They Got Next: Analyzing the WNBA

Posted on June 15, 2008 by


Perhaps today, and certainly by Thursday, the 2007-08 NBA season will come to a close.  And when that happens, what will basketball fans have to talk about? 

I think there is an obvious answer. The WNBA is currently playing its 12th season.  This means we have 11 complete seasons of statistics just waiting to be analyzed (stats you can find at and  With stats in hand (and I thank my student Ryan Knapp for helping me get this data) there are a number of stories we can tell.

Before we get to the stories, we need to first analyze the data.  And that process should begin with reviewing why we have the data in the first place.

The purpose of tabulating statistics in team sports is to separate the player from the team.  As noted before, we know who won and lost.  What we want to know is which players are responsible for the outcome observed.

Linking Wins to the Efficiency Stats in the WNBA

In basketball the process of assigning responsibility begins with linking wins to the player statistics.  And as detailed in The Wages of Wins (both in the book and at the website), wins are linked to player statistics in basketball by regressing wins on offensive and defensive efficiency. 

Before we get to the results of that regression, let’s quickly look at wins and efficiency differential in the WNBA in 2007. 

Table One: Efficiency Differential in the WNBA in 2007

As Table One reveals, the leader in efficiency differential in the regular season was the Detroit Shock.  Because Wins Produced is derived from the link between wins and the aforementioned efficiency metrics, we would expect to be able to connect what we see at the team level for the Shock to the individual players. To do this, though, we first have to -as noted above – regress wins on the two efficiency measures.

In the NBA, this regression can explain 94% of team wins.  The WNBA began in 1997 with only a 28 game regular season.  Today the teams only play 34 regular season contests.  With a shorter schedule, blowouts will have a bigger impact on the aggregate statistics (in other words, these don’t even out completely over the season).  Consequently, the efficiency metrics only explain 79% of wins in the WNBA.

Although the explanatory power is lower, the basic results are quite similar to what we see in the NBA. 

Tables Two and Three: The Value of Player Statistics in the NBA and WNBA

Tables One and Two reveal that the value of points and possessions are quite similar in both leagues. This is because in both leagues, teams get a similar number of points from each possession.  In the NBA – from 1990-91 to 2007-08 – each NBA team scored (on average) 1.02 points per possession.  In the WNBA – from 1997 to 2007 – each team averaged 0.94 points per possession.  Such a small difference in points per possession results in very similar marginal values for each statistic.  Furthermore – and this is important to note — the simple measure Win Score will be exactly the same for both leagues.

Assigning Responsibility in Detroit

Having gone to all the trouble of determining the values listed in Tables Two and Three, though, we are going to focus on Wins Produced.  For example, consider Table Four.

Table Four: The Detroit Shock in 2007

Table Four reports the Wins Produced of the players employed by the Detroit Shock in 2007.  The Shock won 24 games in 2007.  The summation of Wins Produced comes to 21.8, for a difference of 2.2.  Only Minnesota (3.3), Seattle (2.5), and Phoenix (2.3) had a larger discrepancy (the average difference – in absolute terms — was 1.5).

Again, our objective is to assign responsibility.  When we look at the players listed in Table Four, we see that the Deanna Nolan – who was named to the All-WNBA first team in 2007 – lead the Shock in Wins Produced.  Nolan, though, was not the only productive player on the Shock.

Just like in the NBA, average WP40 (note that the WNBA only plays 40 minutes) is 0.100.  An average team will produced 0.500 wins per game (or win half their games).  This means an average player will produce 0.100 wins per 40 minutes.  

With this benchmark in mind, we see that of the ten players who played at last 100 minutes for the Shock, five were above average.  And in terms of WP40, Cheryl Ford – a 2006 All-WNBA player — led the way (meanwhile, Swin Cash – who had problems in 2007 with Bill Laimbeer (the Shock’s head coach) – was actually below average).

Despite the talent the Shock had assembled, Detroit lost in the WNBA Finals to the Phoenix Mercury.  The WNBA Finals go five games, and although Detroit took Phoenix to the limit, the Mercury prevailed in the deciding game.

This year the Shock are again among the top teams in the league while the Mercury – with a record of 2-6 – are struggling.  And of course we wonder why the Shock are still on top while the Mercury have declined.  The answer – and this will not surprise – is in the stats.  But the story will have to wait for another day.

– DJ

The WoW Journal Comments Policy

Our research on the NBA was summarized HERE.

The Technical Notes at provides substantially more information on the published research behind Wins Produced and Win Score

Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:

Simple Models of Player Performance

Wins Produced vs. Win Score

What Wins Produced Says and What It Does Not Say

Introducing PAWSmin — and a Defense of Box Score Statistics

Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.