Context-neutral player and team value in runs, 2015–2026, and the 2027 forecast: every element, how it was derived, and how it was tested.
The MLB Run Ledger values every major-league player in runs above or below a league-average player, per plate appearance for hitters and per batter faced for pitchers, and adds players up to teams. It is built from the complete MLB Stats API play-by-play for every regular-season game from 2014 through 2026, rebuilt one plate appearance at a time, plus FanGraphs fielding, positional and projection inputs. The dashboard shows the ledger for 2017–2026 (2017 is FanGraphs' first prospect board) and a forecast for 2027.
The value is context-neutral: an event is worth what that event is worth on average across all base-out situations, so a hitter is not credited for when his hits happened, a pitcher is not charged for his fielders, and a runner is credited for what he did on the bases relative to an average runner on the same kind of ball. Every component adds to zero across the league, and the players on a team add up to a team total that tracks actual run differential.
The 2027 forecast combines three seasons of each player's ledger, pulls every component toward average by an amount fitted to how much it repeats (pitchers' strikeouts, walks and contact toward what their pitch tracking predicts instead), ages each component with curves fitted on 2015–2026, takes playing time and teams from FanGraphs Depth Charts, and turns team totals into projected wins. In an out-of-sample test on 2026 it cut the typical miss for regular hitters from 27.6 runs per 600 PA (using last year's rate) to 16.9, and for pitchers from 17.0 to 13.0 runs per 600 batters faced.
Minor leaguers enter the forecast through a translation of their Triple-A and Double-A seasons (Section 14): the same ledger built from minor-league play-by-play, carried over to MLB piece by piece with weights fitted on the players who made the jump. Fitted on earlier seasons and tested on 2025–26, it cut the typical miss for hitters coming straight from the minors from 21.7 to 20.3 runs per 600 PA.
FanGraphs WAR is shown beside the ledger's runs throughout the dashboard for reference; Section 11 explains how the two scales relate (a full-time average player is about 2 WAR) and how team WAR is built. Section 16 sets out what the ledger borrows from other methods and what it does differently.
The dashboard's Method view is the short version. This paper documents every element, how it was derived, the tests that chose between alternatives, and the known limitations.
Each plate appearance is valued once from the offense's side (batter and runners) and once from the defense's side (pitcher, catcher and fielders). Both sides use the same run values, so the league's batting runs add to zero, the league's pitching runs add to zero, and a team's offense minus its run prevention is a run differential. Values are always relative to the league average for the season, never to a replacement level, so a full-time average player is worth zero.
Components add to a player's runs; players add to a team. Everything not credited to a player (timing, home runs beyond what the contact predicted, batted-ball conversion that fielding metrics do not assign, the pitching side of the park) is collected in one unassigned layer, so the books always balance against actual run differential.
Hitters are rated per plate appearance, pitchers per batter faced, and both are reported as season totals and per 600 units, so part-time players compare with regulars and season totals still add up.
Every hitter is shown against left- and right-handed pitching, every pitcher against left- and right-handed batters, every team lineup against each pitcher hand, and every team staff against each batter hand. Parts of a player's value that do not depend on the matchup are spread by playing-time share, so the two splits always add up to the total.
The positional adjustment is shown for comparing a shortstop with a first baseman, but it is left out of team sums: every team fields one of each position, so the adjustments cancel in a team total and adding them would only add noise.
| Source | What is used | Seasons |
|---|---|---|
MLB Stats API game/{pk}/playByPlay | Every play: event type, base runners and their movements (start, end, out, reason), outs, score, batter and pitcher with handedness, exit velocity, launch angle and trajectory of the batted ball | 2014–2026 |
| MLB Stats API schedule, standings, teams, people | Final regular-season games, wins and losses, team names, player names, birth dates, bats and throws | 2014–2026 |
| FanGraphs leaders API (players) | DRS and its parts, Statcast fielding parts, positional adjustment; WAR, batting and baserunning runs for comparison (WAR is also shown beside the ledger in the dashboard) | 2015–2026 |
| FanGraphs leaders API (teams) | Team totals used in the Brewers study and the fielding-metric test | 2019, 2021–2026 |
| FanGraphs Depth Charts projections | 2027 playing time (PA, batters faced, games, starts, innings), team and projected WAR | 2027 |
| FanGraphs RosterResource depth-chart pages | Check that the projection feed's teams match current rosters | 2027 |
All requests are plain HTTPS. FanGraphs answers curl but refuses Python's default client, so the FanGraphs scripts shell out to curl. The play-by-play is requested with a field filter that keeps about one fifth of the full feed (about 13 MB compressed per season).
Each play in the feed is either a plate appearance (single, walk, strikeout, field out, fielder's choice, error, sacrifice, interference and so on: 26 event types) or a play that ends without one (a caught stealing or pickoff that ends an inning, a walk-off wild pitch). Within a plate appearance, runner movements are split by the index of the pitch they happened on:
The base state is tracked by runner id through each half-inning. Three details matter:
One data quirk is handled explicitly: once (2023-10-01, game 716358) the feed records a run as a movement ending at 4B instead of score; it is treated as a run.
Every game is checked three ways: after each plate appearance the tracked base state must match the feed's own post-play runners, the outs must match the feed's count, and the runs from plate appearances and runner events must add to the final score. All 30,052 regular-season games from 2014 to 2026 pass all three checks with zero mismatches.
| Season | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Final games | 2,430 | 2,429 | 2,428 | 2,430 | 2,431 | 2,429 | 898 | 2,429 | 2,430 | 2,430 | 2,429 | 2,430 | 2,429 |
The rebuild produces three tables per season: one row per plate appearance (about 182,000 a season), one row per runner per plate appearance with start base, destination and out reason (about 290,000), and one row per mid-PA runner event (about 6,000).
For each season, the expected runs from each of the 24 base-out states to the end of the half-inning are the average actual runs scored from that state, using innings 1–8 (later innings are cut short by walk-offs and skipped bottom halves). The 2026 table:
| Runners | 0 outs | 1 out | 2 outs |
|---|---|---|---|
| empty | 0.503 | 0.269 | 0.105 |
| 1B | 0.887 | 0.537 | 0.238 |
| 2B | 1.181 | 0.663 | 0.316 |
| 1B 2B | 1.539 | 0.917 | 0.455 |
| 3B | 1.440 | 0.908 | 0.369 |
| 1B 3B | 1.767 | 1.181 | 0.526 |
| 2B 3B | 2.053 | 1.357 | 0.508 |
| loaded | 2.511 | 1.642 | 0.744 |
The run value of a plate appearance is its change in run expectancy plus the runs it scored:
RE24(PA) = RE(end state) - RE(start state) + runs scored on the PA (RE = 0 after the third out)
The start state is the state after any mid-PA runner events, so the batter is never charged or credited for a steal or wild pitch during his plate appearance. The league mean RE24 per PA, μ, is slightly negative (−0.0035 in 2026) because runner events between pitches carry some of the run expectancy; every value below is expressed relative to μ, so the league average PA is worth exactly zero.
| Event (LW1) | Runs vs average PA, 2026 | Count |
|---|---|---|
| Home run | +1.398 | 5,575 |
| Triple | +1.071 | 671 |
| Double | +0.762 | 7,493 |
| Single | +0.474 | 26,110 |
| Reached on error | +0.449 | 1,041 |
| Hit by pitch | +0.357 | 2,118 |
| Interference | +0.337 | 83 |
| Walk | +0.335 | 15,792 |
| Intentional walk | +0.234 | 545 |
| Bunt out | −0.191 | 1,136 |
| Fly-ball out | −0.249 | 24,860 |
| Line-drive out | −0.262 | 11,006 |
| Strikeout | −0.272 | 40,700 |
| Ground-ball out | −0.275 | 37,880 |
| Popup out | −0.279 | 8,839 |
Valuing outs by type matters little in total (a strikeout costs about the same as a ground-ball out once double plays and runner advancement are averaged in), but it keeps the batting value honest and makes the decomposition below exact.
For every plate appearance the gap between its actual value and its wOBA-style value splits exactly into three layers:
RE24 - LW0 = (LW1 - LW0) detail: events wOBA lumps together (context-neutral)
+ (LW2 - LW1) timing: the base-out state the event came in (context)
+ (RE24 - LW2) play: what happened beyond the event class in that state
(runner advancement, double plays, errors on the play)
This identity is the backbone of the ledger: the detail layer stays with the batter, the timing layer is left out, and the play layer is credited to whoever was responsible (runners, the batter on ball-in-play outs, or nobody).
The model began with a question: did the 2026 Brewers (103–59, +214 run differential, World Series champions, projected for about 81 wins) do the little things right in ways projection systems miss? Their record matched their Pythagorean expectation exactly (103.0), so the gap was in runs, not in turning runs into wins, and FanGraphs WAR accounted for only about 96 of the 103 wins.
