhole staff	0.348 → 0.325	-0.023	0.0889	0.1158n=387 Good	0.0868n=730 Average	0.1130	+0.088	-0.007	-0.002 ok
		split	starter x0.9765 at 0.611 share (log5 0.1130 vs his 0.0868) + bullpen x0.8695 at 0.389 share (log5 0.1007 vs pen 0.0770) = x0.9349 \| starter alone would have been -0.008
		Home runswhole staff	0.121 → 0.123	+0.002	0.0304	0.0338n=387 Average	0.0311n=730 Average	0.0346	+0.011	+0.002	+0.000 ok
		split	starter x1.0228 at 0.611 share (log5 0.0346 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0337 vs pen 0.0303) = x1.0126 \| starter alone would have been +0.003
Neutralising Austin Wells to the qualified-population mean moves the club's KPI run forecast by 0.1437 runs, and the weighted stage by 0.1121. Interactions make these non-additive across players, so they do not sum to the published movement.
Ryan McMahon	3B	Hitswhole staff	0.701 → 0.689	-0.012	0.2169	0.2083n=354 Below average	0.2195n=730 Average	0.2109	-0.028	+0.009	-0.000 ok
		split	starter x1.0122 at 0.611 share (log5 0.2109 vs his 0.2195) + bullpen x0.9367 at 0.389 share (log5 0.1951 vs pen 0.2033) = x0.9829 \| starter alone would have been +0.009
		Strikeoutswhole staff	0.908 → 0.913	+0.005	0.2212	0.2850n=354 Below average	0.2144n=730 Average	0.2770	+0.211	-0.022	-0.004 ok
		split	starter x0.9719 at 0.611 share (log5 0.2770 vs his 0.2144) + bullpen x1.0585 at 0.389 share (log5 0.3017 vs pen 0.2354) = x1.0056 \| starter alone would have been -0.026
		Walkswhole staff	0.262 → 0.244	-0.018	0.0889	0.0816n=354 Average	0.0868n=730 Average	0.0796	-0.024	-0.007	+0.001 ok
		split	starter x0.9756 at 0.611 share (log5 0.0796 vs his 0.0868) + bullpen x0.8651 at 0.389 share (log5 0.0706 vs pen 0.0770) = x0.9326 \| starter alone would have been -0.006
		Home runswhole staff	0.106 → 0.107	+0.001	0.0304	0.0308n=354 Average	0.0311n=730 Average	0.0315	+0.001	+0.002	0 ok
		split	starter x1.0228 at 0.611 share (log5 0.0315 vs his 0.0311) + bullpen x0.9965 at 0.389 share (log5 0.0307 vs pen 0.0303) = x1.0126 \| starter alone would have been +0.002
Neutralising Ryan McMahon to the qualified-population mean moves the club's KPI run forecast by 0.0599 runs, and the weighted stage by 0.0467. Interactions make these non-additive across players, so they do not sum to the published movement.
STARTER Cam Schlittler21.4633 BF	Starter	Hits	2.854 → 3.050	+0.196	0.2169	0.1805n=741 Very good	0.23119/9 qualified lineup mean	0.1929	-0.781	+0.305	-0.039 ok
		Strikeouts	7.006 → 5.737	-1.268	0.2212	0.3029n=741 Very good	0.17749/9 qualified lineup mean	0.2481	+1.753	-0.941	-0.237 ok
		Walks	1.500 → 1.497	-0.003	0.0889	0.0701n=741 Good	0.08879/9 qualified lineup mean	0.0700	-0.404	-0.004	+0.001 ok
		Home runs	0.294 → 0.295	+0.001	0.0304	0.0249n=741 Good	0.03059/9 qualified lineup mean	0.0250	-0.118	+0.002	-0.000 ok
RELIEVER Will Warren2.7878 BF	Reliever	Hits	0.639 → 0.680	+0.041	0.2169	0.2293n=673 Below average	0.23119/9 qualified lineup mean	0.2441	+0.035	+0.040	+0.002 ok
		Strikeouts	0.647 → 0.520	-0.127	0.2212	0.2320n=673 Average	0.17749/9 qualified lineup mean	0.1866	+0.030	-0.122	-0.004 ok
		Walks	0.225 → 0.225	-0.001	0.0889	0.0807n=673 Average	0.08879/9 qualified lineup mean	0.0805	-0.023	-0.001	0 ok
