Win Shares According to Plus/Minus

Now that the season is wrapped up and we're pretty much done picking through the wreckage of the NCAA tournament, I wanted to look back at the season as a whole and the production of individual players.  If we can ignore the last game, the season was a great one for Kansas basketball.  Ranked #1 for the majority of the year and a lot of growth for some of the young players with continued solid play from the older guys.  In my first look back here, I've taken the plus/minus numbers and calculated a plus/minus "Win Share".  Basically, each players share of the cumulative plus/minus for the season.  The new NCAA Basketball Reference tracks "win shares" for each past season and I thought it'd fun to see how my numbers compare to theirs once they are released. 

Below I've figured the per game averages for both my plus/minus and Game Score.  The final results are actually pretty close to each other except for a few discrepancies, which we'll look into and try to diagnose.  The college basketball win share definition is fairly complicated, as you can read about here.  As we all know, my system's very simple and the game score is slightly more complicated.  So if these turn out close, I'll be satisfied.  For turning the per game averages to a win share, I divided the player total by the team total to get a percentage of the total score.  That number was multiplied by 33 because Kansas finished with 33 wins.  There are some big picture issues with this but I think it works for now.  Onto the numbers...

First up, we have the numbers from 2008-2009, just to see if I'm wasting my time.

Off a decent amount on Cole Aldrich, Tyshawn Taylor, and Markieff Morris.  Not out of the ballpark though, so I feel halfway comfortable sharing the numbers for this year. 

2009-2010 win shares with plus/minus and game score.

I've got a feeling that plus/minus is shooting a little high on Aldrich again and a little low on Marcus Morris.  Wouldn't be surprised if the game score numbers end up very close to the basketball reference numbers at the end.  One thing I don't like doing but I haven't thought of a way to get around yet is how to deal with the losses.  The numbers are from 36 games and to find win shares, I divided by 33.  Not sure if that matters or if I'm missing an obvious way to handle that, I figure we've got enough smart guys around here to figure something out. 

And the last thing I've got is a graph comparing each players plus/minus and their game score.  There's no real clear pattern of one system favoring a certain position or type of player.  Collins, Henry, and Taylor are all slightly lower on mine than on game score.  But Reed, Johnson, and Morningstar are liked more by the plus/minus.  My numbers like Aldrich a little too much but that doesn't hold true for either of the Morris brothers.  I'm guessing the difference in the numbers comes primarily from me counting only field goal attempts, so someone that shoots often is getting punished.

Thoughts?