The first measurement, FanGraphs' team RE24 minus wRAA, put Milwaukee at +60 runs of situational value and showed the gap persisting from year to year across teams (r about 0.6). That persistence turned out to be a ballpark artifact: FanGraphs' RE24 is park-adjusted and its wRAA is not. The residual of FanGraphs' RE24 against our unadjusted RE24 is about −90 runs for Colorado and correlates 0.85–0.87 with FanGraphs' own park adjustment (Batting minus wRAA). Subtracting the two FanGraphs figures mixes a park factor into the gap. The ledger therefore computes RE24 itself and never subtracts FanGraphs' two numbers.
| Brewers offense | 2026 | 2025 | Repeats across teams (r, 2025→2026) | In a context-neutral ledger? |
|---|---|---|---|---|
| Gap, RE24 minus wOBA-style runs | +49.3 (1st) | +43.5 | 0.20 | |
| Detail (outs by type, reached on error) | −2.0 | +7.2 | 0.10 | Yes |
| Timing | +14.3 | +8.3 | −0.08 | No |
| Play, on hits (runner advancement) | +10.2 | +9.7 | 0.02 | Yes, to runners |
| Play, on ball-in-play outs | +24.7 (1st) | +20.5 | 0.41 | Yes, to batters |
The persistent piece was value on ball-in-play outs: chiefly avoiding double plays (34.5% of double-play-chance grounders became double plays against a 45.1% league rate; +13.5 runs at league-average opportunity rates, 1st in MLB; year-to-year r = 0.39 across teams) and scoring the runner from third on outs (+6 in 2026, +12 in 2025, but r = 0.03, mostly noise). Timing did not repeat at all. Reaching on error was not a factor; neither the Brewers' hitters nor their ballpark stood out.
| Brewers run prevention | 2026 | 2025 | Repeats (r) |
|---|---|---|---|
| Strikeouts and walks | +73 (3rd) | +19 | 0.59 |
| Contact quality allowed | +36 (6th) | +41 (2nd) | 0.45 |
| Fielders' batted-ball conversion | +32 (5th) | +19 | 0.71 |
| Running game | +5 | +3 | 0.58 |
| On the play | +7.5 | +3 | 0.04 |
| Home runs beyond expected | −29 (28th) | 0 | 0.38 |
| Timing | −10 | +5 | −0.03 |
| Total (actual) | +115 (+108) | +89 (+87) |
Two conclusions shaped the model. First, the context layers that do not repeat (timing, home runs beyond expected) should be left out of player value, and the ones that do (double-play avoidance, contact quality, conversion) should be measured and credited to the right player. Second, contact quality matters: Milwaukee's staff was top six at suppressing contact in both seasons with a 40% different staff, which FIP-based measures cannot see. That is why pitchers are credited with the expected value of the contact they allow, not just strikeouts, walks and home runs.
A hitter's runs are the sum of four components (five for catchers): batting, situational outs, baserunning, fielding, and catcher defense. All are runs relative to an average player and all add to zero across the league.
Batting is the LW1 value of each plate appearance minus a park adjustment:
batting(PA) = LW1(event class) - mu - (PF(park) - 1) x league runs per PA
PF(team) = 1 + 0.5 x ( [runs per PA in that team's home games, both clubs, two seasons]
/ [runs per PA in that team's road games, both clubs, two seasons] - 1 )
The park factor pools the season and the one before (so 2015 uses 2014) and is regressed halfway to neutral because two seasons of home/road splits are noisy. Runs include runner-event runs. After the adjustment the league total is recentered to exactly zero. The 2026 factors at the ends of the range:
| Park | PF 2026 |
|---|---|
| COL | 1.105 |
| ATH | 1.089 |
| PHI | 1.072 |
| LAD | 1.034 |
| WSH | 1.029 |
| … | |
| SD | 0.957 |
| MIA | 0.948 |
| STL | 0.948 |
| TEX | 0.936 |
| SEA | 0.935 |
Pitchers' values are not park-adjusted, because their main inputs (strikeouts, walks and the expected value of contact from exit velocity and launch angle) depend little on the park. The park's effect on run prevention lands in the unassigned layer.
This component carries the repeatable part of the Brewers finding. On a hitter's ball-in-play outs, the play layer (RE24 minus LW2) is the value of what happened beyond the average for that kind of out in that base-out state: double plays avoided or hit into, forced runners, runners moved up. After runner credits are taken out (below), the remainder belongs to the batter, together with the deviations of runners forced on his ground-ball outs (a double play is the batter's doing, not the runner's).
The opportunities are plate appearances with runners on and fewer than two outs. To avoid rewarding a hitter for batting behind high-OBP teammates, the value is rated per opportunity and applied at the league-average rate of those opportunities:
situational = (situational runs / situational PAs) x 0.257 x PA (25.7% of league PAs are situational in 2026)
The result is recentered to zero per PA. Situational value is the least repeatable hitter component (year-to-year r about 0.20 per PA for regulars), and the forecast pulls it almost entirely to average.
Baserunning combines advancement on balls in play and the running game.
Advancement. For every runner on every plate appearance except home runs (including the batter-runner), the destination is compared with what an average runner achieved in the same cell:
cell = (event class, start base-out state, outs, runner's base, exit-velocity band)
exit-velocity bands: under 85, 85-95, 95+ mph; the band is dropped when a cell has fewer than 30 runners
d_adv = bases advanced (0 if out) - cell average
d_out = 1 if out on the bases, else 0, - cell out rate
The run value of an extra base and of an out on the bases are not assumed: they are fitted each season by regressing every plate appearance's play-layer residual on the sum of its runners' deviations (no intercept). The fitted weights are stable across seasons:
| Season | Runs per extra base | Runs per out on the bases |
|---|---|---|
| 2015 | +0.221 | −0.345 |
| 2016 | +0.230 | −0.324 |
| 2017 | +0.227 | −0.363 |
| 2018 | +0.231 | −0.338 |
| 2019 | +0.230 | −0.380 |
| 2020 | +0.213 | −0.403 |
| 2021 | +0.222 | −0.361 |
| 2022 | +0.225 | −0.339 |
| 2023 | +0.234 | −0.351 |
| 2024 | +0.233 | −0.347 |
| 2025 | +0.213 | −0.387 |
| 2026 | +0.225 | −0.365 |
Each runner's credit is w_adv x d_adv + w_out x d_out. Credits go to the runner, except the batter-runner's own deviation (to the batter: baserunning on hits, situational on outs) and runners forced on ground-ball outs (to the batter, situational). Whatever the runner credits leave of the play residual on hits stays unassigned (much of it is outfield arms and hit location).
Running game. Steals, caught stealing and pickoffs are valued by their change in run expectancy (split equally among the runners on a double steal). Wild pitches, passed balls and balks are the battery's, not the runner's.
Opportunity normalization. A runner's chances depend on how often teammates put the ball in play behind him. Advancement is rated per opportunity (runner-on-base plate appearances) and applied at the league rate of opportunities per stint on base; steals are centered on the league steal value per stint:
advancement = (credits / opportunities) x (league opportunities / league stints) x stints steals = steal runs - league steal runs per stint x stints stint = one continuous stay on base (game, half-inning, runner)
The baserunning total is recentered to zero per stint and split between pitcher hands by the share of the runner's opportunities against each hand.
Fielding is the one major component the play-by-play cannot assign to individual fielders without spray angles and fielder positioning, so it comes from published metrics. To choose between Defensive Runs Saved and Statcast's fielding run value, each team's non-catcher fielding under each metric was compared with the ledger's own team measure of batted-ball conversion (actual results minus expected results from exit velocity and launch angle, plus the play layer on defense):
| Non-catcher fielding metric | r with team conversion, 2025 | r, 2026 |
|---|---|---|
| Statcast FRV (range + arm + infield DP + 1B receiving) | 0.47 | 0.76 |
| DRS minus catcher parts (rSZ, rCERA, rSB) | 0.86 | 0.82 |
| 50/50 blend | 0.78 | 0.86 |
DRS tracks what actually happened on batted balls best, which suits a ledger of what happened. Part of that agreement is by construction (DRS credits plays made by zone, which is closer to an outcome measure than Statcast's catch-probability model), so for forecasting a tracking-based metric may be the better input; the dashboard shows the Statcast figure in every player's breakdown as a second opinion.
fielding(player) = sum over non-pitcher positions of [DRS - rSZ - rCERA - rSB]
- league DRS per inning x innings at non-pitcher positions
pitcher fielding = DRS at P - league pitcher DRS per inning x innings at P
DRS does not add to zero across the league (it averaged about +19 runs per team in 2026), so it is centered per defensive inning, separately for position players and pitchers. Fielding is hand-neutral and spread across a hitter's splits by plate-appearance share. Traded players' fielding is allocated to their clubs by plate-appearance share.
Catchers are credited with Statcast framing, blocking and throwing (FanGraphs API fields fFRP, bFRP, tFRP), centered per catcher inning. Those runs show up in the pitching numbers too (framing changes strikeouts and walks; throwing and blocking change the running game), so the same runs are taken back out of the team's pitchers, in proportion to batters faced:
pitcher strikeouts & walks -= team catchers' framing x (pitcher BF / team BF) pitcher running game -= team catchers' (throwing + blocking) x (pitcher BF / team BF)
Statcast catcher data begins in 2016; 2015 catcher defense is zero.
Field-name trap. In FanGraphs' fielding API, fFRP is framing and cFRP is first-base receiving, not catcher framing. The mapping was verified against the CSV exports (within about a run on every row).
The positional adjustment is FanGraphs' Positional runs. It is shown as a separate column and added only in "Value with position" for comparing players across positions; it is never part of team totals.
A pitcher's runs prevented are the sum of strikeouts and walks, contact quality, running game, and his own fielding. Each plate appearance's value from the defensive side splits exactly into layers; the pitcher owns the first two.
for each PA (sign flipped so positive = runs prevented):
strikeouts & walks : K, BB, IBB, HBP at their LW1 value; every batted ball at the league
average batted-ball value b
contact quality : xv - b, the expected value of the batted ball minus the league average
conversion : what the fielders made of it (fielders, park, luck)
HR vs expected : home runs beyond what the contact predicted (park, luck)
timing : LW2 - LW1 (context)
on the play : RE24 - LW2 (fielders and runners)
Every batted ball's expected value comes from the league average of balls hit with the same exit velocity and launch angle, in 3 mph × 4° cells; sparse cells fall back to 6 mph × 8°, then to the trajectory average, then to the overall average (a cell needs at least 25 balls and at least one that stayed in the park). Bunts are kept apart from swings. About 0.4% of batted balls lack exit velocity or launch angle and use their trajectory's average.
The residual must be split carefully. Splitting by the actual result makes every in-park ball look like a fielding win and every home run a loss (about ±110 runs per team, offsetting). Instead each cell gives a home-run probability p and the mean value m of the balls that stayed in:
xv = p x v(HR) + (1 - p) x m HR part = (1[HR] - p) x (v(HR) - m) averages zero within a cell in-park part = v1 - m (non-home-runs only) averages zero within a cell
Steals, caught stealing, pickoffs, wild pitches, passed balls and balks while a pitcher is on the mound are valued by their change in run expectancy against the league rate per batter faced; the catchers' throwing and blocking share is then removed (above). The pitcher's own fielding is his DRS at pitcher, centered. Strikeouts and walks include his share of his catchers' framing, removed.