		Home runs	0.086 → 0.086	+0.000	0.0304	0.0309n=673 Average	0.03059/9 qualified lineup mean	0.0310	+0.001	+0.000	0 ok
RELIEVER David Bednar2.5888 BF	Reliever	Hits	0.525 → 0.560	+0.035	0.2169	0.2027n=277 Good	0.23119/9 qualified lineup mean	0.2163	-0.037	+0.037	-0.002 ok
		Strikeouts	0.679 → 0.551	-0.129	0.2212	0.2624n=277 Good	0.17749/9 qualified lineup mean	0.2127	+0.107	-0.113	-0.015 ok
		Walks	0.203 → 0.202	-0.000	0.0889	0.0783n=277 Good	0.08879/9 qualified lineup mean	0.0781	-0.028	-0.001	+0.000 ok
		Home runs	0.063 → 0.063	+0.000	0.0304	0.0243n=277 Very good	0.03059/9 qualified lineup mean	0.0244	-0.016	+0.000	-0.000 ok
RELIEVER Paul Blackburn2.5785 BF	Middle Reliever (High Leverage)	Hits	0.914 → 0.975	+0.061	0.2169	0.1987n=329 Good	0.23119/9 qualified lineup mean	0.2120	-0.047	+0.037	-0.002 ok
		Strikeouts	0.848 → 0.677	-0.171	0.2212	0.2046n=329 Average	0.17749/9 qualified lineup mean	0.1634	-0.043	-0.113	+0.007 ok
		Walks	0.465 → 0.464	-0.001	0.0889	0.0946n=329 Average	0.08879/9 qualified lineup mean	0.0944	+0.015	-0.001	0 ok
		Home runs	0.067 → 0.067	+0.000	0.0304	0.0265n=329 Good	0.03059/9 qualified lineup mean	0.0266	-0.010	+0.000	0 ok
RELIEVER Brent Headrick2.4959 BF	Setup Man	Hits	0.585 → 0.624	+0.039	0.2169	0.2125n=326 Average	0.23119/9 qualified lineup mean	0.2265	-0.011	+0.035	-0.001 ok
		Strikeouts	0.757 → 0.617	-0.140	0.2212	0.2849n=326 Very good	0.17749/9 qualified lineup mean	0.2323	+0.159	-0.109	-0.022 ok
		Walks	0.329 → 0.329	-0.001	0.0889	0.0848n=326 Average	0.08879/9 qualified lineup mean	0.0846	-0.010	-0.001	0 ok
		Home runs	0.068 → 0.068	+0.000	0.0304	0.0240n=326 Very good	0.03059/9 qualified lineup mean	0.0241	-0.016	+0.000	-0.000 ok
RELIEVER John Schreiber2.2453 BF	Middle Reliever (High Leverage)	Hits	0.908 → 0.968	+0.060	0.2169	0.2088n=246 Average	0.23119/9 qualified lineup mean	0.2227	-0.018	+0.032	-0.001 ok
		Strikeouts	0.591 → 0.469	-0.122	0.2212	0.1770n=246 Below average	0.17749/9 qualified lineup mean	0.1404	-0.099	-0.098	+0.016 ok
		Walks	0.340 → 0.340	-0.001	0.0889	0.0991n=246 Below average	0.08879/9 qualified lineup mean	0.0989	+0.023	-0.000	0 ok
		Home runs	0.091 → 0.091	+0.000	0.0304	0.0297n=246 Average	0.03059/9 qualified lineup mean	0.0298	-0.002	+0.000	0 ok
RELIEVER Ryan Weathers1.6748 BF	Reliever	Hits	0.377 → 0.401	+0.024	0.2169	0.2248n=568 Below average	0.23119/9 qualified lineup mean	0.2393	+0.013	+0.024	+0.001 ok
		Strikeouts	0.415 → 0.335	-0.080	0.2212	0.2480n=568 Good	0.17749/9 qualified lineup mean	0.2002	+0.045	-0.073	-0.007 ok
		Walks	0.120 → 0.120	-0.000	0.0889	0.0717n=568 Good	0.08879/9 qualified lineup mean	0.0715	-0.029	-0.000	+0.000 ok
		Home runs	0.052 → 0.052	+0.000	0.0304	0.0311n=568 Average	0.03059/9 qualified lineup mean	0.0312	+0.001	+0.000	0 ok
RELIEVER Ryan Yarbrough0.3715 BF	Reliever	Hits	0.077 → 0.082	+0.005	0.2169	0.2080n=253 Average	0.23119/9 qualified lineup mean	0.2218	-0.003	+0.005	-0.000 ok
		Strikeouts	0.072 → 0.058	-0.015	0.2212	0.1943n=253 Below average	0.17749/9 qualified lineup mean	0.1548	-0.010	-0.016	+0.002 ok
		Walks	0.032 → 0.032	-0.000	0.0889	0.0853n=253 Average	0.08879/9 qualified lineup mean	0.0851	-0.001	-0.000	0 ok
		Home runs	0.009 → 0.009	0	0.0304	0.0237n=253 Very good	0.03059/9 qualified lineup mean	0.0238	-0.003	0	0 ok

