Ratings Methodology

League constants, run-value weights, and engine diagnostics for the current season selection. Use the watchlists below to jump straight into a team's ratings tab and inspect its per-game adjusted detail. The shipped JSON endpoints are documented in the API guide.

Runs / PA

0.1523

League run environment per plate appearance.

Runs / Out

0.2430

Core environment scaler used to move MLB priors toward league conditions.

Box Coverage

88%

Share of team-game rows with box-score detail feeding the constants layer.

Engine Version

v3_additive_iterative_10

Current schedule-adjustment model powering team ratings.

Non-Converged Teams

Teams that finished the iterative engine above the convergence threshold.

Low Rated Coverage

Teams with less than 80% of scored games resolved into the adjusted engine.

High Final Delta

Teams whose last iteration still moved by more than 0.05 runs per game.

League Run Values

0.454

Run value of an unintentional walk.

HBP

0.454

Run value of a hit-by-pitch.

0.667

Run value of a single in the current run environment.

1.093

Run value of a double.

1.519

Run value of a triple.

1.988

Run value of a home run.

0.284

Run value of a stolen base.

-0.568

Run penalty of a caught stealing.

wOBA Weights

wBB

0.980

League-scaled wOBA coefficient for walks.

wHBP

1.025

League-scaled wOBA coefficient for HBP.

w1B

1.261

League-scaled wOBA coefficient for singles.

w2B

1.805

League-scaled wOBA coefficient for doubles.

w3B

2.295

League-scaled wOBA coefficient for triples.

wHR

2.983

League-scaled wOBA coefficient for home runs.

lg wOBA

0.497

League-average wOBA for the selected season.

wOBA Scale

0.792

Scale factor used to convert wOBA differences into runs.

Player Offense Layer

Primary Sort

Off/G

Player rankings now lead with offensive runs above average per game from `player_offense_adjusted`.

Off Runs

wRAA

First-pass `off_runs` is implemented as wRAA using season-native wOBA weights and wOBA scale.

wRC+

Support metric

Still shown for familiarity, but now calculated from season-native constants rather than fixed stock weights.

BB%

NIBB / PA

Player walk rate now uses non-intentional walks when IBB data is present, which better reflects earned plate-discipline value.

XBH Splits

100%

When player-level 2B/3B splits are missing, unsplit non-HR XBH are allocated using the observed season 2B/3B mix, and the mart now flags exact versus estimated split rows.

Qualifier

50 PA

The mart keeps every player row and exposes a `qualified` flag. Product views can filter or annotate from that flag.

Fallback

Legacy safe

If `player_offense_adjusted` is unavailable, product surfaces fall back to `hitting_advanced` and keep the page online.

Player Pitching Layer

Primary Sort

Pitch/G

Pitcher rankings now lead with runs saved per 9 innings versus the season league-average RA9.

Pitch Runs

RA9-based

First-pass `pitch_runs_saved` is based on RA9 relative to the season league average and scaled by workload.

FIP-

Support metric

FIP is retained, with `FIP-` indexed to league average so lower is better and 100 is average. FIP now uses non-intentional walks when IBB data is present.

K-BB%

NIBB-aware

Walk-based rates and K-BB% now use non-intentional walks where IBB detail exists, improving comparability for richer sources.

BABIP-A

Support metric

BABIP-against is now surfaced as a contact-management support metric using hits in play and outs in play.

LOB%

Support metric

LOB% is now exposed as a sequencing / strand-rate context metric, not a standalone rating.

GO/FO

Detail signal

GO/FO is surfaced where GO and FO fields are available, helping richer NCAA/NWAC rows show batted-ball shape.

Qualifier

10 IP

The mart keeps every pitcher row and exposes qualification separately instead of filtering in SQL.

Fallback

Legacy safe

If `player_pitching_adjusted` is unavailable, product surfaces fall back to `pitching_advanced`.

Player Baserunning Layer

Primary Sort

BsR/G

Baserunning now has its own player-facing layer instead of living only as an offense side field.

wSB

Relative running value

Uses season-native SB/CS run values, but now scores players relative to the league-average running value expected from their on-base opportunities.

wGDP

GO proxy context

GDP value now uses a ground-ball opportunity proxy when GO data exists, making richer NCAA/NWAC rows more informative than lower-detail sources.

BsR

wSB + wGDP

First-pass standalone baserunning score combines running value and GDP burden.

Opp Proxy

1B + NIBB + HBP

Running opportunity proxy is built from singles, non-intentional walks, and hit-by-pitches instead of using PA directly.

Qualifier

50 PA

Shares the current hitter qualifier so baserunning views can stay aligned with player opportunity.

Fallback

Layered safe

If `player_baserunning_adjusted` is unavailable, product surfaces fall back to baserunning fields derived from player offense or legacy hitting views.