Batted-ball conversion (fielders), home runs beyond what the contact predicted, and timing are left out of pitcher value. In the Brewers decomposition these layers carried −29 (home runs beyond expected) and −10 (timing) runs in 2026, luck that worked against the club and that the ledger correctly does not charge to the staff.
Season splits are descriptive and unregressed. A part-time hitter's 120 PA against left-handers say little on their own; the forecast regresses each player's split gap toward the typical gap for his handedness (below).
| Group | Rate vs right-handers minus rate vs left-handers, runs per 600 | Share of PA/BF against left-handed opponents |
|---|---|---|
| Left-handed hitters | +15.9 | 22.8% |
| Right-handed hitters | −7.6 | 30.7% |
| Switch hitters | −0.9 | 28.8% |
| Left-handed pitchers (vs RHB minus vs LHB) | −12.1 | 30.2% |
| Right-handed pitchers | +9.6 | 49.4% |
Pooled 2023–2026 ledgers. Hitters' rate is batting runs; pitchers' is strikeouts and walks plus contact quality.
Batting is summed from the team's own plate appearances against each hand (exact). Situational outs and baserunning come from each hitter's split times his share of time with the club. Fielding and all pitching do not depend on the opposing pitcher's hand, so they are spread by the team's share of plate appearances against each hand. The two splits add exactly to the season total; actual run differential is not split, because runs allowed have no opposing-pitcher hand.
Each pitcher's strikeouts and walks, contact quality and running game by batter side, times his share of time with the club, plus team fielding spread by the share of batters faced from each side. Offense is excluded from this view. vs LHB plus vs RHB equals the staff-and-defense total.
players total = sum over hitters (batting + situational + baserunning + fielding + catcher) x team share
+ sum over pitchers (strikeouts & walks + contact + running game + fielding) x team share
unassigned = (actual run differential - league mean run differential) - players total
A player's team share is his share of plate appearances (hitters) or batters faced (pitchers) with that club, so traded players' value is divided between their teams; batting is summed directly from each team's own plate appearances.
The unassigned layer is everything the ledger deliberately does not credit to a player: offensive and defensive timing, home runs beyond expected, batted-ball conversion that DRS does not assign, the park's effect on run prevention, runner advancement on hits not explained by runner credits, and small differences in innings played (home teams skip the bottom of the ninth in wins). It is the measure of how much a team's record was shaped by things other than its players' context-neutral value. Its spread across teams and the agreement of players' totals with actual run differential are in the validation table below.
| Season | Hitters | Batting vs FG Batting | Baserunning vs FG BsR | Value vs FG WAR | Pitchers vs FG WAR | Team total vs run diff. | Unassigned SD (runs) | League runs / PA |
|---|---|---|---|---|---|---|---|---|
| 2015 | 268 | 0.991 | 0.78 | 0.90 | 0.69 | 0.89 | 48 | 0.110 |
| 2016 | 268 | 0.987 | 0.83 | 0.92 | 0.64 | 0.92 | 45 | 0.115 |
| 2017 | 288 | 0.985 | 0.84 | 0.91 | 0.69 | 0.93 | 50 | 0.120 |
| 2018 | 278 | 0.987 | 0.81 | 0.92 | 0.75 | 0.96 | 44 | 0.119 |
| 2019 | 273 | 0.986 | 0.78 | 0.91 | 0.73 | 0.96 | 55 | 0.121 |
| 2020 | 267 | 0.986 | 0.67 | 0.88 | 0.71 | 0.93 | 20 | 0.126 |
| 2021 | 262 | 0.988 | 0.83 | 0.91 | 0.76 | 0.94 | 48 | 0.122 |
| 2022 | 277 | 0.990 | 0.83 | 0.92 | 0.70 | 0.98 | 29 | 0.118 |
| 2023 | 293 | 0.989 | 0.79 | 0.91 | 0.63 | 0.98 | 32 | 0.118 |
| 2024 | 286 | 0.989 | 0.82 | 0.91 | 0.67 | 0.95 | 37 | 0.119 |
| 2025 | 277 | 0.991 | 0.81 | 0.92 | 0.65 | 0.96 | 35 | 0.118 |
| 2026 | 267 | 0.986 | 0.76 | 0.88 | 0.70 | 0.92 | 41 | 0.118 |
Correlations across hitters with 300+ PA and pitchers with 250+ batters faced (120 and 100 in the 60-game 2020 season). FG = FanGraphs. Team correlations across the 30 clubs.
The dashboard shows FanGraphs WAR next to the ledger's runs everywhere a player or team total appears: the hitters, pitchers and teams tables, the leader cards, the player and team cards (season by season, with a career total), and the Top 100 rookies. It is FanGraphs' own figure, shown for reference, not an output of this model. For the 2027 forecast it is FanGraphs Depth Charts' projected WAR. It is left out of the views split by hand, because WAR exists only for the full season.
team WAR = sum over hitters and pitchers of FanGraphs WAR x team share
The team share is the same share of plate appearances or batters faced that divides a traded player's runs, so team WAR and the players total are split the same way. (FanGraphs' own team pages split by the games a player actually played for each club, which can differ slightly for traded players.) Summed over the 30 clubs this gives exactly FanGraphs' league total, 1,000 WAR in every full season and 369 in the 60-game 2020, which checks that the shares account for every player:
| Season | League WAR (clubs summed) | Team WAR vs wins | Players total vs wins | Team WAR vs players total |
|---|---|---|---|---|
| 2017 | 1,000 | 0.91 | 0.90 | 0.94 |
| 2018 | 1,000 | 0.92 | 0.92 | 0.96 |
| 2019 | 1,000 | 0.93 | 0.93 | 0.95 |
| 2020 | 369 | 0.80 | 0.88 | 0.89 |
| 2021 | 1,000 | 0.92 | 0.86 | 0.94 |
| 2022 | 1,000 | 0.98 | 0.97 | 0.98 |
| 2023 | 1,000 | 0.90 | 0.90 | 0.97 |
| 2024 | 1,000 | 0.94 | 0.95 | 0.95 |
| 2025 | 1,000 | 0.92 | 0.91 | 0.97 |
| 2026 | 1,000 | 0.88 | 0.85 | 0.93 |
Correlations across the 30 clubs. Team WAR is every player's FanGraphs WAR times his share of time with the club; players total is this ledger's runs with the same shares.
Team WAR and the ledger's players total agree closely (r = 0.89–0.98), and each tracks wins about as well as the other. Neither includes the timing, sequencing and park effects that sit in the ledger's unassigned layer.
Two scales. WAR counts wins above a replacement-level player; the ledger counts runs against an average player. Fitted across 2017–2026 (2020 left out), FanGraphs WAR is very nearly a fixed amount per unit of playing time plus a fixed fraction of the ledger's value:
hitters (300+ PA, 2,501 seasons): WAR = 2.00 x PA/600 + 0.082 x value with position r = 0.94 pitchers (250+ batters faced, 2,513 seasons): WAR = 1.79 x BF/600 + 0.076 x runs prevented r = 0.89
So a full-time average player (zero here) is worth about 2 WAR, and roughly 12 of the ledger's runs make one FanGraphs win for hitters (13 for pitchers). That is more than the ~10 runs per win of the league itself, mainly because the two methods disagree on the details (fielding source, baserunning, contact quality), which flattens the fitted slope. Pitchers agree less because FanGraphs' pitcher WAR is FIP-based and ignores the contact quality this ledger credits.
The depth charts' projected WAR for the 2027 clubs sums to 1,160, more than any played season's 1,000, because depth charts plan full seasons for regulars that injuries then cut (the same reason 2027's projected batting-title qualifiers outnumber actual ones). It tracks this forecast's projected team runs at r = 0.92. The dashboard shows FanGraphs' figure as published, without rescaling.
Every player who appeared in MLB in 2026, plus depth-chart players back from a lost season (21 hitters, 50 pitchers) and depth-chart players up from Triple-A or Double-A (81 and 113; Section 14): 761 hitters and 975 pitchers in all. Pitchers who only batted a few times and position players who only pitched are left out of the other list; two-way players are in both. The rates below describe players with MLB history; anyone with Triple-A or Double-A time in the window also gets the minor-league translation as his starting point (Section 14).
Each component's 2027 rate combines the player's last three seasons, weighted toward the most recent (Marcel's weights: hitters 5/4/3, pitchers 3/2/1), and adds k units of league-average play so that small samples are pulled toward average:
rate(c) = sum_y w_y x runs_y(c) / ( sum_y w_y x units_y + k_c ) units = PA or BF then add the aging adjustment for c (next section)
Because every component is already relative to league average, "league-average play" contributes zero runs and k acts purely as a pull toward average (league centering, Section 15, then sets that added play a little below average). For pitchers' strikeouts and walks and contact the added play sits at the pitcher's pitch quality instead (below). Each k is fitted by back-test rather than assumed: a grid search (0 to 10,000) minimizing the PA-weighted squared error of each component's per-PA rate when 2023–24 predict 2025, for players with 100+ units in the target season. The final k averages that fit with the fit for 2024–25 predicting 2026.
| Component | k fitted on 2025 | Final k | |
|---|---|---|---|
| Hitters | Batting | 2,500 | 3,250 |
| Hitters | Situational outs | 10,000 | 10,000 |
| Hitters | Baserunning | 4,000 | 3,500 |
| Hitters | Fielding | 2,500 | 3,250 |
| Hitters | Catcher defense | 1,000 | 1,500 |
| Pitchers | Strikeouts & walks | 800 | 1,250 |
| Pitchers | Contact quality | 3,000 | 5,000 |
| Pitchers | Running game | 2,000 | 3,000 |
| Pitchers | Pitcher fielding | 4,000 | 5,500 |
A larger k means the component repeats less. Situational outs hit the top of the grid: they are pulled essentially all the way to average. Pitchers' strikeouts and walks and contact are fitted together with pitch quality (below); their final k pools 2023–2026.
For two pitcher components the k units of prior play do not sit at league average but at what the pitcher's pitches predict. Every MLB pitch from 2021 (the first season of the current tracking system) through 2026 is pulled from the Stats API feed with its tracking: 4.3 million pitches, 99% or more of each season tracked (fetch_pitches.py).
