2025 MLB

Where the Wins Came From

979693929090898989898886848483838280807777757573726968686545

Every 2025 season begins as a single number, wins above replacement, and ends as a clubhouse full of players. This is the trip from one to the other, in three charts.

The thirty teams, by record. Click one to follow it through every chart.

Primer

What goes into a win above replacement

Before the next chart turns each team into a single number, here is where that number comes from. WAR, wins above replacement, is built one player at a time, by adding up the runs a player creates or saves and dividing by runs-per-win. Hover any bar or distribution to read it.

Figure 1.

Cumulative team WAR tracks with wins, but WAR distributions vary widely across teams.

  1. AFor each team, WAR is calculated by adding up player WAR and plotting it against wins, revealing the expected positive correlation between player production (WAR) and team record (wins).
  2. BAn interactive table that allows sorting of wins, the wins-vs-WAR differential, and the Gini coefficient. These show the team standings, a measure of over- or underperformance relative to player production, and how concentrated production is across the roster (0 means perfectly equal production, 1 means production concentrated among fewer players).
  3. CTeam-level WAR distributions show how value is spread across batters/fielders and pitchers.
  4. DEach team's record relative to .500 is decomposed into cumulative player Win Probability Added, showing how individual players collectively push a team above or below breaking even.

A team's WAR is only the sum of its players'. So before asking how a club assembled its wins, look at the raw material: every player in the league, sized by what he was worth.

Figure 2.

League-wide player distribution and per-player WAR component breakdown.

  1. AAll players are plotted against the full distribution of WAR. Users can filter by team, by batting and pitching roles, and by position to see how class is distributed across the league. Ohtani, a league-outlier two-way player, is shown in his own tranche with individual batter and pitcher profiles.
  2. BSelecting a player reveals the individual components that build his WAR profile. Shows Runs above Replacement (RAR) values per component (offense, defense, baserunning, pitching, playing-time) and percentile rankings.

The crowd has a shape, but it is still everyone at once. The last chapter takes these same players and sorts them back onto the teams they actually won games for.

Primer

How cumulative win probability is built

The last chapter measures players a second way: not the talent they had, but the win probability they actually swung, play by play. Here is how that running total is built before you read it on a team.

Figure 3.

Team record as a player-by-player tug-of-war of win probability.

  1. AFor a selected team, each player's per-game WPA is shown as a distribution of positive and negative contributions.
  2. BPlayer Net Win Probability Added vs Involvement. Players further along the x-axis are involved in more game-swinging moments. The y-axis measures conversion of that involvement, did the player add win probability in the situations he was put in?
  3. CPlayer Net Win Probability Added vs Wins above Replacement. Given WAR is a benchmark for value, this plot asks whether that value was delivered in win-probability-adding moments. In other words, was value delivered in the moments that mattered?

Three views of one season, the talent a team had, the players that talent was made of, and the way each club turned it into wins.