TB against Tyler Wells (RHP)

The pipeline applied no club-wide rescale to TB hitter lines.
Opposing staff exposure. These hitters face the starter for 21.4633 of 36.206 projected batters (59.3%) and the 3 projected relievers for the remaining 40.7%. Every metric below is the exposure-weighted blend of both matchups; the starter-only figure is shown beside it. Real 2026 game logs put a starter at 21.75 batters faced on average (median 23, p10-p90 17-27) and a whole staff at 37.9: this starter is inside that band and this staff is inside its. The per-player columns are the measured alternative: each subject moved by his own KPI pairing. Exposure for a hitter is his projected plate appearances; for the starter it is batters faced.

Player	Pos	Metric	Published → KPI	Δ	League	Own rate	Opponent rate	Log5	Own skill	Opponent	Remainder
Yandy Díaz	DH	Hitswhole staff	0.949 → 0.850	-0.099	0.2169	0.2464n=678 Very good	0.1805n=741 Very good	0.2063	+0.129	-0.159	-0.016 ok
		split	starter x0.8374 at 0.593 share (log5 0.2063 vs his 0.1805) + bullpen x0.9796 at 0.407 share (log5 0.2413 vs pen 0.2123) = x0.8953 \| starter alone would have been -0.154
		Strikeoutswhole staff	0.881 → 1.123	+0.242	0.2212	0.1455n=678 Very good	0.3029n=741 Very good	0.2067	-0.330	+0.356	-0.089 ok
		split	starter x1.4203 at 0.593 share (log5 0.2067 vs his 0.3029) + bullpen x1.0639 at 0.407 share (log5 0.1548 vs pen 0.2340) = x1.2752 \| starter alone would have been +0.370
		Walkswhole staff	0.280 → 0.240	-0.040	0.0889	0.0835n=678 Average	0.0701n=741 Good	0.0658	-0.024	-0.082	+0.005 ok
		split	starter x0.7876 at 0.593 share (log5 0.0658 vs his 0.0701) + bullpen x0.9589 at 0.407 share (log5 0.0801 vs pen 0.0853) = x0.8574 \| starter alone would have been -0.059
		Home runswhole staff	0.139 → 0.119	-0.020	0.0304	0.0319n=678 Average	0.0249n=741 Good	0.0261	+0.006	-0.024	-0.001 ok
		split	starter x0.8197 at 0.593 share (log5 0.0261 vs his 0.0249) + bullpen x0.904 at 0.407 share (log5 0.0288 vs pen 0.0275) = x0.854 \| starter alone would have been -0.025
Neutralising Yandy Díaz to the qualified-population mean moves the club's KPI run forecast by -0.1520 runs, and the weighted stage by -0.0669. Interactions make these non-additive across players, so they do not sum to the published movement.
Jonathan Aranda	1B	Hitswhole staff	0.848 → 0.758	-0.090	0.2169	0.2302n=661 Good	0.1805n=741 Very good	0.1921	+0.056	-0.152	-0.007 ok
		split	starter x0.8345 at 0.593 share (log5 0.1921 vs his 0.1805) + bullpen x0.9791 at 0.407 share (log5 0.2254 vs pen 0.2123) = x0.8934 \| starter alone would have been -0.140
		Strikeoutswhole staff	1.316 → 1.623	+0.308	0.2212	0.2420n=661 Average	0.3029n=741 Very good	0.3282	+0.087	+0.341	+0.019 ok
		split	starter x1.3559 at 0.593 share (log5 0.3282 vs his 0.3029) + bullpen x1.0562 at 0.407 share (log5 0.2557 vs pen 0.2340) = x1.2339 \| starter alone would have been +0.468
		Walkswhole staff	0.341 → 0.293	-0.048	0.0889	0.1007n=661 Good	0.0701n=741 Good	0.0797	+0.049	-0.079	-0.009 ok
		split	starter x0.7908 at 0.593 share (log5 0.0797 vs his 0.0701) + bullpen x0.9597 at 0.407 share (log5 0.0967 vs pen 0.0853) = x0.8596 \| starter alone would have been -0.071
		Home runswhole staff	0.144 → 0.123	-0.021	0.0304	0.0325n=661 Average	0.0249n=741 Good	0.0266	+0.009	-0.023	-0.002 ok
		split	starter x0.8198 at 0.593 share (log5 0.0266 vs his 0.0249) + bullpen x0.9041 at 0.407 share (log5 0.0293 vs pen 0.0275) = x0.8541 \| starter alone would have been -0.026
