MLB Prediction System

KPI Player Effects

How each KPI adds or removes a hit, a strikeout, a walk or a home run, per player, against the arm he actually faces

Slate 20261013Games 1Schema kpi_player_effects_v2Built 2026-10-12 18:53Z

What this document measures

Every counting stat below has a KPI on both sides of the matchup, measured on the same denominator: a plate appearance. That is what lets the hitter's reading and the starter's reading be combined instead of listed side by side. The combination is the standard Log5 odds ratio, which is exactly additive in log odds, so the hitter's push and the pitcher's push separate cleanly and the remainder is reported rather than absorbed.

Hits use a derived KPI. Hit rate is not in config/kpi_registry.json: registering it would move the registry hash recorded in the model manifest and fail the live layer closed until a retrain. It is computed with the same strict as-of cutoff and the same shrinkage, stored in its own namespace, and never read by the runtime.

These per-player numbers are measured, not published. The live KPI layer moves team runs and the win probability. The hitter lines in the artifact were then scaled by one club-wide ratio, identical for the leadoff hitter and the ninth bat. The per-hitter columns here are what that single ratio discards.

The whole staff, not only the starter. A hitter takes roughly four in ten of his plate appearances against relievers. Each club's bullpen here is the pregame bullpen: the seven busiest healthy relievers of the last 30 days, the same arms the KPI forecast weighted and the simulation pitched, each sharing the relief work by the batters faced the forecast gave him. The bullpen's workload tonight is the club's real mean staff batters faced minus the starter's projection. The published staff alone holds only the relievers one simulation draw used, so it is shown beside the bullpen for comparison, not used.

Reading a delta. Own skill is measured against the league. The opponent effect is measured against the hitter's own rate, which is the only one that may adjust a published line, because the published line already carries his own skill. Adding both to a published line would count the hitter twice.

Coverage. 18 hitter-versus-starter pairings; 18 have both sides above their sample floor on hits and 18 carry a non-zero hit adjustment. A side below its floor falls back to the league rate and is marked n<floor, which is a stated absence of evidence rather than a measured zero. All four outcome metrics have value documents. Handedness source: model bundle pitcher_hands.json.

Largest per-player movements on the slate

Hitter	Club	Metric	Starter faced (+ bullpen)	Published	KPI adjusted, whole staff	Δ	Hitter band	Pitcher band	Game
Jo Adell	CLE	Strikeouts	Drew Rasmussen	1.109	1.272	+0.163	Average	Good	20261013_CLE@TB
Angel Martínez	CLE	Strikeouts	Drew Rasmussen	1.004	1.156	+0.152	Average	Good	20261013_CLE@TB
Patrick Bailey	CLE	Strikeouts	Drew Rasmussen	1.024	1.168	+0.144	Below average	Good	20261013_CLE@TB
Nathaniel Lowe	CLE	Strikeouts	Drew Rasmussen	0.910	1.045	+0.136	Average	Good	20261013_CLE@TB
Travis Bazzana	CLE	Strikeouts	Drew Rasmussen	0.837	0.966	+0.129	Average	Good	20261013_CLE@TB
Steven Kwan	CLE	Walks	Drew Rasmussen	0.406	0.288	-0.118	Very good	Very good	20261013_CLE@TB
José Ramírez	CLE	Strikeouts	Drew Rasmussen	0.680	0.795	+0.115	Very good	Good	20261013_CLE@TB
Chase DeLauter	CLE	Strikeouts	Drew Rasmussen	0.672	0.784	+0.112	Good	Good	20261013_CLE@TB
José Ramírez	CLE	Walks	Drew Rasmussen	0.360	0.254	-0.106	Good	Very good	20261013_CLE@TB
Chase DeLauter	CLE	Walks	Drew Rasmussen	0.321	0.226	-0.095	Good	Very good	20261013_CLE@TB
Brayan Rocchio	CLE	Strikeouts	Drew Rasmussen	0.562	0.656	+0.094	Good	Good	20261013_CLE@TB
Travis Bazzana	CLE	Walks	Drew Rasmussen	0.296	0.209	-0.088	Good	Very good	20261013_CLE@TB
Nathaniel Lowe	CLE	Walks	Drew Rasmussen	0.260	0.182	-0.077	Average	Very good	20261013_CLE@TB
Steven Kwan	CLE	Strikeouts	Drew Rasmussen	0.431	0.508	+0.077	Very good	Good	20261013_CLE@TB
Steven Kwan	CLE	Hits	Drew Rasmussen	0.929	0.854	-0.075	Good	Good	20261013_CLE@TB
Chase DeLauter	CLE	Hits	Drew Rasmussen	0.922	0.848	-0.074	Very good	Good	20261013_CLE@TB
Jo Adell	CLE	Hits	Drew Rasmussen	0.859	0.788	-0.071	Average	Good	20261013_CLE@TB
José Ramírez	CLE	Hits	Drew Rasmussen	0.830	0.761	-0.069	Below average	Good	20261013_CLE@TB
Patrick Bailey	CLE	Walks	Drew Rasmussen	0.220	0.154	-0.066	Average	Very good	20261013_CLE@TB
Angel Martínez	CLE	Hits	Drew Rasmussen	0.808	0.743	-0.066	Good	Good	20261013_CLE@TB

