Completed-season analytics · current edition 2025
Independent MLB player and game analytics built from event-level evidence — CORE rate impact, CROWN cumulative value, PULSE recent form and SIGNAL evidence quality — kept inspectable from the leaderboard all the way down to the plate appearance it came from.
This is a finished season, read backwards. The source publishes twice a year and never mid-season, so there is no in-season 2026 edition here and nothing on this site updates during a season — a deliberate constraint, not a gap.
Highest 2025 CROWN · partial ledger
+55.1runs above replacement·+5.25wins
CROWN provisional
Batting accounts for +50.9 of Judge's +55.1 runs in this partial ledger. Fielding and catching are not measured in this edition.
This total adds only the components that exist. A missing component is not a zero. Position and league adjustments are not applied in this edition.
| Component | Runs | Coverage |
|---|---|---|
| Batting | +50.92 | published component |
| Baserunning | -0.48 | single-identifiable-runner standalone terminal SB/CS/PO/POCS only; multi-runner or unidentified actions, advancement, and strict compound secondaries remain unallocated |
| Fielding | not measured — component does not exist | unavailable — no fielding component exists in this release |
| Catching | not measured — component does not exist | unavailable — no catching component exists in this release |
| Position | not measured — component does not exist | not applied — no positional scarcity model in v1 |
| League | not measured — component does not exist | not applied — no league-strength model in v1 |
| Replacement | +4.61 | published component |
Each node is a CROWN component in runs, placed where it happens.
2021-2025 regular seasons, complete through 2025-09-28 · 2025 postseason: 47 games through 2025-11-01 · CORE Actual: research · CROWN: provisional · PULSE: research · nothing here is validated
Every figure links to the surface it came from. Hitter and pitcher CORE are different questions on different denominators and are never ranked against one another.
Every player published in the 2025 release is reachable by name. You can also browse the complete list, open the ratings, or compare two players.
CORE is rate, not volume: value per role-standard workload. CORE Actual research
View the full ratings →runs above average per 600 PA (B-CORE) · minimum 400 fitted PA
| # | Player | CORE | ±SE | fitted PA |
|---|---|---|---|---|
| 1 | Aaron Judge | +41.6 | ±7.8 | 655 |
| 2 | Shohei Ohtani | +28.9 | ±7.4 | 704 |
| 3 | Cal Raleigh | +26.4 | ±8.5 | 675 |
| 4 | Juan Soto | +26.4 | ±7.4 | 688 |
| 5 | George Springer | +25.2 | ±8.0 | 562 |
| 6 | Kyle Schwarber | +25.0 | ±8.6 | 702 |
| 7 | Nick Kurtz | +24.8 | ±8.3 | 470 |
| 8 | Trent Grisham | +22.2 | ±8.2 | 560 |
SIGNAL evidence band weak, limited by identifiability.
runs prevented above average per 750 batters faced · minimum 400 fitted BF
| # | Player | CORE | ±SE | fitted BF |
|---|---|---|---|---|
| 1 | Paul Skenes | +31.2 | ±7.2 | 733 |
| 2 | Nathan Eovaldi | +28.6 | ±7.3 | 492 |
| 3 | Trevor Rogers | +27.6 | ±7.3 | 423 |
| 4 | Tarik Skubal | +26.2 | ±7.3 | 748 |
| 5 | Cristopher Sánchez | +26.1 | ±7.5 | 807 |
| 6 | Garrett Crochet | +25.7 | ±7.5 | 814 |
| 7 | Freddy Peralta | +23.5 | ±7.5 | 723 |
| 8 | Hunter Brown | +23.1 | ±7.7 | 729 |
SIGNAL evidence band weak, limited by identifiability.
runs prevented above average per 250 batters faced · minimum 150 fitted BF
| # | Player | CORE | ±SE | fitted BF |
|---|---|---|---|---|
| 1 | Randy Rodríguez | +8.5 | ±2.8 | 171 |
| 2 | Adrián Morejón | +7.9 | ±2.5 | 274 |
| 3 | Bryan Abreu | +7.6 | ±2.7 | 254 |
| 4 | Seranthony Domínguez | +7.3 | ±2.8 | 237 |
| 5 | Aaron Ashby | +7.2 | ±2.7 | 261 |
| 6 | Garrett Whitlock | +6.9 | ±2.7 | 262 |
| 7 | Dennis Santana | +6.7 | ±2.1 | 187 |
| 8 | Mike Vasil | +6.4 | ±3.1 | 399 |
SIGNAL evidence band weak, limited by identifiability.
