Tuesday, May 08, 2007

Storm 2006 Plus/Minus

I will start right off saying that my numbers differ from those compiled by Paul Swanson of the Minnesota Lynx and shared at the Lynx Lane fan site. I finished double checking each and every game of the season against the individual game stats posted and archived by Paul on the Official Lynx Blog. Paul performs a HUGE service for statistics fans and posts the +/- data for every WNBA game all season long. He started last season and they are all archived by day on the blog site.

Why do I still bother doing my own tedious plus/minus calculations with Paul making it easy? Well, Paul only gives us the individual player value and I am also interested in effective lineups and player pairs. With so many players returning this year, it is interesting to me to see which lineups have worked well in the past.

Why are the numbers different? Well, after going through all 34 games again checking each one individually against Paul's numbers, I determined that Paul had only one anomaly where he had a wrong value for a single player (probably scanned off a faulty play by play). All the remaining discrepancies were due to a difference in approach. I credit free throws to the players on the floor when the foul occurs in all instances, regardless of when they are subbed out of the game. Paul's numbers attribute fouls to players in the game when the play by play data posts a point. I believe that Paul is using an automated scanner (boy I wish I had one!!!). Since the play by play data sometimes subs the player before the free throws, or in between the free throws and or after the free throws, his stats are occasionally off mine by 1-4 points per game. The differences will mean that at least two players will be different in any game where this occurs, though in some games it was as many as six players (three substitutions during free throw attempts throughout the game). I am very confident with my numbers having tracked down every discrepency to this with the single player exception where again, I am certain of my accuracy (I ran that game manually five times.)

With that aside, here is what I have for the year using formulas or approaches borrowed from many people smarter about this stuff than I am.

Floor% : How often did the player score when TRYING to score.
Eff FG%: Balanced shooting percentage weighted for 3 point shooting.
Stop%: Weak team oriented view of individual defense.
Off Rtg: Points per 100 possessions
Def Rtg: Weak team oriented view of individual points allowed per 100 possessions
W-L%: Dean Oliver value for percentage of wins attributable to the player
Net Avg: Dean Oliver value for points per game using partial values for shots and assists
Pts Prod/G: Dean Oliver value for points produced per game
IR: Individual Rating from a formula I stole from Kevin Pelton
Net: Combines team performance vs. opponent for a player on the court and off
Net40: Roland Rating or Net per 40 minutes
NetG: Net per game

2006 Storm Player Results

Best Floor%
Lauren Jackson: 61%
Janel Burse: 52%
Wendy Palmer: 51%
Barbara Turner: 49%
Iziane Castro-Marques: 46%

Best EFG%
Lauren Jackson: 57%
Iziane Castro-Marques: 53%
Janel Burse: 51%
Wendy Palmer: 50%
Sue Bird: 50%

Best Stop %
Cisti Greenwalt: 50%
Wendy Palmer: 44%
Lauren Jackson: 42%
Janel Burse: 39%
Shaun Gortman: 38%

Best Def Rtg (Lower is better)
Lauren Jackson: 103.18
Cisti Greenwalt: 120.36
Wendy Palmer: 122.69
Janel Burse: 124.73
Shaun Gortman: 125.07

Best Off Rtg
Lauren Jackson: 133.42
Wendy Palmer: 104.82
Barbara Turner: 104.27
Janel Burse: 104.05
Sue Bird: 103.70

Best W/L%- Only one player here.
Lauren Jackson: 93%

Best Net Avg- Only one player here.
Lauren Jackson: 5.47

Best Pts Prod/G
Lauren Jackson: 18.01
Betty Lennox: 12.85
Sue Bird: 12.42
Janel Burse: 10.85
Wendy Palmer: 9.57

Best IR
Lauren Jackson: 119.00
Sue Bird: 93.53
Barbara Turner: 91.06
Iziane Castro-Marques: 90.71
Janel Burse: 88.39

Best Net +/-
Lauren Jackson: +343
Iziane Castro-Marques: +122
Janel Burse: +91
Sue Bird: +84
Wendy Palmer: +57

Best Net40
Wendy Palmer: +19.00
Lauren Jackson: +16.18
Iziane Castro Marques: +7.45
Janel Burse: +4.94
Sue Bird: +3.15

Best Net/G
Lauren Jackson: +11.43
Wendy Palmer: +11.40
Iziane Castro-Marques: +3.59
Janel Burse: +3/37
Sue Bird: +2.47

So the first thing I take away from this is that this would really have been a different team if they had not lost Wendy Palmer early in the season. She was strong in all the key statistical areas. How good can they be with her back in the lineup this year?

Tomorrow I will take a look at the most effective lineups.


p_d_swanson said...


I'm working manually as well. Perhaps the differences you've noted stem from the source material for play-by-play information. The data listed at WNBA.com and other sites can be incomplete or incorrect if the courtside inputters make revisions after the initial event transaction has been recorded...

Patrick said...

Thanks for the clarification. Do you ever sleep? 34 games for one team takes me forever, but for every team in the league! ACK!

p_d_swanson said...

It generally takes about 20 minutes to log each game, so a full slate might be two hours' worth. Not all that bad...

But to clarify, there are some technical issues with the software used to export the live data to Yahoo! or WNBA.com or ESPN that can create discrepancies. For example, Player A is charged with a foul with 4:39 left in the first quarter. That line goes up on the website instantly. But while the opponent is shooting her free throws, the table crew clarifies with the referee or official scorer that the foul instead was on Player B. That correction isn't seen on the live stats, as Player A is still shown with the foul...

It's the same way with substitutions. If they aren't all logged correctly immediately, it can cause havoc with how things are displayed in real time on the web. The final game file isn't uploaded to the league's bulletin board until all these validation errors are fixed, so the version I wind up with is already cleaned...