It's been a long and busy winter for me, one which has taken me away from women's basketball and towards the dubious goal of "career advancement". (Oh, the things you'll do for money.) With a temporary respite in chasing the mighty dollar, I'm hoping to get back into the swing of things, despite the fact that a good six months or so has been eaten from my basketball knowledge. (*)
The focus of this article is an article that Bill James wrote several years ago comparing pitchers Mitt Pappas to Don Drysdale. The former is not in the Baseball Hall of the Fame but the latter is despite identical win-loss records. James's contention is that the greater amount of variance in Drysdale's won-loss record makes him the more valuable player. By variance, I mean that there is a greater "spread" among Drysdale's seasons - he has several good seasons and several crappy ones, whereas Pappas is Mr. Consistency. However, this variance means that it's very likely for Drysdale to have a breakout season which pushes him over the top, whereas Pappas is less likely to have such seasons.
Variance in WNBA career performance, however, is an article for another time. I began to think about variance in game performance, particularly the kind of variance where one player has - or appears to have - a breakout game. Generally, most WNBA teams have a "go to" player who scores the majority of their points. You can probably name your team's go-to player. Sometimes, that player - or some other player - will score a whole lot of points.
Furthermore, I began to think about a certain kind of variance, the kind of variance represented in cases where a player scored the bulk of a team's points during the game - the kind of games which are basically one-player games, where someone explodes against another team. Clearly, if your best player scores 30 or 40 points against some other team, your team is more likely to win the game, right? And clearly, these are the kinds of performances a fan wants to see.
But do those performances actually lead to victories? As it turns out, the Atlanta Dream have a certain Ms. A. McCoughtry who is known for scoring a lot of her team's points. I don't have hard data but I would estimate that the team's best player usually scores about 20 percent of her team's total points in a game. So I created a chart of all of Atlanta's games for the 2011 season. I listed the total score and compared that score to the score of the highest scoring player, and then calculated a percentage. Furthermore, I marked the result of the game, whether or not it was a loss or a win for the team.
|1||New York||88||19||Castro Marques||21.6%||L|
|10||New York||87||27||De Souza||31.0%||W|
Looking at the chart, you can tell that McCoughtry is clearly the go-to player for Atlanta. Four times during the season, she topped the 40 percent mark of total points scored for her team. During a game against Tulsa, she almost scored 50 percent of her team's points!
So how did Atlanta do in those four games? Two were regular season games, both won by the Dream. But two of those games were WNBA Finals games, and both of those games were lost by the Dream. McCoughtry scored 44.6 percent of the total points for the Dream in Game 1 - scoring 33 against the Lynx - but losing.
I decided to run a correlation between the proportion of total team points scored and whether or not the Atlanta Dream won. Correlation indicates the mathematical relationship between two variables. The question then becomes, "If someone scores a huge percentage of the team's points, is the Dream more likely to win?" A correlation of 0.5 (one is the highest) would be a strong indicator; a correlation of zero would be random chance and a correlation of -1 would be an inverse relationship.
What was the correlation? -0.06! That's virtually identical to random chance - no correlation at all between the proportion of total points scored and whether or not the Dream won. This shouldn't surprise anyone, and I didn't expect a correlation (but wished to explore if one existed). There can be several theories as to why there's no correlation:
1) There is more than one way to win a basketball game. One can distribute scoring duties among several players rather than have one player bear the burden.
2) Just because one player carries the burden, doesn't mean the other team doesn't have a player that can have pull similar duties. Ask Cappie Pondexter about Angel McCoughtry in 2010.
3) If a team is flattened, one player carrying 30 percent of the scoring burden doesn't mean much.
4) A team may have one player carrying the burden simply because there are no other good players on the team. I'm sure if you looked through some Tulsa box scores you'd find some player who had "good" nights relative to the rest of the team - but the Shock were terrible in 2011.
But surely, those nights when McCoughtry is the best player - as opposed to say, De Souza or Harding - should mean something. So I ran a correlation comparing wins or losses to games where McCoughtry is the best player. There should be a relationship between wins there, right.
No. The correlation was -0.02. If McCoughtry is the best player, she's going to be the best player during both wins *and* losses.
It just goes to show. Women's basketball is not a one-woman game, not yet. At least, not until Griner shows up.
(*) The reply from some is "what knowledge?"