Neutralising Jonathan Aranda to the qualified-population mean moves the club's KPI run forecast by 0.2774 runs, and the weighted stage by 0.1221. Interactions make these non-additive across players, so they do not sum to the published movement.
Junior Caminero	3B	Hitswhole staff	0.852 → 0.762	-0.091	0.2169	0.2330n=695 Good	0.1805n=741 Very good	0.1945	+0.065	-0.146	-0.008 ok
		split	starter x0.835 at 0.593 share (log5 0.1945 vs his 0.1805) + bullpen x0.9792 at 0.407 share (log5 0.2281 vs pen 0.2123) = x0.8937 \| starter alone would have been -0.141
		Strikeoutswhole staff	0.901 → 1.133	+0.232	0.2212	0.1853n=695 Good	0.3029n=741 Very good	0.2581	-0.144	+0.328	-0.036 ok
		split	starter x1.393 at 0.593 share (log5 0.2581 vs his 0.3029) + bullpen x1.0607 at 0.407 share (log5 0.1965 vs pen 0.2340) = x1.2577 \| starter alone would have been +0.354
		Walkswhole staff	0.339 → 0.292	-0.047	0.0889	0.1100n=695 Good	0.0701n=741 Good	0.0872	+0.085	-0.075	-0.016 ok
		split	starter x0.7925 at 0.593 share (log5 0.0872 vs his 0.0701) + bullpen x0.9601 at 0.407 share (log5 0.1057 vs pen 0.0853) = x0.8607 \| starter alone would have been -0.070
		Home runswhole staff	0.229 → 0.196	-0.033	0.0304	0.0522n=695 Very good	0.0249n=741 Good	0.0429	+0.087	-0.022	-0.015 ok
		split	starter x0.8228 at 0.593 share (log5 0.0429 vs his 0.0249) + bullpen x0.9059 at 0.407 share (log5 0.0472 vs pen 0.0275) = x0.8566 \| starter alone would have been -0.041
Neutralising Junior Caminero to the qualified-population mean moves the club's KPI run forecast by -0.0804 runs, and the weighted stage by -0.0354. Interactions make these non-additive across players, so they do not sum to the published movement.
Liam Hicks	C	Hitswhole staff	0.829 → 0.741	-0.088	0.2169	0.2335n=551 Good	0.1805n=741 Very good	0.1950	+0.066	-0.143	-0.008 ok
		split	starter x0.8351 at 0.593 share (log5 0.1950 vs his 0.1805) + bullpen x0.9792 at 0.407 share (log5 0.2287 vs pen 0.2123) = x0.8938 \| starter alone would have been -0.137
		Strikeoutswhole staff	0.515 → 0.665	+0.150	0.2212	0.1080n=551 Very good	0.3029n=741 Very good	0.1563	-0.445	+0.321	-0.131 ok
		split	starter x1.447 at 0.593 share (log5 0.1563 vs his 0.3029) + bullpen x1.0669 at 0.407 share (log5 0.1152 vs pen 0.2340) = x1.2922 \| starter alone would have been +0.230
		Walkswhole staff	0.321 → 0.276	-0.045	0.0889	0.0994n=551 Average	0.0701n=741 Good	0.0786	+0.041	-0.074	-0.008 ok
		split	starter x0.7905 at 0.593 share (log5 0.0786 vs his 0.0701) + bullpen x0.9596 at 0.407 share (log5 0.0954 vs pen 0.0853) = x0.8594 \| starter alone would have been -0.067
		Home runswhole staff	0.136 → 0.116	-0.020	0.0304	0.0319n=551 Average	0.0249n=741 Good	0.0262	+0.006	-0.022	-0.001 ok
		split	starter x0.8197 at 0.593 share (log5 0.0262 vs his 0.0249) + bullpen x0.904 at 0.407 share (log5 0.0289 vs pen 0.0275) = x0.854 \| starter alone would have been -0.025
Neutralising Liam Hicks to the qualified-population mean moves the club's KPI run forecast by -0.0260 runs, and the weighted stage by -0.0115. Interactions make these non-additive across players, so they do not sum to the published movement.
Chandler Simpson	LF	Hitswhole staff	0.970 → 0.870	-0.100	0.2169	0.2576n=606 Very good	0.1805n=741 Very good	0.2162	+0.162	-0.145	-0.020 ok
		split	starter x0.8394 at 0.593 share (log5 0.2162 vs his 0.1805) + bullpen x0.9799 at 0.407 share (log5 0.2524 vs pen 0.2123) = x0.8966 \| starter alone would have been -0.156
		Strikeoutswhole staff	0.451 → 0.587	+0.136	0.2212	0.0898n=606 Very good	0.3029n=741 Very good	0.1311	-0.524	+0.326	-0.161 ok
		split	starter x1.4603 at 0.593 share (log5 0.1311 vs his 0.3029) + bullpen x1.0683 at 0.407 share (log5 0.0959 vs pen 0.2340) = x1.3007 \| starter alone would have been +0.208