Adjustment Engine

Teams Rated

Number of teams present in the adjusted engine output.

Converged Teams

Teams whose final-step delta finished below the convergence threshold.

Avg Iterations

10.0

Mean number of additive adjustment passes recorded in the final mart.

Worst Delta

0.107100

Largest final-step change among all teams in the selected season.

Avg Rated Coverage

91%

Average share of games with an opponent that resolved into the adjusted engine.

Avg SOS

-0.743

Average opponent adjusted run margin across the league.

Raw Audit Summary

Avg Def Signal

0.000

Average direct game-level raw defense signal across all rated teams in the selected season.

Avg |Def Resid|

1.339

Average absolute cleanup needed after the direct defense signal is stacked back into raw run margin.

Worst |Def Resid|

4.456

Largest absolute raw defense cleanup among team seasons in the selected season.

Residual > Signal

Teams where the raw defense cleanup magnitude exceeds the direct defense signal magnitude.

Avg BsR Player

-0.002

Average player-layer baserunning contribution per game before blending with game signals.

Avg BsR Game

0.272

Average GDP-aware game-level baserunning contribution per game in the raw layer.

Avg BsR Sig %

Average share of scored team games with direct GDP context available for the raw baserunning signal.

Avg Def Sig %

Average share of scored team games with direct defensive signal support from fielding counts or opponent GDP context.

Avg Opp GDP %

Average share of scored team games where opponent GDP context was available to enrich the defense signal.

Validation Summary

AdjRM Corr

0.919

Correlation between adjusted run margin and team win percentage for the selected season.

RawRM Corr

0.954

Correlation between raw run margin and team win percentage, shown as a baseline against the adjusted engine.

Adj Gain

-0.034

Difference between adjusted and raw run-margin correlation with win percentage.

Adj Corr 80%+ Cov

0.923

Adjusted run-margin correlation with win percentage for teams with at least 80% rated-game coverage.

Avg Win Error

3.19

Average absolute difference between expected wins from the adjusted engine and actual wins.

Worst Win Error

11.30

Largest absolute expected-wins miss among teams in the selected season.

Avg |Luck|

3.19

Average absolute luck term from the adjusted engine. Lower suggests the expected-win layer is matching observed records more closely.

|Luck| >= 2

Teams whose expected wins differ from actual wins by at least two games.

|Win Err| >= 3

Teams whose expected-win miss is at least three games.

Validation Breakdown

80%+ Cov |Err|

3.44

Average absolute expected-win miss for teams with at least 80% rated-game coverage.

<80% Cov |Err|

1.10

Average absolute expected-win miss for teams with weaker adjusted-game connectivity.

Converged |Err|

—

Average absolute expected-win miss for teams whose final iteration converged.

Non-Conv |Err|

3.19

Average absolute expected-win miss for teams that still had meaningful movement left at the final iteration.

High-Conf |Err|

—

Average absolute expected-win miss for converged teams with at least 80% rated-game coverage.

Low-Conf |Err|

3.19

Average absolute expected-win miss for teams with weak coverage or a non-converged result.

Low Coverage Teams

Teams below the 80% rated-game coverage line in the validation layer.

Non-Converged Teams

Teams whose validation rows still carry a non-converged engine result.

Method Notes

Raw Layer

team_ratings_raw

Builds run-differential components from weighted player-adjusted offense and pitching, blends raw baserunning with GDP-aware game signals, and anchors raw defense to a direct game-level proxy before the final reconciliation cleanup.

Raw Coverage Signals

3 shares

Team surfaces now expose offense split-detail coverage, pitching contact-detail coverage, and baserunning GDP-context coverage so richer NCAA/NWAC-backed rows are visible in product.

Raw Audit Fields

Def/BsR split

Team views now show direct raw defense signal, defense residual cleanup, and player-versus-game baserunning contributions so the raw decomposition is inspectable.

Adjusted Layer

team_ratings_adjusted

Applies additive opponent adjustment iteratively across all four components, with offense/pitching and baserunning/defense paired against complementary opponent strengths.

Game Detail

ratings_game_detail

Centers game-level offense, pitching, defense, and baserunning proxies back to each team’s raw season values, with GDP-over-expected and error-context now feeding the defense/baserunning side of the engine.

Player Layers

3 marts

Player-facing product views now read `player_offense_adjusted`, `player_pitching_adjusted`, and `player_baserunning_adjusted` first.

XBH Split Rate

100%

Share of non-HR extra-base hits with observed 2B/3B split detail in this season.

Home Margin

0.653

Observed average home-field run margin from scored games.

Environment Scale

1.420

Run-environment multiplier relative to the MLB baseline used in v1 constants.