Per game

Club Player

20261013_CLE@TB — team layer withheld (values_fresh); runs moved away 0.7279 / home 0.0905

Foster Griffin (LHP, 21.9 BF)
H 0.2244 Below average · SO 0.2193 Average · BB 0.0672 Very good · HR 0.0313 Average

Drew Rasmussen (RHP, 21.8 BF)
H 0.1960 Good · SO 0.2559 Good · BB 0.0557 Very good · HR 0.0247 Good

CLE hitters vs Drew Rasmussen and the bullpen

The pipeline applied no club-wide rescale to CLE hitter lines. The columns below are the measured alternative: each hitter moved by his own KPI pairing with the whole opposing staff, the starter and the full relief corps combined at the share of his plate appearances each takes. Click a name for the starter-only and bullpen decomposition and the contextual readings.

Player	Pos	Faces	PA	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	PA	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Steven Kwan	CF	Drew Rasmussen + bullpen	4.26	0.929 → 0.854	-0.075	0.431 → 0.508	+0.077	0.406 → 0.288	-0.118	0.052 → 0.046	-0.005

José Ramírez	3B	Drew Rasmussen + bullpen	4.35	0.830 → 0.761	-0.069	0.680 → 0.795	+0.115	0.360 → 0.254	-0.106	0.119 → 0.107	-0.013

Chase DeLauter	RF	Drew Rasmussen + bullpen	4.14	0.922 → 0.848	-0.074	0.672 → 0.784	+0.112	0.321 → 0.226	-0.095	0.132 → 0.118	-0.014

Jo Adell	DH	Drew Rasmussen + bullpen	3.99	0.859 → 0.788	-0.071	1.109 → 1.272	+0.163	0.174 → 0.121	-0.053	0.148 → 0.132	-0.015

Nathaniel Lowe	1B	Drew Rasmussen + bullpen	3.70	0.771 → 0.708	-0.063	0.910 → 1.045	+0.136	0.260 → 0.182	-0.077	0.141 → 0.126	-0.015

Angel Martínez	LF	Drew Rasmussen + bullpen	3.83	0.808 → 0.743	-0.066	1.004 → 1.156	+0.152	0.132 → 0.091	-0.041	0.158 → 0.141	-0.017

Travis Bazzana	2B	Drew Rasmussen + bullpen	3.73	0.721 → 0.661	-0.060	0.837 → 0.966	+0.129	0.296 → 0.209	-0.088	0.115 → 0.103	-0.012

Patrick Bailey	C	Drew Rasmussen + bullpen	3.57	0.726 → 0.665	-0.060	1.024 → 1.168	+0.144	0.220 → 0.154	-0.066	0.125 → 0.112	-0.013

Brayan Rocchio	SS	Drew Rasmussen + bullpen	3.38	0.705 → 0.648	-0.058	0.562 → 0.656	+0.094	0.184 → 0.128	-0.056	0.088 → 0.079	-0.009

CLE starter against the TB lineup

Player	Pos	Faces	BF	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	BF	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Foster Griffin	Starter	TB	21.91	5.969 → 6.312	+0.343	4.438 → 3.973	-0.465	1.349 → 1.208	-0.141	0.280 → 0.267	-0.013

CLE bullpen: the full relief corps (7 arms)

Exposure. Starter 21.9 batters faced, bullpen 15.4 (41.2% of the game), from the club's real 37.3 staff batters faced per game over 18 complete games in the last 30 days. The published staff alone would have given the bullpen 35.3% across 2 arms, because it holds only the relievers the selected simulation draw used.