		Walkswhole staff	0.167 → 0.143	-0.025	0.0889	0.0505n=606 Poor	0.0701n=741 Good	0.0394	-0.153	-0.075	+0.031 ok
		split	starter x0.7816 at 0.593 share (log5 0.0394 vs his 0.0701) + bullpen x0.9575 at 0.407 share (log5 0.0483 vs pen 0.0853) = x0.8532 \| starter alone would have been -0.037
		Home runswhole staff	0.038 → 0.032	-0.006	0.0304	0.0092n=606 Poor	0.0249n=741 Good	0.0075	-0.084	-0.022	+0.015 ok
		split	starter x0.8162 at 0.593 share (log5 0.0075 vs his 0.0249) + bullpen x0.902 at 0.407 share (log5 0.0083 vs pen 0.0275) = x0.8511 \| starter alone would have been -0.007
Neutralising Chandler Simpson to the qualified-population mean moves the club's KPI run forecast by -0.0661 runs, and the weighted stage by -0.0291. Interactions make these non-additive across players, so they do not sum to the published movement.
Victor Mesa Jr.	RF	Hitswhole staff	0.801 → 0.715	-0.086	0.2169	0.2223n=259 Average	0.1805n=741 Very good	0.1852	+0.021	-0.140	-0.003 ok
		split	starter x0.8331 at 0.593 share (log5 0.1852 vs his 0.1805) + bullpen x0.9789 at 0.407 share (log5 0.2177 vs pen 0.2123) = x0.8925 \| starter alone would have been -0.134
		Strikeoutswhole staff	0.986 → 1.231	+0.245	0.2212	0.2073n=259 Average	0.3029n=741 Very good	0.2857	-0.053	+0.313	-0.013 ok
		split	starter x1.3784 at 0.593 share (log5 0.2857 vs his 0.3029) + bullpen x1.059 at 0.407 share (log5 0.2195 vs pen 0.2340) = x1.2483 \| starter alone would have been +0.373
		Walkswhole staff	0.265 → 0.228	-0.038	0.0889	0.0888n=259 Average	0.0701n=741 Good	0.0701	-0.000	-0.072	+0.000 ok
		split	starter x0.7886 at 0.593 share (log5 0.0701 vs his 0.0701) + bullpen x0.9591 at 0.407 share (log5 0.0852 vs pen 0.0853) = x0.858 \| starter alone would have been -0.056
		Home runswhole staff	0.177 → 0.152	-0.026	0.0304	0.0421n=259 Good	0.0249n=741 Good	0.0346	+0.045	-0.021	-0.008 ok
		split	starter x0.8212 at 0.593 share (log5 0.0346 vs his 0.0249) + bullpen x0.905 at 0.407 share (log5 0.0381 vs pen 0.0275) = x0.8553 \| starter alone would have been -0.032
Neutralising Victor Mesa Jr. to the qualified-population mean moves the club's KPI run forecast by 0.5287 runs, and the weighted stage by 0.2326. Interactions make these non-additive across players, so they do not sum to the published movement.
Jonny DeLuca	CF	Hitswhole staff	0.823 → 0.736	-0.087	0.2169	0.2346n=363 Good	0.1805n=741 Very good	0.1960	+0.067	-0.137	-0.008 ok
		split	starter x0.8353 at 0.593 share (log5 0.1960 vs his 0.1805) + bullpen x0.9793 at 0.407 share (log5 0.2298 vs pen 0.2123) = x0.8939 \| starter alone would have been -0.136
		Strikeoutswhole staff	1.039 → 1.294	+0.255	0.2212	0.2132n=363 Average	0.3029n=741 Very good	0.2930	-0.030	+0.306	-0.007 ok
		split	starter x1.3745 at 0.593 share (log5 0.2930 vs his 0.3029) + bullpen x1.0585 at 0.407 share (log5 0.2257 vs pen 0.2340) = x1.2458 \| starter alone would have been +0.389
		Walkswhole staff	0.169 → 0.144	-0.025	0.0889	0.0572n=363 Poor	0.0701n=741 Good	0.0448	-0.119	-0.070	+0.024 ok
		split	starter x0.7828 at 0.593 share (log5 0.0448 vs his 0.0701) + bullpen x0.9578 at 0.407 share (log5 0.0548 vs pen 0.0853) = x0.8541 \| starter alone would have been -0.037
		Home runswhole staff	0.126 → 0.108	-0.018	0.0304	0.0303n=363 Average	0.0249n=741 Good	0.0249	-0.000	-0.021	0 ok
		split	starter x0.8194 at 0.593 share (log5 0.0249 vs his 0.0249) + bullpen x0.9039 at 0.407 share (log5 0.0274 vs pen 0.0275) = x0.8538 \| starter alone would have been -0.023