RESEARCH — retrospective, context-limited rate estimates. NOT validated, NOT a WAR replacement, NOT comparable across roles. Park/defense/quality-of-opposition adjustments are NOT yet applied.
Selection: top 8 by published CORE value within each cohort, at the same opportunity floor the ratings route publishes (hitters 400 fitted PA, starters 400 BF, relievers 150 BF); ties broken by player id ascending
CROWN is volume and rate together, above a role-specific replacement baseline. CROWN provisional
Open the CROWN ledger →| # | Player | Role | CROWN runs | Wins | Opportunities | Largest component | Components present |
|---|---|---|---|---|---|---|---|
| 1 | Aaron Judge | batter | +55.1 | +5.25 | 655 | Batting | 3 of 7 |
| 2 | Shohei Ohtani | batter | +47.3 | +4.51 | 704 | Batting | 4 of 8 |
| 3 | Juan Soto | batter | +41.2 | +3.93 | 688 | Batting | 3 of 7 |
| 4 | George Springer | batter | +34.4 | +3.28 | 562 | Batting | 3 of 7 |
| 5 | Paul Skenes | starter | +34.2 | +3.26 | 733 | Pitching | 2 of 6 |
| 6 | Kyle Schwarber | batter | +34.0 | +3.24 | 702 | Batting | 3 of 7 |
| 7 | Cal Raleigh | batter | +32.5 | +3.10 | 675 | Batting | 3 of 7 |
| 8 | Cristopher Sánchez | starter | +32.2 | +3.07 | 807 | Pitching | 2 of 6 |
| 9 | Garrett Crochet | starter | +32.0 | +3.05 | 814 | Pitching | 2 of 6 |
| 10 | Tarik Skubal | starter | +29.9 | +2.85 | 748 | Pitching | 2 of 6 |
this release is NOT better than WAR and is not comparable to it. Three of its four inputs carry a SIGNAL band of weak, fielding and catching do not exist, pitching still contains defense, and position and league adjustments are not applied
Selection: top 10 by published partial CROWN runs; ties broken by player id ascending
The same ledger the ranking is built from, opened up. CROWN provisional
starter · 2025
+34.2runs, partial ledger
Pitching accounts for +30.5 of Skenes' +34.2 runs in this partial ledger. Fielding and catching are not measured in this edition.
| Component | Runs | Running total | Coverage |
|---|---|---|---|
| Pitching | +30.53 | +30.53 | DEFENSE-ENTANGLED: team defense and park effects remain inside this value (component ledger: defense_battery_unallocated). No fielding component may coexist with it |
| Fielding | not measured | — | unavailable — and blocked while pitching is defense-entangled |
| Catching | not measured | — | unavailable — no catching component exists in this release |
| Position | not measured | — | not applied — no positional scarcity model in v1 |
| League | not measured | — | not applied — no league-strength model in v1 |
| Replacement | +3.70 | +34.23 | published component |
| CROWN (partial) | +34.23 | +34.23 | sum of the components that exist |
Bars run left to right from zero: each component starts where the one above it ended. A dashed slot is a component that does not exist in this release.
