Skip to content

fivethirtyeight/WNBA-stats

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 

Repository files navigation

WNBA-stats

This repo contains player advanced stats and Elo ratings for WNBA history.

Player Stats

The file wnba-player-stats.csv contains season-level advanced stats for WNBA players by team for the 1997-2019 seasons, from Basketball-Reference.com. It also contains my own Composite Rating, which blends PER and Win Shares per 40 into a single metric that mimics RAPTOR player ratings.

Category Description
player_ID BB-Ref player ID
Player Player name
year_ID Season
Age Age (as of Jul. 1)
Tm Team played for
tm_gms Team's scheduled games
Tm_Net_Rtg Team's net efficiency (offensive rating minus defensive rating)
Pos Player's position played
G Games played
MP Minutes played
MP_pct Percentage of available minutes played
PER Player Efficiency Rating
TS_pct True Shooting Percentage
ThrPAr Three-point Attempt Rate (3PA/FGA)
FTr Free Throw Rate (FTA/FGA)
ORB_pct Offensive rebound percentage
TRB_pct Total rebound percentage
AST_pct Assist percentage
STL_pct Steal percentage
BLK_pct Block percentage
TOV_pct Turnover percentage
USG_pct Usage percentage
OWS Offensive Win Shares
DWS Defensive Win Shares
WS Total Win Shares
WS40 Win Shares per 40 minutes
Composite_Rating Estimated net points added per 100 possessions
Wins_Generated Wins implied by Composite Rating

Composite Rating is determined by the following formula (based on NBA player stats):

Rating = -5.237248 + 0.1741241*PER + 26.0059929*WS40

Individual ratings are then adjusted so the team's weighted average Composite Rating (times 4.064, a scalar to account for score effects) equals the team's Net Rating. Wins Generated are derived by divvying up the team's Net Rating-implied wins according to each player's contribution to the team's Net Rating.

Elo Ratings

The file wnba-team-elo-ratings.csv contains Elo Ratings for every team in WNBA history on a game-by-game basis. The ratings were developed by FiveThirtyEight's Jay Boice, similar to the basic ratings for the NBA. The ratings change after every game based on the winner's pregame win probability, with more unexpected wins resulting in more points shifting from the loser's rating to the winner's.

Category Decription
season Year of game
date Date of game
team1 First team listed's ID
team2 Second team listed's ID
name1 Team1's full name
name2 Team2's full name
neutral Was game at a neutral site? (1=yes)
playoff Was game in playoffs? (1=yes)
score1 Team1's points in game
score2 Team2's points in game
elo1_pre Team1's pregame Elo rating
elo2_pre Team2's pregame Elo rating
elo1_post Team1's postgame Elo rating
elo2_post Team2's postgame Elo rating
prob1 Team1's pregame odds of winning
is_home1 Was Team1 the home team? (1=yes)

Some other pertinent information about WNBA Elo ratings:

  • Home court advantage is 80 (NBA=100)
  • K-factor is 32 (NBA=20)
  • Teams are reverted by 1/2 between seasons (NBA=1/4)
  • In playoffs, elo difference is multiplied by 1.25
  • Margin of victory -- relative to expectation -- matters, same as NBA
  • Expansion teams start at 1300

There were five teams that moved to different cities over the years; in those cases, ratings were carried over from the previous team:

  • 2003: ORL -> CON
  • 2003: UTA -> SAS
  • 2010: DET -> TUL
  • 2016: TUL -> DAL
  • 2018: SAS -> LVA