Neutralising Jonny DeLuca to the qualified-population mean moves the club's KPI run forecast by 0.2032 runs, and the weighted stage by 0.0894. Interactions make these non-additive across players, so they do not sum to the published movement.
Richie Palacios	2B	Hitswhole staff	0.699 → 0.623	-0.076	0.2169	0.2110n=396 Average	0.1805n=741 Very good	0.1754	-0.020	-0.126	+0.003 ok
		split	starter x0.8311 at 0.593 share (log5 0.1754 vs his 0.1805) + bullpen x0.9786 at 0.407 share (log5 0.2065 vs pen 0.2123) = x0.8912 \| starter alone would have been -0.118
		Strikeoutswhole staff	0.695 → 0.877	+0.182	0.2212	0.1749n=396 Good	0.3029n=741 Very good	0.2448	-0.160	+0.283	-0.041 ok
		split	starter x1.4 at 0.593 share (log5 0.2448 vs his 0.3029) + bullpen x1.0615 at 0.407 share (log5 0.1856 vs pen 0.2340) = x1.2622 \| starter alone would have been +0.278
		Walkswhole staff	0.305 → 0.263	-0.043	0.0889	0.1099n=396 Good	0.0701n=741 Good	0.0871	+0.072	-0.065	-0.014 ok
		split	starter x0.7925 at 0.593 share (log5 0.0871 vs his 0.0701) + bullpen x0.96 at 0.407 share (log5 0.1055 vs pen 0.0853) = x0.8607 \| starter alone would have been -0.063
		Home runswhole staff	0.098 → 0.084	-0.014	0.0304	0.0273n=396 Average	0.0249n=741 Good	0.0224	-0.011	-0.019	+0.002 ok
		split	starter x0.819 at 0.593 share (log5 0.0224 vs his 0.0249) + bullpen x0.9036 at 0.407 share (log5 0.0247 vs pen 0.0275) = x0.8534 \| starter alone would have been -0.018
Neutralising Richie Palacios to the qualified-population mean moves the club's KPI run forecast by 0.0866 runs, and the weighted stage by 0.0381. Interactions make these non-additive across players, so they do not sum to the published movement.
Taylor Walls	SS	Hitswhole staff	0.664 → 0.591	-0.073	0.2169	0.2028n=430 Below average	0.1805n=741 Very good	0.1682	-0.048	-0.124	+0.006 ok
		split	starter x0.8296 at 0.593 share (log5 0.1682 vs his 0.1805) + bullpen x0.9784 at 0.407 share (log5 0.1984 vs pen 0.2123) = x0.8902 \| starter alone would have been -0.113
		Strikeoutswhole staff	1.033 → 1.274	+0.242	0.2212	0.2418n=430 Average	0.3029n=741 Very good	0.3280	+0.070	+0.278	+0.015 ok
		split	starter x1.356 at 0.593 share (log5 0.3280 vs his 0.3029) + bullpen x1.0563 at 0.407 share (log5 0.2555 vs pen 0.2340) = x1.234 \| starter alone would have been +0.368
		Walkswhole staff	0.285 → 0.245	-0.040	0.0889	0.1012n=430 Good	0.0701n=741 Good	0.0801	+0.042	-0.064	-0.008 ok
		split	starter x0.7909 at 0.593 share (log5 0.0801 vs his 0.0701) + bullpen x0.9597 at 0.407 share (log5 0.0972 vs pen 0.0853) = x0.8596 \| starter alone would have been -0.060
		Home runswhole staff	0.051 → 0.043	-0.008	0.0304	0.0144n=430 Poor	0.0249n=741 Good	0.0118	-0.054	-0.019	+0.010 ok
		split	starter x0.817 at 0.593 share (log5 0.0118 vs his 0.0249) + bullpen x0.9025 at 0.407 share (log5 0.0130 vs pen 0.0275) = x0.8518 \| starter alone would have been -0.009
Neutralising Taylor Walls to the qualified-population mean moves the club's KPI run forecast by 0.0184 runs, and the weighted stage by 0.0081. Interactions make these non-additive across players, so they do not sum to the published movement.
STARTER Freddy Peralta21.6754 BF	Starter	Hits	3.814 → 3.803	-0.011	0.2169	0.2195n=730 Average	0.21638/9 qualified lineup mean	0.2189	+0.057	-0.013	-0.000 ok
		Strikeouts	4.724 → 5.113	+0.389	0.2212	0.2144n=730 Average	0.23928/9 qualified lineup mean	0.2321	-0.147	+0.391	-0.008 ok
		Walks	1.829 → 2.002	+0.173	0.0889	0.0868n=730 Average	0.09738/9 qualified lineup mean	0.0950	-0.047	+0.181	-0.004 ok
		Home runs	0.398 → 0.481	+0.083	0.0304	0.0311n=730 Average	0.03688/9 qualified lineup mean	0.0376	+0.015	+0.138	+0.003 ok