Pitch values in the ledger's runs. A count's value is the mean strikeouts-and-walks charge (Section 8) of the plate appearances that reached it. Each pitch is worth the change it makes: the next count's value minus this count's, or the plate appearance's charge minus this count's on the pitch that ends it; the contact charge sits on the ball-in-play pitch. A plate appearance's pitch values add up exactly to its charges minus the 0-0 value, so summed over a season they reproduce the ledger's strikeouts and walks (r = 0.992 or higher every season, pitchers with 250+ batters faced).
The stuff model. A ridge-regularized quadratic regression per pitch group (fastball, cutter, breaking ball, offspeed) × same or opposite hand × two strikes or not, on velocity, induced vertical break, horizontal break, spin, extension and release height and side (break and release side mirrored for left-handers), plus each non-fastball's gap in velocity and break from the pitcher's main fastball, and the count. It explains little of any single pitch (0.5% of strikeouts-and-walks value, 0.1% of contact, out of sample), but summed over a season it is a stable trait. Each season is scored only by models fitted on other seasons; scored in sample, the polynomial learns part of the pitchers' own results (its in-sample fit is nearly double the out-of-sample one).
Pitch quality (PQ) is a pitcher's summed predicted pitch values per batter faced, in pitcher runs, weighted 3/2/1 over the three seasons like the rest of the forecast. It predicts next season's contact better than this season's contact does (r = 0.41 vs 0.32, 661 pitcher pairs with 250+ batters faced in both seasons); for strikeouts and walks the results themselves predict better (0.58 vs 0.42), so pitch quality only adds to them.
rate(c) = ( sum_y w_y x runs_y(c) + k_c x beta_c x PQ(c) x U / (U + kq_c) ) / ( U + k_c ) U = sum_y w_y x BF_y c = strikeouts & walks, contact
beta scales pitch quality into the component's runs and kq shrinks it for a pitcher with few batters faced; both are fitted with k by the same grid search (beta 0 to 2 by 0.1, kq 0 to 3,200, k up to 30,000). With beta = 0 this is the plain forecast. Its value moves from season to season (for 2025, fitted alone, it carried little weight), so the final settings pool the four targets with pitch data, 2023–2026, each predicted from its two seasons before:
| Component | k (fit on 2025) | beta | kq | Final k (2023–26) | beta | kq |
|---|---|---|---|---|---|---|
| Strikeouts & walks | 800 | 0.3 | 200 | 1,250 | 0.7 | 400 |
| Contact quality | 3,000 | 0.4 | 0 | 5,000 | 0.5 | 50 |
In the forecast's own out-of-sample test (settings fitted on 2025, 2024–25 predicting 2026, 250+ batters faced) pitch quality cuts the typical miss from 13.29 to 13.03 runs per 600 batters faced (r 0.453 to 0.481): strikeouts and walks 8.16 to 8.00, contact 8.73 to 8.52; for 250–499 batters faced 15.80 to 15.57, for 500+ 11.13 to 10.83. Holding out each season from 2023 to 2026 in turn (settings fitted on the other three, two-season histories; pitch_quality.py):
| Typical miss per 600 BF | 2023 | 2024 | 2025 | 2026 | Pooled | Better in bootstrap |
|---|---|---|---|---|---|---|
| Without pitch quality | 13.73 | 12.44 | 12.78 | 13.31 | 13.07 | |
| Stuff (used) | 13.50 | 12.45 | 12.90 | 12.97 | 12.96 | 92% |
| Stuff plus location | 13.51 | 12.44 | 12.88 | 13.05 | 12.98 | 88% |
| Stuff, contact only | 13.55 | 12.50 | 12.75 | 13.14 | 12.99 | 92% |
| Strikeouts & walks: without / with | 8.15 / 8.07 | 7.87 / 7.74 | 7.39 / 7.51 | 8.15 / 7.87 | ||
| Contact quality: without / with | 9.13 / 8.84 | 8.02 / 7.97 | 8.22 / 8.03 | 8.78 / 8.56 |
Bootstrap: share of 4,000 resamples of pitchers (paired, within each held-out season) in which the pooled miss is below the plain forecast's. The two component rows compare the plain forecast with the stuff version.
Minor leaguers get the same measure from Triple-A, the only fully pitch-tracked level (from 2023): Triple-A pitches are scored by this MLB model and enter the minor-league translation as two more pieces (Section 14).
Contact improves in every held-out season and strikeouts and walks in three of four; in total the stuff version wins two seasons, ties 2024 and loses 2025. Adding plate location (the mean pitch value of the pitch's location cell for its group, count and platoon) explains far more of single pitches but nothing more of next season, because whatever location captures about command is already in the results. The gain is modest, about 0.1 to 0.3 runs per 600 batters faced, the size of the aging refit's.
The player's platoon gap is regressed toward the typical gap for his handedness, using the harmonic sample of his weighted plate appearances against each hand, and the split is built around his overall rate so the overall projection is unchanged:
gap_r = g(hand) + (observed gap - g(hand)) x n / (n + k_gap) n = 1 / (1/n_L + 1/n_R) k_gap = 3,500 (hitters), 900 (pitchers) share_L = (weighted PA vs LHP + 200 x league share for his hand) / (weighted PA + 200) rate vs L = rate - (1 - share_L) x gap_r rate vs R = rate + share_L x gap_r
With every setting fitted on the 2025 target, 2024–25 histories predict 2026:
| Players | Forecast r | Forecast miss (RMSE per 600) | Last year's rate: r | Last year's rate: miss | |
|---|---|---|---|---|---|
| Hitters (300+ PA) | 248 | 0.402 | 16.89 | 0.212 | 27.65 |
| Pitchers (250+ BF) | 250 | 0.481 | 13.03 | 0.338 | 17.02 |
| Pitchers, without pitch quality | 250 | 0.453 | 13.29 |
The correlations look modest because single-season player value is noisy: for regulars with 300+ PA in both 2025 and 2026, total runs per PA correlate only about 0.39 from one season to the next (batting 0.41, baserunning 0.48, fielding 0.43, situational 0.20; pitchers' strikeouts and walks 0.60, contact 0.33). A forecast cannot correlate with next season much better than next season correlates with itself.
Aging is measured from the same players in consecutive seasons. For every pair of seasons (y, y+1) and every player with at least 200 units (PA or batters faced) in both, the change in each component's rate is recorded against his age on July 1 of the second season, weighted by the harmonic mean of his two seasons' playing time:
delta = rate(y+1) - rate(y) per component, per PA or BF weight = 1 / (1/units(y) + 1/units(y+1)) age = age on July 1 of season y+1
The curves use 9 season pairs, 2015-2016 through 2025-2026. Pairs touching 2020 are skipped: in a 60-game season almost no player reaches 200 units and the season's rates are far noisier.
Same-player changes mix real aging with two biases: players who keep their jobs after a good season tend to fall back toward average (regression to the mean), and players who decline sharply lose their jobs and leave the sample (survivorship). Both shift the average change down at every age. The forecast already pulls every player toward average through k, so applying the raw level as well would count regression twice. Each fitted curve therefore keeps only its shape: the weighted mean change across the sample is subtracted, so young players are adjusted up and older players down, relative to each other. On the original three-pair fit, keeping the level made the 2026 back-test worse (17.6 vs 17.1 runs per 600 for hitters); removing it improved on no aging at all.
Five shapes were fitted on the pairs ending by 2025 (so the 2026 test stays out of sample):
a + b(age - 28) + c(age - 28)^2, weighted least squares);Each component then takes the shape that best predicts that component in the held-out 2026 season (RMSE of the component's rate per 600):
| Component | None | Line 2023–25 | Line 2015–25 | Curve 2015–25 | By age 2015–25 |
|---|---|---|---|---|---|
| Batting | 14.205 | 13.925 | 13.862 | 13.859 | 13.868 |
| Baserunning | 2.520 | 2.532 | 2.495 | 2.500 | 2.502 |
| Fielding | 7.605 | 7.629 | 7.599 | 7.608 | 7.613 |
| Pitchers: strikeouts & walks | 8.145 | 8.043 | 8.003 | 7.996 | 8.018 |
| Pitchers: contact quality | 8.554 | 8.562 | 8.524 | 8.524 | 8.545 |
Batting, pitchers' strikeouts and walks, and contact quality take the curve; baserunning and fielding take the straight line. Choosing per component matters for fielding: one quadratic for every component made fielding U-shaped, with 38- and 40-year-olds "improving". That is a selection effect, not aging: aging players who field poorly move to designated hitter, so their fielding runs per PA rise toward zero. The straight line does not bend back.
Total-value back-test on 2026 for each choice:
| Aging | Hitters RMSE | Hitters r | Pitchers RMSE | Pitchers r |
|---|---|---|---|---|
| none | 17.33 | 0.371 | 13.21 | 0.470 |
| straight line, 2023–25 | 17.08 | 0.387 | 13.12 | 0.477 |
| straight line, 2015–25 | 16.89 | 0.402 | 13.04 | 0.481 |
| quadratic curve, 2015–25 | 16.87 | 0.403 | 13.03 | 0.481 |
| age-by-age average, 2015–25 | 16.88 | 0.403 | 13.04 | 0.482 |
| per-part best | 16.89 | 0.402 | 13.03 | 0.481 |
The gains are real but modest, as expected: aging moves a typical player by a few runs per 600, while single-season noise is about 17 runs per 600 for hitters. The baserunning and fielding gains over no aging are very small and should be read as directional.
