Win probability, WPA and leverage

The chance a team wins: before a game, during it, and what each play was worth.

What it measures

  • Pregame win probability: the chance each team wins, from the two ratings and where the game is played.
  • In-game win probability: the home team's chance to win after each play.
  • Win probability added (WPA): how much a player's plays moved his team's chance to win, added up.
  • Leverage: how much is riding on a plate appearance, compared with an average one.

How it is built

  • The pregame number comes from the gap between the two ratings, plus home field.
  • The in-game number works from the inning, the outs, the runners on base and the score, and leans toward the pregame favorite: a strong team trailing by one in the third is in better shape than a weak one.
  • WPA credits the change in win probability on each play to the hitter and charges it to the pitcher.

How to read it

  • Probabilities run from 0 to 1. One exception: matchup returns percentages (56.6, not 0.566).
  • A WPA of +1.0 over a season is one win added by the player's plays alone.
  • A leverage of 1.0 is an average situation; 2.0 means twice as much is at stake.
  • The PEAR field of a game is the projected margin in runs, written with the favorite: LSU -0.83.

In the API

  • games and schedule: home_win_prob and the PEAR line for every game.
  • matchup: the same for any two teams.
  • play-by-play/{game_id}: home_wp after every play, and win_prob_pregame.
  • players: WPA and pLI, the player's average leverage.

Limits

  • WPA depends on when a player's plays happened, so it says more about what happened than about skill. For skill, use PRV.
  • The in-game number knows the ratings but not who is pitching or who is left in the bullpen.
Advertisement