RELIEVER Bryan Baker4.0796 BF	Closer	Hits	0.547 → 0.545	-0.002	0.2169	0.1836n=244 Very good	0.21638/9 qualified lineup mean	0.1831	-0.136	-0.003	+0.000 ok
		Strikeouts	0.960 → 1.036	+0.075	0.2212	0.2490n=244 Good	0.23928/9 qualified lineup mean	0.2686	+0.114	+0.073	+0.006 ok
		Walks	0.292 → 0.319	+0.028	0.0889	0.0897n=244 Average	0.09738/9 qualified lineup mean	0.0981	+0.003	+0.034	+0.000 ok
		Home runs	0.049 → 0.059	+0.010	0.0304	0.0254n=244 Good	0.03688/9 qualified lineup mean	0.0308	-0.020	+0.026	-0.004 ok
RELIEVER Steven Matz2.8075 BF	Middle Reliever	Hits	1.145 → 1.142	-0.003	0.2169	0.2166n=342 Average	0.21638/9 qualified lineup mean	0.2160	-0.001	-0.002	0 ok
		Strikeouts	1.141 → 1.238	+0.098	0.2212	0.1865n=342 Below average	0.23928/9 qualified lineup mean	0.2024	-0.097	+0.051	-0.006 ok
		Walks	0.305 → 0.334	+0.029	0.0889	0.0781n=342 Good	0.09738/9 qualified lineup mean	0.0855	-0.030	+0.024	-0.003 ok
		Home runs	0.240 → 0.290	+0.050	0.0304	0.0362n=342 Poor	0.03688/9 qualified lineup mean	0.0438	+0.016	+0.018	+0.003 ok
RELIEVER Cole Sulser2.383 BF	Setup Man	Hits	0.855 → 0.853	-0.002	0.2169	0.2203n=236 Average	0.21638/9 qualified lineup mean	0.2197	+0.008	-0.001	0 ok
		Strikeouts	1.099 → 1.187	+0.088	0.2212	0.2332n=236 Average	0.23928/9 qualified lineup mean	0.2519	+0.029	+0.043	+0.002 ok
		Walks	0.312 → 0.342	+0.030	0.0889	0.0788n=236 Good	0.09738/9 qualified lineup mean	0.0863	-0.024	+0.020	-0.002 ok
		Home runs	0.104 → 0.126	+0.022	0.0304	0.0301n=236 Average	0.03688/9 qualified lineup mean	0.0365	-0.001	+0.015	-0.000 ok
RELIEVER Cam Booser1.6856 BF	Reliever	Hits	0.336 → 0.335	-0.001	0.2169	0.1990n=152 Good	0.21638/9 qualified lineup mean	0.1985	-0.030	-0.001	+0.000 ok
		Strikeouts	0.455 → 0.490	+0.035	0.2212	0.2702n=152 Good	0.23928/9 qualified lineup mean	0.2907	+0.083	+0.030	+0.004 ok
		Walks	0.109 → 0.119	+0.011	0.0889	0.0646n=152 Very good	0.09738/9 qualified lineup mean	0.0709	-0.041	+0.014	-0.004 ok
		Home runs	0.052 → 0.063	+0.011	0.0304	0.0310n=152 Average	0.03688/9 qualified lineup mean	0.0375	+0.001	+0.011	+0.000 ok
RELIEVER Kevin Kelly1.596 BF	Reliever	Hits	0.331 → 0.330	-0.001	0.2169	0.2073n=279 Average	0.21638/9 qualified lineup mean	0.2067	-0.015	-0.001	0 ok
		Strikeouts	0.359 → 0.388	+0.029	0.2212	0.2248n=279 Average	0.23928/9 qualified lineup mean	0.2430	+0.006	+0.029	+0.000 ok
		Walks	0.091 → 0.100	+0.009	0.0889	0.0573n=279 Very good	0.09738/9 qualified lineup mean	0.0629	-0.051	+0.013	-0.004 ok
		Home runs	0.043 → 0.052	+0.009	0.0304	0.0270n=279 Good	0.03688/9 qualified lineup mean	0.0327	-0.005	+0.010	-0.001 ok
RELIEVER Ian Seymour0.8959 BF	Reliever	Hits	0.189 → 0.188	-0.001	0.2169	0.2104n=551 Average	0.21638/9 qualified lineup mean	0.2098	-0.006	-0.001	0 ok
		Strikeouts	0.249 → 0.268	+0.019	0.2212	0.2784n=551 Very good	0.23928/9 qualified lineup mean	0.2993	+0.051	+0.016	+0.003 ok
		Walks	0.063 → 0.069	+0.006	0.0889	0.0706n=551 Good	0.09738/9 qualified lineup mean	0.0774	-0.016	+0.007	-0.001 ok
		Home runs	0.035 → 0.042	+0.007	0.0304	0.0387n=551 Poor	0.03688/9 qualified lineup mean	0.0467	+0.007	+0.006	+0.002 ok
RELIEVER Tyler Wells0.3349 BF	Reliever	Hits	0.065 → 0.065	-0.000	0.2169	0.1936n=273 Very good	0.21638/9 qualified lineup mean	0.1930	-0.008	-0.000	0 ok
		Strikeouts	0.085 → 0.092	+0.007	0.2212	0.2541n=273 Good	0.23928/9 qualified lineup mean	0.2738	+0.011	+0.006	+0.001 ok
		Walks	0.025 → 0.027	+0.002	0.0889	0.0743n=273 Good	0.09738/9 qualified lineup mean	0.0815	-0.005	+0.003	-0.000 ok
		Home runs	0.010 → 0.012	+0.002	0.0304	0.0300n=273 Average	0.03688/9 qualified lineup mean	0.0363	-0.000	+0.002	0 ok