Final curves, refit on all pairs through 2025–26. Values are the change per year of age, in runs per 600 PA (hitters) or batters faced (pitchers), at each age:
| Component (shape) | 22 | 24 | 26 | 28 | 30 | 32 | 34 | 36 | 38 | 40 |
|---|---|---|---|---|---|---|---|---|---|---|
| Batting (curve) | +4.2 | +2.9 | +1.6 | +0.4 | −0.7 | −1.8 | −2.8 | −3.7 | −4.6 | −5.4 |
| Baserunning (line) | +0.3 | +0.2 | +0.1 | 0.0 | 0.0 | −0.1 | −0.2 | −0.3 | −0.4 | −0.5 |
| Fielding (line) | +0.5 | +0.3 | +0.2 | +0.1 | −0.1 | −0.2 | −0.3 | −0.5 | −0.6 | −0.7 |
| Pitchers: strikeouts & walks (curve) | +3.1 | +2.1 | +1.2 | +0.3 | −0.4 | −0.9 | −1.4 | −1.8 | −2.1 | −2.2 |
| Pitchers: contact quality (curve) | +0.6 | +0.5 | +0.4 | +0.2 | 0.0 | −0.3 | −0.6 | −0.9 | −1.3 | −1.7 |
Hitting peaks around 28–29 and then declines at an accelerating rate; pitchers' strikeouts and walks follow a similar but gentler arc, and their contact quality declines late. Speed and defense decline steadily from the late 20s.
A player's weighted history sits on average about two years before the forecast season (with 5/4/3 weights). Each component is aged over that gap, evaluated at the midpoint age:
years = 2027 - (weighted mean season of the history, weighted by w x units) mid_age = clamp(age in 2027 - years / 2, 20, 40) rate += delta_c(mid_age) x years
A 23-year-old with three full seasons gains about +2 runs per 600 per year of the gap in batting; a 37-year-old loses about −4. A player with only one season of history (a 2026 rookie) has a gap of about one year.
Through the 2026 season the forecast projected only players who had appeared in MLB; everyone else on a 2027 depth chart was a flat newcomer allowance. This section adds Triple-A and Double-A players. It builds the same ledger in the minors and measures, from the players who made the jump, how much of each piece of minor-league value shows up in MLB the next season. A minor-league line is never taken at face value: it passes through a level gap, a partial carry-over for each piece, a pull toward average for small samples, and an age adjustment.
The Stats API serves minor-league play-by-play in the same shape as MLB (sports 11–14 are Triple-A, Double-A, High-A and Single-A), so the same reconstruction runs unchanged: base-out states, runner movements, batted-ball type and handedness. Across 2021–26 it reconciles to the final score in every game but one, with a handful of base-state resyncs per 150,000 plate appearances. Each plate appearance also keeps its park, because affiliates sometimes play home games elsewhere. Seasons start in 2021, when the minors were reorganized (there was no 2020 season). Exit velocity and launch angle exist only where the park has Hawk-Eye: every Triple-A park from 2023, some in 2022, none in 2021, none at Double-A or High-A, and the Florida State League's parks at Single-A. All four levels are built; the forecast uses Triple-A and Double-A (see "High-A and Single-A" below).
| Season | AAA PA | AA PA | A+ PA | A PA | AAA exit velocity | A exit velocity |
|---|---|---|---|---|---|---|
| 2021 | 145,997 | 129,504 | 133,870 | 135,845 | 0% | 27% |
| 2022 | 171,720 | 156,510 | 146,833 | 149,397 | 38% | 28% |
| 2023 | 174,963 | 156,701 | 147,812 | 148,834 | 100% | 27% |
| 2024 | 172,462 | 152,818 | 147,055 | 148,374 | 99% | 29% |
| 2025 | 171,483 | 152,649 | 147,225 | 149,076 | 99% | 29% |
| 2026 | 171,799 | 156,270 | 151,429 | 151,705 | 99% | 29% |
Exit velocity = share of batted balls with exit velocity and launch angle. Double-A and High-A have none; Single-A has it only in the Florida State League's parks. In 2021 the leagues had interim names (Triple-A East / West, Double-A Northeast / Central / South, High-A East / Central / West, Low-A East / Southeast / West).
Every number is measured against the player's own league in that season (International or Pacific Coast League; Eastern, Southern or Texas League), so the environments drop out: Triple-A's automated strike zone, the bigger bases and pickoff limits, the high-altitude Pacific Coast parks. Events are valued at the same season's MLB linear weights, so a walk is worth the same in every league and only the player's mix of events against his league's mix counts. Batting is split into the pieces that carry over differently. For a plate appearance with result c (MLB value v_c; b_avg = MLB's average batted-ball value):
K = v_K - b_avg strikeouts, valued against a ball in play BB = v_c - b_avg walks, HBP, intentional walks, interference TYPE = x_traj - b_avg batted-ball mix: MLB's mean value for the trajectory QUAL = x_evla - x_traj exit velocity x launch angle cell value beyond it RES = v_c - x_best results beyond the expected value K + BB + TYPE + QUAL + RES = the MLB ledger's batting value
The trajectory and exit-velocity values come from MLB's own batted balls that season, with the home-run probability split used throughout this model (Section 8). One trap: the trajectory cells are keyed on the recorded trajectory, never on the result. The MLB contact model separates bunt hits from bunt outs by result, which is harmless when describing a season but would leak the outcome into a predictor.
Each piece is park-adjusted by venue, additively. The venue's mean per plate appearance minus the mean in its home teams' road games, over two seasons, regressed by n / (n + 10,000 PA); then every league is re-centered on zero. Situational outs and baserunning use exactly the MLB method (Section 6) with the league's own run expectancy. Pitchers carry the same pieces with the sign flipped (strikeouts / walks = PK + PBB; contact = PTYPE + PQUAL; results PRES kept apart, as in MLB) and a running-game term. There is no minor-league fielding: the minors have no DRS.
Triple-A is pitch-tracked by the same Hawk-Eye system as MLB in every park from 2023 (about a third of 2022's pitches; Double-A and High-A have none). Every Triple-A pitch is scored by the MLB stuff model of Section 12, fitted on MLB seasons through that season, so the score reads on the MLB scale: what the pitch would be worth in MLB. Summed per pitcher-season and divided by tracked batters faced, it is centered on that season's Triple-A average (an untracked pitcher counts as average) and enters as one more piece of strikeouts / walks (PQK) and of contact (PQC), shrunk by its own split-half reliability like every other piece. It is the steadiest piece in the minors (odd vs even plate appearances r = 0.98 and 0.95), because a pitcher's velocity and movement barely change from one outing to the next, and it carries over at 0.42 and 0.45 of a run. For contact it is the only minor-league piece that carries over well. In the back-test it lowers the typical miss in every group, pitchers with no MLB time most (better in 98% of 4,000 paired bootstrap resamples):
| Pitchers | n | Without pitch quality: RMSE (r) | With: RMSE (r) |
|---|---|---|---|
| all | 928 | 14.29 (0.37) | 14.23 (0.38) |
| minors only | 105 | 15.81 (0.12) | 15.59 (0.18) |
| mostly minors | 281 | 15.89 (0.29) | 15.82 (0.28) |
| mostly MLB | 542 | 13.39 (0.45) | 13.36 (0.45) |
The translation's own back-test (2025 and 2026 pooled, each fitted on earlier targets), with and without the two pitch-quality pieces; groups as in the back-test table below.
A piece that repeats within a season is a skill; one that does not is mostly noise. Each piece is pulled toward its league average by its own split-half reliability before it is translated:
x~ = (weighted rate) x n / (n + k_j) n = minor-league PA (or BF) in the window k_j = h (1 - r) / r r = odd-vs-even PA correlation, h = typical half size
| Piece | Hitters AAA r (k) | Hitters AA r (k) | Pitchers AAA r (k) | Pitchers AA r (k) |
|---|---|---|---|---|
| strikeouts | 0.73 (64) | 0.76 (53) | 0.59 (110) | 0.60 (104) |
| walks / HBP | 0.56 (139) | 0.56 (138) | 0.54 (133) | 0.56 (123) |
| batted-ball mix | 0.31 (384) | 0.31 (389) | 0.14 (1,000) | 0.17 (747) |
| contact quality | 0.62 (108) | – | 0.09 (1,702) | – |
| results beyond contact | 0.15 (978) | 0.34 (345) | 0.13 (1,100) | 0.09 (1,671) |
| all batting pieces | 0.26 (487) | 0.33 (357) | 0.22 (549) | 0.20 (622) |
| pitch quality: strikeouts / walks | – | – | 0.98 (20) | – |
| pitch quality: contact | – | – | 0.95 (20) | – |
Split-half correlation of a player's odd and even plate appearances (100+ in each half, every season pooled, about 175 PA per half); k in plate appearances or batters faced. Exit-velocity pieces only from seasons with tracking; none at Double-A.
Strikeouts settle in about 60 plate appearances; results on balls in play need about 1,000. A hitter's total batting value over half a Triple-A season repeats at only r = 0.26, which is why the translation works piece by piece. For pitchers, contact quality allowed hardly repeats at all, as in MLB.
The projection from level L is a level gap plus the carried-over pieces plus an age term; levels are mixed by discounted playing time:
milb_L = a_L + sum_j b_Lj x~_Lj + g_L x (age - 26) milb = sum_L s_L milb_L s_L = d_L U_L / sum(d U) U = weighted units
It enters the forecast as the starting point a player's MLB record is weighed against. R and U_mlb are his MLB runs and units weighted 5/4/3 (pitchers 3/2/1), k is the MLB forecast's own regression constant (Section 12), and m is the minor-league share of his history:
rate = (R + K x m x milb) / (U + K) K = k + sum_L d_L U_L m = sum(d U) / (sum(d U) + U_mlb)
With no minor-league time, K = k and m = 0: exactly the MLB forecast, pulled toward average. With no MLB time it is exactly the translation. A veteran's rehab stint has a tiny m and barely moves him. Given d (and k for the single-piece parts), the rate is linear in a, b and g, so they come from weighted least squares, weighted by the player's MLB playing time in the target season; d is chosen on a grid. Batting, strikeouts / walks and contact use one carry-over per piece for both levels, with the Double-A piece scaled by a fitted factor. Separate slopes per level fit no better (17.24 against 17.25 runs per 600) and gave unstable, sometimes negative, Double-A slopes. The fit uses 2,698 hitter and 3,407 pitcher player-seasons (targets 2022–26).