Each trailing window sits beside that player’s own full-season anchor, so form is never read without its baseline. PULSE research
Open PULSE →| Player | Window vs anchor | Anchor | Trailing 50 PA | Move |
|---|---|---|---|---|
| Tyrone Taylor50 of 330 PA | -0.0437 | +0.1024 | +0.1461 | |
| Josh Naylor50 of 581 PA | +0.0424 | +0.1883 | +0.1459 | |
| Drake Baldwin50 of 419 PA | +0.0662 | +0.2063 | +0.1401 | |
| Bo Naylor50 of 394 PA | -0.0062 | +0.1290 | +0.1351 | |
| Spencer Horwitz50 of 403 PA | +0.0422 | +0.1665 | +0.1243 |
| Player | Window vs anchor | Anchor | Trailing 50 PA | Move |
|---|---|---|---|---|
| Riley Greene50 of 632 PA | +0.0204 | -0.1484 | -0.1688 | |
| Sean Murphy50 of 315 PA | +0.0159 | -0.1464 | -0.1622 | |
| Josh Lowe50 of 419 PA | -0.0200 | -0.1759 | -0.1560 | |
| Ke'Bryan Hayes50 of 542 PA | -0.0447 | -0.1925 | -0.1478 | |
| Carson Kelly50 of 404 PA | -0.0239 | -0.1633 | -0.1394 |
a chronological forward test found these short windows have HIGHER error than a cutoff-frozen league average, so movement here is a record of what happened and carries no demonstrated forecasting value
Selection: hitters with at least 300 season PA whose trailing-50-PA window clears the sample floor, ranked by trailing window minus that player's own full-season anchor; ties broken by player id ascending
Every rating on this site is assembled from plate appearances like this one. Here is the whole accounting for a single swing.
2025-09-28 · Twins 1 · Phillies 2
+0.370win probability, home perspective
The home side went from 0.281 to 0.651 at a leverage index of 2.64, and the play was worth +1.000 runs against the base-out state it began in.
State before
Before: 0 out, bases empty
RE 0.501
State after
After: 0 out, bases empty
RE 0.501
| Field | Value |
|---|---|
| Inning | bottom 8 |
| Batter | Max Kepler |
| Event | Home run · HR/L9D |
| Runs scored | 1 |
| Run expectancy before | 0.5005 |
| Run expectancy after | 0.5005 |
| Run value | +1.0000 |
| Win probability before | 0.2806 |
| Win probability after | 0.6511 |
| WPA | +0.3704 |
| Leverage index | 2.639 |
| Source line | 13191 |
run value is run expectancy after minus run expectancy before, plus runs scored on the play; both expectancies are the published 24-state matrix this game's own run value was computed against win probability and WPA are stated from the home team's perspective; the visiting perspective is the exact complement
Win probability before, after and WPA are each rounded to four decimals in the source payload, so the difference of the first two can sit one unit in the last place from the third. The published WPA is the model's value, not a subtraction of two rounded displays.
Selection: largest absolute win-probability change among plate-appearance-ending events of the most recent fully admitted slate (2025-09-28); ties broken by game id then sequence
SIGNAL rates the evidence behind each construct, never a player’s ability. Six axes stay separate and the overall band is the minimum across them. SIGNAL research
Open SIGNAL →| Construct | Overall | sample | identifiability | granularity | noise | availability | lineage |
|---|---|---|---|---|---|---|---|
| base out run expectancy | moderate | strong | strong | strong | strong | moderatelimiting | strong |
| baserunning impact rate | weak | weaklimiting | moderate | weaklimiting | weaklimiting | weaklimiting | strong |
| batting impact rate | weak | strong | weaklimiting | moderate | moderate | strong | strong |
| event run value | moderate | strong | moderatelimiting | moderatelimiting | strong | strong | strong |
| pitching impact rate | weak | strong | weaklimiting | moderate | moderate | strong | strong |
| win probability neutral | moderate | moderatelimiting | moderatelimiting | strong | moderatelimiting | moderatelimiting | strong |
six axes published SEPARATELY; no weighted composite is accepted as scientific truth; the overall band is the minimum across axes and names its limiting axis; low granularity cannot be erased by high lineage; high sample cannot erase poor identifiability
The event corpus reaches back further than the ratings do, and the rule changes that forbid reading straight across it are declared rather than assumed.