Reliever	Role	In the corps because	Relief BF, 30d	Apps	Share	BF tonight
Joey Cantillo	Long Reliever	pregame bullpen, published staff	45	3	24.0%	3.68
Hunter Gaddis	Reliever	pregame bullpen	53	13	14.1%	2.16
Tim Herrin	Reliever	pregame bullpen	35	9	14.1%	2.16
Erik Sabrowski	Reliever	pregame bullpen	37	10	13.9%	2.14
Shawn Armstrong	Reliever	pregame bullpen	19	6	11.8%	1.82
Cade Smith	Reliever	pregame bullpen	49	12	11.7%	1.80
Tanner Bibee	Middle Reliever	pregame bullpen, published staff	3	1	10.4%	1.60

CLE relievers against the TB lineup

Player	Pos	Faces	BF	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	BF	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Joey Cantillo	Long Reliever	TB (24% of relief)	3.68	0.716 → 0.758	+0.042	2.074 → 1.863	-0.211	0.704 → 0.634	-0.070	0.000 → 0.000	0

Hunter Gaddis	Reliever	TB (14% of relief)	2.16	0.462 → 0.489	+0.027	0.479 → 0.429	-0.050	0.190 → 0.170	-0.019	0.064 → 0.061	-0.003

Tim Herrin	Reliever	TB (14% of relief)	2.16	0.438 → 0.464	+0.026	0.488 → 0.437	-0.051	0.230 → 0.206	-0.023	0.059 → 0.056	-0.003

Erik Sabrowski	Reliever	TB (14% of relief)	2.14	0.389 → 0.412	+0.024	0.715 → 0.650	-0.065	0.274 → 0.247	-0.027	0.055 → 0.053	-0.003

Shawn Armstrong	Reliever	TB (12% of relief)	1.82	0.377 → 0.400	+0.022	0.396 → 0.355	-0.042	0.190 → 0.171	-0.019	0.051 → 0.048	-0.002

Cade Smith	Reliever	TB (12% of relief)	1.80	0.344 → 0.364	+0.021	0.596 → 0.542	-0.054	0.140 → 0.125	-0.015	0.046 → 0.044	-0.002

Tanner Bibee	Middle Reliever	TB (10% of relief)	1.60	0.944 → 0.998	+0.054	0.967 → 0.863	-0.104	0.571 → 0.512	-0.059	0.141 → 0.135	-0.006

TB hitters vs Foster Griffin and the bullpen

The pipeline applied no club-wide rescale to TB hitter lines. The columns below are the measured alternative: each hitter moved by his own KPI pairing with the whole opposing staff, the starter and the full relief corps combined at the share of his plate appearances each takes. Click a name for the starter-only and bullpen decomposition and the contextual readings.

Player	Pos	Faces	PA	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	PA	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Yandy Díaz	DH	Foster Griffin + bullpen	4.36	1.139 → 1.136	-0.003	0.648 → 0.687	+0.040	0.305 → 0.279	-0.026	0.150 → 0.147	-0.002

Jonathan Aranda	1B	Foster Griffin + bullpen	4.15	1.056 → 1.054	-0.003	1.047 → 1.103	+0.056	0.294 → 0.269	-0.025	0.145 → 0.143	-0.002

Junior Caminero	3B	Foster Griffin + bullpen	4.04	0.982 → 0.979	-0.002	0.728 → 0.770	+0.042	0.367 → 0.336	-0.031	0.223 → 0.220	-0.003

Ryan Vilade	RF	Foster Griffin + bullpen	3.77	0.882 → 0.880	-0.002	0.880 → 0.927	+0.047	0.279 → 0.255	-0.024	0.154 → 0.152	-0.002

Jonny DeLuca	CF	Foster Griffin + bullpen	3.88	0.992 → 0.990	-0.002	0.841 → 0.887	+0.047	0.188 → 0.171	-0.017	0.130 → 0.128	-0.002