| Piece | Levels | Carry-over b | k | Full Triple-A season | |
|---|---|---|---|---|---|
| Hitters | strikeouts | both (Double-A × 0.70) | 0.81 | 64 | 0.71 |
| Hitters | walks / HBP | both (Double-A × 0.70) | 0.74 | 139 | 0.56 |
| Hitters | batted-ball mix | both (Double-A × 0.70) | 1.45 | 384 | 0.78 |
| Hitters | contact quality | both (Double-A × 0.70) | 0.91 | 108 | 0.73 |
| Hitters | results beyond contact | both (Double-A × 0.70) | 1.14 | 978 | 0.36 |
| Hitters | situational outs | Triple-A | 0.35 | 300 | 0.21 |
| Hitters | situational outs | Double-A | 0.12 | 300 | 0.07 |
| Hitters | baserunning | Triple-A | 0.89 | 300 | 0.54 |
| Hitters | baserunning | Double-A | 0.58 | 300 | 0.35 |
| Pitchers | strikeouts | both (Double-A × 1.00) | 0.54 | 110 | 0.44 |
| Pitchers | walks / HBP | both (Double-A × 1.00) | 0.24 | 133 | 0.18 |
| Pitchers | pitch quality: strikeouts / walks | both (Double-A × 1.00) | 0.42 | 20 | 0.40 |
| Pitchers | batted-ball mix allowed | both (Double-A × 1.00) | 0.62 | 1,000 | 0.19 |
| Pitchers | contact quality allowed | both (Double-A × 1.00) | 0.84 | 1,702 | 0.18 |
| Pitchers | results beyond contact | both (Double-A × 1.00) | 0.04 | 1,100 | 0.01 |
| Pitchers | pitch quality: contact | both (Double-A × 1.00) | 0.45 | 20 | 0.43 |
| Pitchers | running game | Triple-A | 0.61 | 300 | 0.37 |
| Pitchers | running game | Double-A | 0.34 | 300 | 0.20 |
b multiplies the reliability-shrunk rate. The last column is what a run above the league over a full season (450 PA or batters faced) is worth in MLB the next season.
| Part | AAA gap / 600 | AA gap / 600 | AAA age / yr | AA age / yr | d (AAA / AA) | |
|---|---|---|---|---|---|---|
| Hitters | batting | −15.7 | −20.9 | −1.48 | −2.82 | 3.00 / 1.00 |
| Hitters | situational outs | +0.2 | +0.5 | −0.08 | −0.07 | 3.00 / 3.00 |
| Hitters | baserunning | −0.1 | −0.5 | −0.06 | −0.26 | 2.00 / 2.00 |
| Pitchers | strikeouts / walks | −5.0 | −6.5 | −0.58 | −0.17 | 0.50 / 0.50 |
| Pitchers | contact allowed | −1.2 | +0.7 | +0.42 | −0.01 | 3.00 / 0.25 |
| Pitchers | running game | −0.4 | −0.4 | −0.06 | −0.06 | 1.50 / 1.50 |
Gap = what an average player at the level, aged 26, projects to in MLB the next season. Age = change per year older than 26 (negative: the young project better). d = how much a minor-league plate appearance counts against an MLB one when both exist.
For hitters, strikeouts, contact quality and batted-ball mix carry over at roughly 0.7–0.8 of a full-season run, walks at about half and results beyond contact at about a third. For pitchers only strikeouts carry over well; walks carry over weakly and contact allowed hardly at all. Triple-A pitch quality (above) carries over far better than the contact pieces.
| Group | n | Pieces: RMSE (r) | Bottom line: RMSE (r) | Before: RMSE (r) | Bias: pieces / before | |
|---|---|---|---|---|---|---|
| Hitters | all | 767 | 17.25 (0.34) | 17.48 (0.31) | 17.62 (0.28) | −1.6 / +0.3 |
| Hitters | minors only | 99 | 20.32 (0.25) | 21.05 (0.20) | 21.68 (0.11) | −6.1 / −7.8 |
| Hitters | mostly minors | 241 | 17.66 (0.24) | 17.66 (0.22) | 18.44 (0.06) | −2.4 / +2.0 |
| Hitters | mostly MLB | 427 | 16.55 (0.42) | 16.79 (0.40) | 16.54 (0.42) | −0.5 / +0.8 |
| Pitchers | all | 928 | 14.23 (0.38) | 14.63 (0.33) | 14.22 (0.37) | −1.4 / −1.1 |
| Pitchers | minors only | 105 | 15.59 (0.18) | 16.10 (−0.04) | 15.54 (−0.02) | −5.1 / −4.8 |
| Pitchers | mostly minors | 281 | 15.82 (0.28) | 16.12 (0.24) | 15.75 (0.27) | −2.2 / −0.9 |
| Pitchers | mostly MLB | 542 | 13.36 (0.45) | 13.79 (0.42) | 13.39 (0.45) | −0.7 / −0.6 |
2025 and 2026 test seasons pooled; each fitted only on earlier targets (2022–24 for 2025, 2022–25 for 2026). Players with minor-league time in the three seasons before and 100+ MLB PA or batters faced in the test season; weighted by playing time, runs per 600. 'Before' is the forecast without the minors: MLB history pulled to average, and the flat newcomer level for players with no MLB time. 'Minors only': no MLB time in the window; 'mostly minors': more weighted minor-league than MLB time. Bias = projected minus actual.
For hitters the translation beats the forecast without it in every group, and the piece-by-piece version beats translating total value. Players coming straight from the minors gain the most: the typical miss falls from 21.7 to 20.3 runs per 600 and the ranking correlation rises from 0.11 to 0.25. For pitchers the translation ranks players a little better but does not shrink the typical miss, because only strikeouts carry over. Both versions under-project call-ups who reach 100 PA by about 5 runs per 600: players who hit keep playing, so the test group is selected toward success. Two ideas were tried and dropped: minor-league strikeouts as a predictor of MLB contact allowed (14.11 against 14.12), and a single regression constant for every piece (slopes above 1 and a meaningless Double-A exit-velocity slope).
FanGraphs' Future Value and tool grades were tested as additions to the translation. The grades come from The Board, using each season's end-of-season list (2021–26, joined to MLB ids through FanGraphs' minor-league leaderboards); for target season T the T−1 list is used, which is what was known when a forecast for T would have been made. They entered as extra terms in the minor-league projection, weighted like the stats, and were fitted and tested the same way. A second test caps every player's weight at 150 PA or batters faced and counts every call-up with 30+. That guards against a selection trap: teams keep playing high-graded prospects who struggle and drop low-graded players who struggle, so weighting by playing time could hide what the grades know.
| Test | Group | n | Stats only | + FV | + tools | Grades only | Before | |
|---|---|---|---|---|---|---|---|---|
| Hitters | playing-time weights, 100+ | all with minor-league time | 767 | 17.25 (0.34) | 17.27 (0.33) | 17.27 (0.34) | 17.58 (0.28) | 17.62 (0.28) |
| Hitters | playing-time weights, 100+ | no MLB time | 99 | 20.32 (0.25) | 20.68 (0.21) | 20.50 (0.23) | 21.18 (0.04) | 21.68 (0.11) |
| Hitters | playing-time weights, 100+ | mostly minors | 241 | 17.66 (0.24) | 17.52 (0.25) | 17.56 (0.25) | 18.41 (0.07) | 18.44 (0.06) |
| Pitchers | playing-time weights, 100+ | all with minor-league time | 928 | 14.12 (0.38) | 14.12 (0.38) | 14.10 (0.39) | 14.15 (0.38) | 14.09 (0.38) |
| Pitchers | playing-time weights, 100+ | no MLB time | 105 | 15.73 (0.11) | 15.82 (0.10) | 15.70 (0.14) | 15.69 (−0.03) | 15.54 (−0.02) |
| Pitchers | playing-time weights, 100+ | mostly minors | 281 | 15.81 (0.30) | 15.83 (0.29) | 15.80 (0.29) | 15.91 (0.27) | 15.67 (0.26) |
| Hitters | weights capped at 150, 30+ | all with minor-league time | 984 | 21.81 (0.37) | 21.79 (0.37) | 21.79 (0.37) | 22.14 (0.33) | 23.11 (0.26) |
| Hitters | weights capped at 150, 30+ | no MLB time | 163 | 24.89 (0.27) | 24.98 (0.27) | 24.97 (0.26) | 25.33 (0.24) | 25.64 (0.10) |
| Hitters | weights capped at 150, 30+ | mostly minors | 346 | 23.89 (0.30) | 23.82 (0.31) | 23.80 (0.31) | 24.48 (0.22) | 26.40 (0.05) |
| Pitchers | weights capped at 150, 30+ | all with minor-league time | 1287 | 17.90 (0.28) | 17.90 (0.29) | 17.89 (0.29) | 17.91 (0.29) | 17.93 (0.29) |
| Pitchers | weights capped at 150, 30+ | no MLB time | 208 | 20.81 (0.07) | 20.83 (0.06) | 20.75 (0.10) | 20.81 (0.07) | 20.62 (0.05) |
| Pitchers | weights capped at 150, 30+ | mostly minors | 415 | 18.83 (0.17) | 18.82 (0.17) | 18.84 (0.17) | 18.86 (0.20) | 18.94 (0.20) |
Typical miss in runs per 600 (correlation), 2025 and 2026 pooled, each fitted on earlier seasons only. '+ FV' adds Future Value (per 5 points) and an indicator for players not on the list; '+ tools' adds future hit, game power and speed grades (pitchers: best future pitch and command); 'Grades only' drops the minor-league stats and keeps FV, age and level; 'Before' is the forecast without the minors.
Neither version improved on the stats. Adding Future Value or the tool grades moved the typical miss by no more than 0.4 runs per 600 either way, in both tests and every group, and the fitted Future Value coefficient is close to zero once the stats are in. On their own, grades beat the old flat newcomer level for hitters but fall short of the stats. Future Value tracks age-for-level closely (correlation −0.57 to −0.44: the higher graded are the younger) and the same season's minor-league production (0.19 to 0.34), and the translation already uses both. The translation therefore stays stats only. Grades may still carry information this test cannot see. One example is who reaches MLB and gets playing time, which the forecast takes from FanGraphs' depth charts. Another is the low minors, where there are no Triple-A or Double-A stats. milb_grades_test.py reruns the test.