Published pitching, integer draw beside the continuous model line

Pitcher	Role	BF	IP	expected IP	H int/cont	SO int/cont	BB int/cont	ER int/cont	Rate source
Cam Schlittler	Starter	21.5	5.2	5.703	1 / 2.854	5 / 7.006	1 / 1.500	0 / 0.817	all_players_stats_recent_season_blend
Paul Blackburn	Middle Reliever (High Leverage)	5.4	1.1	1.341	0 / 0.914	1 / 0.848	0 / 0.465	0 / 0.157	all_players_stats_recent_season_blend
John Schreiber	Middle Reliever (High Leverage)	4.3	1.0	1.006	2 / 0.908	0 / 0.591	0 / 0.340	0 / 0.377	all_players_stats_recent_season_blend
Brent Headrick	Setup Man	2.9	1.0	0.671	2 / 0.585	1 / 0.757	0 / 0.329	0 / 0.231	all_players_stats_recent_season_blend
Freddy Peralta	Starter	21.7	5.1	5.344	2 / 3.814	6 / 4.724	1 / 1.829	1 / 1.371	all_players_stats_recent_season_blend
Cole Sulser	Setup Man	4.2	1.0	1.002	1 / 0.855	1 / 1.099	0 / 0.312	0 / 0.385	all_players_stats_recent_season_blend
Bryan Baker	Closer	3.8	1.0	1.002	1 / 0.547	0 / 0.960	1 / 0.292	1 / 0.110	all_players_stats_recent_season_blend
Steven Matz	Middle Reliever	5.5	1.2	1.336	4 / 1.145	2 / 1.141	1 / 0.305	2 / 0.757	all_players_stats_recent_season_blend

Audit: NO DEFECT FOUND

Invariant	Verdict	Evidence
blend arithmetic away_runs	PASS	4.0000 + 0.78 x (4.4268 - 4.0000) = 4.3329, recorded 4.3329
blend arithmetic home_runs	PASS	3.0000 + 0.44 x (3.1160 - 3.0000) = 3.0510, recorded 3.0510
blend arithmetic home_win	PASS	0.4534 + 0.00 x (0.6851 - 0.4534) = 0.4534, recorded 0.4534
published rounding faithful to the blend	PASS	blended 4.3329-3.0510 rounds to 4-3, layer published 4-3
published probability on the score's side	PASS	4-3 with p(home) 0.453398
Log5 additivity136 checks	PASS	136 of 136 pass
away club fields exactly one starter	PASS	1 starters in a staff of 4
away staff totals a regulation workload	PASS	27 outs recorded, 27 expected
away starter carries a continuous line	PASS	expected_IP 5.703
away starter batters faced inside the real band	PASS	Cam Schlittler projected 21.5 batters faced; real starters run p01-p99 4-30, median 23, mean 21.75
away reliever batters faced inside the real band3 checks	PASS	3 of 3 pass
home club fields exactly one starter	PASS	1 starters in a staff of 4
home staff totals a regulation workload	PASS	27 outs recorded, 27 expected
home starter carries a continuous line	PASS	expected_IP 5.344
home starter batters faced inside the real band	PASS	Freddy Peralta projected 21.7 batters faced; real starters run p01-p99 4-30, median 23, mean 21.75
home reliever batters faced inside the real band3 checks	PASS	3 of 3 pass
away hitters satisfy every coherence rule	PASS	9 hitters checked; H<=AB, HR<=H, 2B+3B+HR<=H, SO<=AB, continuous line present
home hitters satisfy every coherence rule	PASS	9 hitters checked; H<=AB, HR<=H, 2B+3B+HR<=H, SO<=AB, continuous line present

Provenance. Read only from published artifacts and stored KPI value documents. Nothing is recomputed, so this document cannot disagree with what was published. Per-player metric adjustments are a measurement of the catalog at player level and are labelled as such throughout; the pipeline itself moved team runs, the win probability, and hitter lines by one club-wide ratio.

Verdict across the slate

How to read this, and what a defect looks like

Slate overview

Per game

Stage chain: every stage that touched the score

KPI team layer: the blend, with its arithmetic

What drove the KPI forecast for away_runs

What drove the KPI forecast for home_runs

Per-player metric movement

CHW against Hagen Smith (LHP)

CLE against Matt Festa (RHP)

Published pitching, integer draw beside the continuous model line

Audit: NO DEFECT FOUND

Stage chain: every stage that touched the score

KPI team layer: the blend, with its arithmetic

What drove the KPI forecast for away_runs

What drove the KPI forecast for home_runs

Per-player metric movement

NYY against Ryan Yarbrough (LHP)

TB against Tyler Wells (RHP)

Published pitching, integer draw beside the continuous model line

Audit: NO DEFECT FOUND

What drove the KPI forecast for `away_runs`

What drove the KPI forecast for `home_runs`

What drove the KPI forecast for `away_runs`

What drove the KPI forecast for `home_runs`