Convergence Watchlist

Teams with the largest remaining final-step movement. Open any row to inspect the ratings tab for game-level detail.

Team	AdjRM	Rated %	Iter	Delta	Conv
Walla Walla NWAC	-4.51	100%	10	0.1071	N
Blue Mountain NWAC	-4.21	100%	10	0.1057	N
Yakima Valley NWAC	-0.16	98%	10	0.1056	N
Centralia NWAC	-1.76	100%	10	0.1044	N
Skagit Valley NWAC	-3.25	100%	10	0.1031	N
Tacoma NWAC	-0.78	97%	10	0.1024	N
Olympic NWAC	-2.72	100%	10	0.1003	N
Spokane NWAC	1.51	96%	10	0.0990	N

Coverage Watchlist

Teams with the thinnest adjusted-game coverage. These ratings may still be directionally useful, but they are less fully connected.

Team	AdjRM	Rated %	Iter	Delta	Conv
Douglas NWAC	-4.99	62%	10	0.0674	N
Clackamas NWAC	-1.41	74%	10	0.0370	N
Mt Hood NWAC	0.45	77%	10	0.0369	N
Linn-Benton NWAC	2.17	82%	10	0.0445	N
Treasure Valley NWAC	-2.18	84%	10	0.0845	N
Lane NWAC	3.24	84%	10	0.0495	N
Lower Columbia NWAC	2.32	86%	10	0.0853	N
Grays Harbor NWAC	-7.99	88%	10	0.0912	N

Raw Audit Watchlist

Teams whose raw defense needed the largest cleanup after the direct game-level signal was applied.

Team	RawDef	Def Resid	RawBsR	BsR P/G	Def Sig %
Grays Harbor NWAC	-4.46	-4.46	-0.11	-0.11 / -0.12	0%
Walla Walla NWAC	-3.21	-3.21	0.02	0.02 / 0.02	0%
Big Bend NWAC	-2.92	-2.92	0.07	-0.02 / 0.23	0%
Douglas NWAC	-2.84	-2.84	0.09	0.06 / 0.16	0%
Spokane NWAC	2.52	2.52	0.23	0.07 / 0.54	0%
Everett NWAC	1.85	1.85	0.37	0.22 / 0.66	0%
Lower Columbia NWAC	1.80	1.80	-0.14	-0.29 / 0.15	0%
Skagit Valley NWAC	-1.65	-1.65	0.01	-0.04 / 0.11	0%

Validation Watchlist

Teams with the largest gap between expected wins from the adjusted engine and actual wins.

Team	W-L	AdjRM	Exp W	Luck	\|Err\|	Rated %	Risk
Blue Mountain NWAC	18-29	-4.21	6.7	11.3	11.3	100%	high 95%+ / drifting
Olympic NWAC	19-29	-2.72	8.4	10.6	10.6	100%	high 95%+ / drifting
Treasure Valley NWAC	23-27	-2.18	16.0	7.0	7.0	84%	high 80-95% / drifting
Chemeketa NWAC	11-35	-1.43	17.6	-6.6	6.6	91%	high 80-95% / drifting
Centralia NWAC	21-27	-1.76	15.6	5.4	5.4	100%	high 95%+ / drifting
Yakima Valley NWAC	30-21	-0.16	24.9	5.1	5.1	98%	high 95%+ / drifting
Lane NWAC	39-18	3.24	43.8	-4.8	4.8	84%	high 80-95% / close
Big Bend NWAC	16-30	-4.13	11.3	4.7	4.7	91%	high 80-95% / drifting

How To Read The Flags

Converged = N means the iterative engine still had a meaningful rating move left at the final pass. That does not make the rating unusable, but it is a good cue to inspect opponent connectivity and game-by-game detail.

Rated coverage tracks how many scored games had an opponent that resolved into the adjusted engine. Lower coverage usually means more schedule isolation or incomplete connectedness.

Matchup detail lines summarize how much richer raw offense split detail, pitching contact detail, and baserunning GDP-context support each team brings into the pregame cards.

Matchup audit lines now also split raw defense into direct signal versus cleanup and raw baserunning into player versus game-signal contributions, so matchup cards expose the same raw audit language as team pages.

Validation summary compares how well `AdjRM` and `RawRM` track win percentage, while the validation breakdown splits expected-win miss size by engine confidence so weak coverage or non-convergence is easier to spot season-wide.

Adjusted defense and baserunning now come from centered game-level proxies that include GDP-over-expected context alongside the earlier SB/CS and error signals, so those components can participate in opponent adjustment instead of remaining static season carry-through fields.

Team pages now support a direct tab=ratings link target, so the watchlists on this page can jump directly to the per-game adjusted detail behind a team's `AdjRM`.