High-A and Single-A were rebuilt the same way (2021–26, about 150,000 plate appearances per level per season, the same reconciliation and the same per-league ledger) and added to the translation as two more levels. Each gets its own gap, age term, playing-time discount and scale of the shared carry-overs, never above the level above it. Single-A's partial exit-velocity coverage is folded into results, so its batted-ball value is not lost.
| Players | n | Four levels | Two levels (production) | Before the minors | |
|---|---|---|---|---|---|
| Hitters | Triple-A or Double-A time | 767 | 17.33 (0.33) | 17.25 (0.34) | 17.62 (0.28) |
| Hitters | Triple-A or Double-A time, minors only | 99 | 20.41 (0.23) | 20.32 (0.25) | 21.68 (0.11) |
| Hitters | Triple-A or Double-A time, mostly minors | 251 | 17.47 (0.23) | 17.38 (0.25) | 18.15 (0.08) |
| Hitters | High-A / Single-A time only | 10 | 17.65 (0.31) | 18.20 (0.34) | 18.20 (0.34) |
| Hitters | any minor-league time | 777 | 17.33 (0.33) | 17.27 (0.33) | 17.63 (0.28) |
| Pitchers | Triple-A or Double-A time | 928 | 14.14 (0.38) | 14.12 (0.38) | 14.09 (0.38) |
| Pitchers | Triple-A or Double-A time, minors only | 105 | 15.81 (0.05) | 15.73 (0.11) | 15.54 (−0.02) |
| Pitchers | Triple-A or Double-A time, mostly minors | 291 | 15.66 (0.28) | 15.60 (0.30) | 15.48 (0.26) |
| Pitchers | High-A / Single-A time only | 17 | 16.46 (0.31) | 15.69 (0.57) | 15.69 (0.57) |
| Pitchers | any minor-league time | 945 | 14.19 (0.37) | 14.15 (0.39) | 14.12 (0.38) |
Typical miss in runs per 600 (correlation), 2025 and 2026 pooled, each fitted on earlier seasons only, players with 100+ MLB PA or batters faced. 'Two levels' gives players with only High-A / Single-A time today's treatment (MLB history pulled to average, or the newcomer level).
They do not help next-season projections. Nearly everyone who reaches MLB has Triple-A or Double-A time in the three seasons before, so the lower levels add only older, noisier information about the same players, and more parameters. Only a handful of test players had nothing above High-A. The four−level fit put the batting gap at −16.1 (Triple−A), −19.8 (Double−A), −21.9 (High−A) and −41.4 (Single−A) runs per 600, with an unstable Single−A age slope: the few players who go straight from A−ball to the majors are exceptional. The forecast therefore uses Triple-A and Double-A. The lower levels stay on disk for projections further ahead, where they would carry the information; milb_levels_test.py reruns the comparison.
The forecast now includes 81 hitters and 113 pitchers coming up from Triple-A or Double-A and 21 hitters and 50 pitchers back from a lost season, all with 2027 depth-chart roles. It also uses the minors as the starting point for 605 hitters and 807 pitchers already in the pool who had minor-league time in 2024–26. That matters most for fringe major leaguers with long Triple-A records, whom the forecast without the minors pulled toward MLB average. Translated players get the positional adjustment of their listed position, the typical platoon gap for their handedness, and no fielding value.
| Player | Age in 2027 | Club | Latest season | PA | Runs / 600 vs league | 2027 MLB / 600 | Projected PA |
|---|---|---|---|---|---|---|---|
| Aidan Miller | 23 | PHI | 2026 AA | 11 | −18.7 | −1.2 | 350 |
| Sebastian Walcott | 21 | TEX | 2026 AA | 171 | +49.3 | +3.3 | 322 |
| Owen Ayers | 26 | CHC | 2026 AA | 278 | +46.7 | −13.5 | 321 |
| Seaver King | 24 | WSH | 2026 AAA | 356 | +23.5 | −3.4 | 266 |
| Ralphy Velazquez | 22 | CLE | 2026 AAA | 369 | +23.2 | +6.1 | 245 |
| Murf Gray | 23 | PIT | 2026 AA | 195 | +18.7 | −7.4 | 196 |
| Nelson Rada | 21 | LAA | 2026 AAA | 585 | −8.2 | −11.2 | 182 |
| Noah Cardenas | 27 | MIN | 2026 AAA | 275 | −11.5 | −28.4 | 179 |
Players up from the minors with the most 2027 depth-chart playing time (offense only: batting, situational outs, baserunning). The projection keeps only part of a minor-league edge, after the level gap, with credit for youth.
Projected plate appearances, batters faced and games started come from FanGraphs Depth Charts, whose rosters and playing time follow FanGraphs' RosterResource depth charts. Only playing time and team are taken from it; every rate is this model's own. On the pull of 2026-11-14: 669 hitters and 828 pitchers in the pool had a 2027 role (89 and 125 of them unsigned free agents, who keep FanGraphs' projected playing time and count for no team until they sign); 92 hitters and 147 pitchers were on no depth chart and get zero playing time, keeping their per-600 rates. A pitcher is a starter if he is projected for 8+ starts or starts in at least half his games.
The feed's team assignments were checked against all 30 RosterResource pages: 1,276 players on the same club in both, none on a different club, no feed free agent on a RosterResource roster, and 9 feed players on no RosterResource page. If the depth-chart files are missing, playing time falls back to Marcel's rule (hitters 0.5 × last season + 0.1 × the season before + 200 PA; pitchers the same plus 260 batters faced for starters or 110 for relievers).
Depth charts also give 2027 playing time to players who did not appear in MLB in 2026. Since the minor-league stage they are projected individually, like everyone else:
What is left is depth-chart time for players with no MLB, Triple-A or Double-A record since 2024 (overseas signings, jumps from A-ball): 70 PA and 191 batters faced across the league. It counts at the measured newcomer level: what players who were not in MLB the season before actually produced, pooled over 2024–26:
newcomer level = −6.5 runs per 600 PA (hitters), −2.5 per 600 BF (pitchers)
The conventional alternative, replacement level, is far harsher. It follows the convention that a replacement-level team wins about 48 games (1,000 WAR a season, 57% to position players, 43% to pitchers), converted to runs at 2026's runs per win:
replacement = -0.57 x 1000 x RPW / league PA = −17.8 per 600 PA
-0.43 x 1000 x RPW / league PA = −13.4 per 600 BF
Measured newcomers play well above replacement because the group includes regulars back from a lost season, not just call-ups. Before the minor-league stage this level filled about 8,400 PA and 18,900 batters faced; now it fills only the small remainder. The dashboard's team popups show both values.
Runs against average add up to zero in every real season. The projections did not: on the 2026-11-14 pull the 30 depth charts summed to +1,419 runs (+47.3 a club; hitters +688, pitchers +731). The back-test shows the same lean at actual playing time, and shows where it sits: regulars were projected about right, part-time players well above what they produced.
The fix moves the pull toward average. Each projection adds k units of league-average play to the player's history (Section 12). That added play now counts at a level m a little below average, one per component, solved so the depth-chart league, newcomers included, sums to zero:
rate(c) = ( sum_y w_y x runs_y(c) + k_c x m_c ) / ( sum_y w_y x units_y + k_c )
= uncentered rate + m_c x k_c / (U + k_c)
m_c solved so that the 30 clubs' projected runs sum to zero
m per 600, hitters: batting −3.5, situational outs −0.3, baserunning −0.7, fielding −1.0, catcher defense −3.6
m per 600, pitchers: strikeouts and walks −3.6, contact −1.1, running game +0.1, pitcher fielding 0.0
A regular with years of history leans little on the added play and moves 1 to 2 runs per 600; part-timers and rookies move 2 to 4. For translated minor leaguers only the league-average part of the prior is centered, since the translated part was fitted to MLB outcomes. Fielding is not shifted for designated hitters, and catcher defense only for catchers. Newcomers stay at the measured level. Because the shift follows each roster's mix, it is not the same for every club: from −53.4 runs (SD) to −42.8 (NYY), a spread of 1.1 wins. Teams leaning on part-timers and rookies give back the most.
| Target | Uncentered league | All: none | All: uniform | All: prior-weighted | Regulars: none | Regulars: uniform | Regulars: prior-weighted | |
|---|---|---|---|---|---|---|---|---|
| Hitters | 2026 | +2.62 | 21.90 | 21.50 | 21.37 | 16.90 | 17.09 | 17.06 |
| Hitters | 2025 | +3.18 | 21.61 | 21.33 | 21.15 | 16.38 | 16.77 | 16.70 |
| Pitchers | 2026 | +1.91 | 16.20 | 16.09 | 16.12 | 13.13 | 13.20 | 13.34 |
| Pitchers | 2025 | +1.78 | 16.27 | 16.17 | 16.10 | 12.71 | 12.88 | 12.84 |
Runs per 600 PA (hitters) or batters faced (pitchers). Each projection is put at the player's actual playing time in the target season; 'uncentered league' is how far the projected league, newcomers at the measured level, came out above zero. Typical miss is weighted by playing time. All = everyone projected who played; regulars = 300+ PA or 250+ batters faced. 2026 is out of sample; 2025 uses the same settings, fitted on it. Uniform = one shift per unit for every player.
Over everyone who played, centering lowers the typical miss in both seasons, for hitters and pitchers. A single uniform shift does nearly as well, a little better for pitchers in 2026 and worse for hitters, so the prior-weighted form is used. For regulars alone it adds 0.1 to 0.3 runs per 600, because they were already projected about right. That group is chosen by what happened (players who play well keep their jobs), which flatters an uncentered projection, so the cost is probably overstated.
team runs = sum of projected (centered) runs of every player on the club's depth chart
(2026 major leaguers, players back from a lost season, players up from AAA / AA)
+ leftover PA x newcomer hitter level + leftover BF x newcomer pitcher level
wins = 81 + (team runs - league mean team runs) / RPW (the mean is zero after centering)
RPW = 4R / (2R)^0.287 = 9.54 R = 2026 runs per team-game (Pythagenpat)
Projected wins sum to exactly 2,430. On the 2026-11-14 pull the top clubs were LAD 95.8, MIL 93.7, BOS 92.0, NYY 90.5, SEA 85.8; the bottom were KC 72.9, CIN 72.4, COL 66.1. Team splits in the forecast follow the same rules as for actual seasons. Translated players get the typical platoon gap for their handedness; leftover newcomer hitters are assigned the league share of plate appearances against left-handers and leftover newcomer pitchers the league share of left-handed batters.