Chandler Simpson	LF	Foster Griffin + bullpen	3.89	1.132 → 1.129	-0.003	0.392 → 0.418	+0.026	0.141 → 0.129	-0.013	0.039 → 0.038	-0.001

Nick Fortes	C	Foster Griffin + bullpen	3.28	0.822 → 0.820	-0.002	0.515 → 0.546	+0.031	0.147 → 0.134	-0.013	0.089 → 0.088	-0.001

Jorge Mateo	2B	Foster Griffin + bullpen	3.38	0.781 → 0.779	-0.002	1.016 → 1.066	+0.050	0.177 → 0.161	-0.015	0.102 → 0.100	-0.002

Taylor Walls	SS	Foster Griffin + bullpen	3.43	0.751 → 0.749	-0.002	0.828 → 0.872	+0.044	0.289 → 0.264	-0.025	0.053 → 0.052	-0.001

TB starter against the CLE lineup

Player	Pos	Faces	BF	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	BF	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Drew Rasmussen	Starter	CLE	21.82	3.261 → 3.321	+0.060	5.640 → 4.799	-0.841	1.135 → 1.129	-0.006	0.272 → 0.257	-0.015

TB bullpen: the full relief corps (7 arms)

Exposure. Starter 21.8 batters faced, bullpen 13.8 (38.8% of the game), from the club's real 35.6 staff batters faced per game over 17 complete games in the last 30 days. The published staff alone would have given the bullpen 32.0% across 2 arms, because it holds only the relievers the selected simulation draw used.

Reliever	Role	In the corps because	Relief BF, 30d	Apps	Share	BF tonight
Griffin Jax	Reliever	pregame bullpen	8	1	23.3%	3.22
Ian Seymour	Long Reliever	pregame bullpen, published staff	16	3	20.9%	2.89
Kevin Kelly	Reliever	pregame bullpen	23	7	19.1%	2.64
Cam Booser	Middle Reliever (High Leverage)	pregame bullpen, published staff	22	8	16.4%	2.26
Bryan Baker	Reliever	pregame bullpen	25	7	16.4%	2.26
Steven Matz	Reliever	pregame bullpen	26	6	2.0%	0.27
Cole Sulser	Reliever	pregame bullpen	27	7	1.9%	0.27

TB relievers against the CLE lineup

Player	Pos	Faces	BF	Hits		Strikeouts		Walks		Home runs
Player	Pos	Faces	BF	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ	published → KPI	Δ
Griffin Jax	Reliever	CLE (23% of relief)	3.22	0.675 → 0.687	+0.012	0.799 → 0.679	-0.120	0.261 → 0.260	-0.001	0.101 → 0.095	-0.006

Ian Seymour	Long Reliever	CLE (21% of relief)	2.89	1.326 → 1.350	+0.024	1.951 → 1.668	-0.283	0.433 → 0.430	-0.002	0.225 → 0.213	-0.012

Kevin Kelly	Reliever	CLE (19% of relief)	2.64	0.548 → 0.558	+0.010	0.594 → 0.502	-0.092	0.151 → 0.151	-0.001	0.071 → 0.067	-0.004

Cam Booser	Middle Reliever (High Leverage)	CLE (16% of relief)	2.26	0.352 → 0.358	+0.006	0.651 → 0.556	-0.096	0.120 → 0.120	-0.001	0.064 → 0.060	-0.004

Bryan Baker	Reliever	CLE (16% of relief)	2.26	0.415 → 0.422	+0.008	0.562 → 0.478	-0.085	0.203 → 0.202	-0.001	0.058 → 0.054	-0.003

Steven Matz	Reliever	CLE (2% of relief)	0.27	0.059 → 0.060	+0.001	0.051 → 0.043	-0.008	0.021 → 0.021	0	0.010 → 0.009	-0.001

Cole Sulser	Reliever	CLE (2% of relief)	0.27	0.058 → 0.059	+0.001	0.062 → 0.052	-0.009	0.021 → 0.021	0	0.008 → 0.007	0

Measured from published artifacts and stored KPI values. Nothing recomputed. Per-player adjustments are a measurement of the catalog, not the published line.