Open the corpus →| Season | Games | Events | Clean games | Conservation |
|---|---|---|---|---|
| 2021 | 2,429 | 187,210 | 2,429 | reconciled |
| 2022 | 2,430 | 187,365 | 2,430 | reconciled |
| 2023 | 2,430 | 190,416 | 2,430 | reconciled |
| 2024 | 2,429 | 188,938 | 2,429 | reconciled |
| 2025 | 2,430 | 189,311 | 2,430 | reconciled |
the event corpus spans every season above; CORE, CROWN, PULSE and SIGNAL are published for 2025 only, so this window shows reconciled coverage and the declared rule-regime breaks — not multi-season ratings, which do not exist here
Compare
Comparison refuses to rank across roles, because the units are not the same question.
Compare these two →Parks
| Park | Factor (100 = neutral) | Home R/G | Road R/G |
|---|---|---|---|
| Rockies | 141.9 | 11.72 | 8.26 |
| Phillies | 115.4 | 9.43 | 8.17 |
| Dodgers | 113.0 | 9.88 | 8.74 |
| Rangers | 81.0 | 7.12 | 8.79 |
| Mariners | 81.8 | 8.11 | 9.91 |
| Marlins | 85.4 | 8.57 | 10.04 |
DESCRIPTIVE ONLY for 2025. A park prior for season t may only use seasons <= t-1, so these factors may serve as priors from season 2026 onward, never for their own season. Multi-year smoothing requires MS-18 backfill.
Open parks →Eras
Every declared rule-regime break records which comparisons it breaks, so a series that crosses one without a flag is treated as invalid rather than interesting.
Open eras →Every sentence in an analytical biography is generated from a measured value and cites the artifact and field it came from. The engine has no path from free text to the page, so it cannot tell you why anything happened.
Read the biographies →In the 2025 regular season Aaron Judge recorded 679 plate appearances with 179 hits, 53 home runs, 137 runs and 114 RBI.
That produced a 0.331 batting average, 0.457 on-base percentage and 0.688 slugging percentage, an OPS of 1.144.
Coverage
2021-2025 regular seasons, complete through 2025-09-28 · 2025 postseason: 47 games through 2025-11-01
The current season is not included. Data as of 2025-09-28.
Sources
16 admitted source captures
Retrosheet event and game-log backbone with a pinned Chadwick identity register. Every capture is hashed and receipted.
Inspect provenance →Status
nothing in this release is validated; every surface carries the status of the artifact behind it
See validation →Every surface answers one research question about the finished 2025 season. Start from the question you are actually asking.
Players
One dossier per published player: season line, rate impact, cumulative ledger, recent form, and how strong the evidence is — with what cannot be measured stated plainly.
Open a player →Compare
Component by component, on the same scale, with missing components shown as missing. No winner sentence — roles are different questions.
Compare players →Teams
Thirty club pages: record against run differential, park environment, and where the batting value actually came from.
Open a club →Games
All 2,430 regular-season games, event by event, with the win-probability path and every plate appearance traceable to its source line.
Open a game →Seasons
Conventional 2025 counting lines, and how scoring moved across the 2021–2025 reconciled record.
Season stats → · History →Context
Descriptive run environments for every 2025 park, and the rule changes that stop seasons being read as one era.
Parks → · Eras →Honors
Awards granted only where the cited metric's evidence clears the bar — and the blocked ones published with their reasons.
Honors → · Biographies →Method
What each metric family measures, what it refuses to claim, and the validation record behind the parsed corpus.
Methodology → · Validation →CORE Actual measures rate impact per role-standard workload. CROWN accumulates total value above a role-specific replacement baseline. PULSE records recent form beside a full-season anchor. SIGNAL rates how strongly the evidence supports each construct. Every number on this site carries its unit, its model, its status and its source lineage.
No odds, picks or betting content. No fantasy advice. No automated collection from MLB digital properties. No republication of proprietary metrics. No fabricated pitch, tracking or contact data — a surface without an admitted lawful source says so plainly rather than filling the gap.