The 2027 forecast season has a Top 100 rookies view: the rookie-eligible players with a depth-chart role, ranked by projected runs above replacement. That is runs against average (hitters with the position adjustment) plus the playing time valued at the replacement gap above, so a regular outranks a cameo; ranked against average alone, a rookie with ten innings would beat an everyday rookie a run below average. 378 of the 423 rookie-eligible players in the forecast have a 2027 role.
Eligibility follows MLB's rule: a player is a rookie unless, in earlier seasons, he exceeded 130 at-bats or 50 innings in the majors, or spent more than 45 days on an active roster during the period of the 26-player limit (Opening Day to August 31), injured-list time excluded. At-bats and innings come from the Stats API's season-by-season MLB stats. Active days are rebuilt from each player's transaction log (on at a call-up, selected contract, waiver claim or activation; off at an option, injured-list placement, designation, outright or release). The people endpoint's roster entries cannot be used: they record organization stints, not active-roster days. Checked against MLB's own 2026 rookie list, the rebuilt rule agreed for 1,435 of 1,470 players who appeared that season; nearly every miss undercounted a fringe player's active days. So MLB's lists settle each player's history through his last MLB season (since 2024), and the counts decide only the rest. Players with 35–45 counted days are flagged as close calls.
| # | Rookie | Club | Pos | Age | Proj. PA / BF | Runs vs avg | Runs above repl. |
|---|---|---|---|---|---|---|---|
| 1 | Kaelen Culpepper | MIN | SS | 24 | 602 | +4.6 | +22.5 |
| 2 | Walker Jenkins | MIN | CF | 22 | 490 | +4.8 | +19.3 |
| 3 | Kade Anderson | SEA | SP | 22 | 558 | +5.0 | +17.5 |
| 4 | Leo Bernal | STL | C | 23 | 434 | +1.9 | +14.8 |
| 5 | River Ryan | DET | SP | 28 | 530 | +2.2 | +14.0 |
| 6 | Hagen Smith | CWS | SP | 23 | 437 | +3.2 | +12.9 |
| 7 | Ethan Salas | SD | C | 21 | 459 | −1.3 | +12.3 |
| 8 | Aidan Miller | PHI | SS | 23 | 350 | +1.8 | +12.2 |
| 9 | Jonah Tong | NYM | SP | 24 | 349 | +2.2 | +9.9 |
| 10 | Joshua Báez | STL | LF | 24 | 518 | −5.4 | +9.9 |
The top 10 of the dashboard's Top 100 rookies view (2027 forecast season).
A scheduled task re-pulls the depth charts every Monday morning, reruns the forecast, re-checks rookie eligibility, rebuilds and republishes the dashboard, and writes what changed (club changes and signings, players on or off a depth chart, team projected-win swings of half a win or more) to a log. Rates do not change week to week; playing time, teams and team totals do.
Almost every ingredient of the ledger comes from earlier public work. What is particular to it is how the pieces are combined: one run currency for both sides of every plate appearance, books that balance against actual run differential, a tested rule for which situational play counts as skill, and a forecast whose settings are all chosen by back-test.
| Piece | Borrowed from | What the ledger changes |
|---|---|---|
| Run expectancy and event run values | Lindsey's run expectancy tables; Palmer's linear weights; Tango, Lichtman and Dolphin's The Book; FanGraphs wOBA and wRAA | Outs are valued by batted-ball type and reaching on an error is its own event (wOBA counts it as an out). Batting still agrees with FanGraphs at r ≥ 0.98 every full season (Sections 4, 6). |
| Baserunning | Lichtman's UBR; FanGraphs BsR | What an average runner does is conditioned on the exit-velocity band of the ball, not just the event and state; rated per time on base and applied at the league-average number of chances (Section 6). |
| Fielding | Sports Info Solutions' DRS (non-catchers); Statcast framing, blocking and throwing (catchers) | Used as published. DRS was chosen because it tracks team batted-ball outcomes best (Section 7). |
| Positional adjustment | FanGraphs | Shown for comparing players, kept out of team sums (Section 7). |
| Pitcher contact quality | Statcast expected statistics (xwOBA, xERA) | Expected value of each batted ball from exit velocity and launch angle, in the hitters' run currency. It replaces FIP's actual home runs (Section 8). |
| Park factors | The usual home/road comparison | Two seasons, regressed halfway to neutral (Section 6). |
| Platoon splits | The Book's regressed platoon gap | The overall rate is preserved, so the two splits add to the total (Sections 9, 12). |
| Forecast weights | Tango's Marcel (5/4/3 hitters, 3/2/1 pitchers) | The pull toward average is fitted per component by back-test instead of Marcel's fixed 1,200 PA (Section 12). |
| Pitch quality | Stuff+ and similar pitch-tracking models | Each pitch is valued in the ledger's own runs and scored only by models fitted on other seasons; it enters only the pitchers' forecast, as the prior, with a weight fitted by back-test (Section 12). |
| Aging | The delta method | Slope only: the level shift in same-player deltas double-counts regression to the mean (Section 13). |
| Minor-league translation | Bill James's major-league equivalencies | Piece by piece in MLB runs, each piece shrunk by its own reliability, entered as the prior in the forecast's own formula; Triple-A pitchers also carry their MLB-scale pitch quality (Section 14). |
| Playing time and teams | FanGraphs Depth Charts (RosterResource) | Playing time and club only; every rate is the ledger's (Section 15). |
| Runs per win | Pythagenpat | Turns projected team runs into wins (Section 15). |
| Splitting each plate appearance among batter, runners, pitcher and fielders | openWAR (Baumer, Jensen and Matthews, 2015) | The closest relative; the differences are below. |
"Different" here means we know of no public system that combines these choices this way; each piece may have precedents beyond those named.
run_value_model/scripts)| Script | Does | Writes |
|---|---|---|
fetch_pbp.py YEAR | Pulls a season of play-by-play (16 threads, slimmed fields) | data/pbp_YEAR.jsonl.gz |
pa_table.py YEAR | Rebuilds every plate appearance, runner and runner event; runs the three reconciliation checks | data/pa_, advance_, runner_events_YEAR.csv.gz |
gap_decomp.py YEAR | Run expectancy, LW0/LW1/LW2, offense layers by team | output/gap_decomp_YEAR.json |
prevention_decomp.py YEARS | Run-prevention layers by team; expected value of contact | output/prevention_YEAR.json |
fetch_fg_players.py YEARS | FanGraphs player batting, fielding and pitching rows | data/fg_players_*_YEAR.json |
ledger.py YEAR | The player and team ledger (needs YEAR and YEAR−1 play-by-play) | output/ledger_YEAR.json |
fetch_pbp.py YEAR aaa, pa_table.py YEAR aaa | The same for Triple-A (aa for Double-A), keeping the park id | data/milb/ |
milb_ledger.py all | Minor-league ledger in MLB runs against each league, with split-half sums | output/milb/ledger_LEVEL_YEAR.json |
milb_translate.py | Reliability, translation fit and back-test, 2027 minor-league projections | output/milb_translation.json, output/milb_projections_2027.json |
milb_grades_test.py, milb_levels_test.py | Tests of prospect grades and of High-A / Single-A as additions to the translation | output/milb_grades_test.json, output/milb_levels_test.json |
fetch_pitches.py YEARS | Every MLB pitch with its tracking (velocity, movement, spin, extension, release, location) | data/pitches/pitches_YEAR.csv.gz |
pitch_quality.py scores | Pitch values, the stuff model and each pitcher's pitch quality, scored leave-season-out (once a season) | output/pitch_quality_scores.json |
pitch_quality.py | The pitch-quality back-test: stuff vs stuff plus location, two protocols, bootstrap | output/pitch_quality_test.json |
fetch_pitches.py YEARS aaa | Every Triple-A pitch with its tracking (full from 2023); scored by pitch_quality.py scores | data/pitches/milb/pitches_aaa_YEAR.csv.gz |
milb_pq_test.py | The translation with and without Triple-A pitch quality | output/milb_pq_test.json |
fetch_depth_charts.py 2027 | FanGraphs Depth Charts playing time and teams | data/fg_depthcharts_*_2027.json |
forecast.py 2027 | Fits k, splits and aging by back-test; projects players and teams | output/forecast_2027.json |
build_dashboard.py | Dashboard for 2017–2026 and the 2027 forecast, with FanGraphs WAR beside the runs and team WAR from player shares | MLB_Run_Ledger.html, output/run_ledger_artifact.html |
build_white_paper.py | This paper, with its tables computed from the outputs | MLB_Run_Ledger_White_Paper.html |
refresh_forecast.py | Weekly: depth charts, forecast, dashboard, change log | output/refresh_log.md |
| Constant | Value | Where it comes from |
|---|---|---|
| Run expectancy table | per season, innings 1–8 | Section 4 |
| Park factor regression | halfway to 1.00, two seasons | Section 6 |
| Advancement cell minimum | 30 runners (else drop the exit-velocity band) | Section 6 |
| Contact cell minimum | 25 balls; 3 mph × 4°, then 6 × 8° | Section 8 |
| Forecast weights | hitters 5/4/3, pitchers 3/2/1 | Marcel |
| Regression k | fitted per component (Section 12) | Back-test |
| Split share prior | 200 PA at the league share for the hand | Section 12 |
| Pitch quality | beta 0.7 / 0.5, kq 400 / 50 (strikeouts & walks / contact) | Back-test, Section 12 |
| Aging sample | 200+ units in both seasons; 2020 skipped | Section 13 |
| Minor-league venue effects | two seasons, regressed by n / (n + 10,000 PA) | Section 14 |
| Minor-league reliability | split-half k = h(1 − r) / r, per piece and level | Section 14 |
| Runs per win | 9.54 | Pythagenpat, Section 15 |
| Newcomer level | −6.5 / 600 PA, −2.5 / 600 BF | Measured, Section 15 |
fFRP is framing and cFRP is first-base receiving.team=0%2Cts; an unencoded comma returns players.