MD Plus/Minus Halfway Through ACC Play

No lofty introductions this time, though I've left my explanation of the methodology below.  This update includes the final non-conference numbers, the ACC numbers through 8 games, and the total numbers so far:

Name               OFF/DEF/TOT (non-conf) OFF/DEF/TOT (ACC 8 gm) OFF/DEF/TOT (overall)

Bowie              -8.24/ 0.66/ -7.58                    -0.03/ 2.98/ 2.95                      -5.25/ 1.50/ -3.75

Gregory            -8.13/ 3.92/ -4.20                    4.71/ -0.33/ 4.38                      -0.79/ 1.49/ 0.70

Hayes               3.84/ 0.75/ 4.59                       0.46/ 3.00/ 3.46                       2.61/ 1.57/ 4.18

Milbourne         4.91/ -0.11/ 4.79                      0.80/ 1.29/ 2.09                       3.42/ 1.21/ 4.63

Mosley              4.74/ 2.22/ 6.96                       -0.50/ -0.81/ -1.31                   2.74/ 1.07/ 3.81

Padgett             -9.21/ 0.44/ -8.77                    -23.35/ -20.36/ -43.71             -14.16/ -6.84/ -21.00

Vasquez           2.85/ -2.27/ 0.58                      1.68/ -1.50/ 0.17                      2.42/ -1.99/ 0.43

Williams           3.88/ -2.45/ 1.43                      3.45/ -0.80/ 2.66                      3.72/ -1.85/ 1.88

These numbers are essentially the cumulative value of the team's offensive and defensive performance with a player on the court, as a mean of the differences between on-court efficiency/40 minutes and the team's actual performance over 40 minutes.  So,

• OFF=avg MD offensive efficiency per 40 minute s above or below the overall performance with player x on the court
• DEF= the same for defensive efficiency, except that the values are reversed to limit confusion (so a positive number is good).
• TOT= OFF+DEF

 Some random thoughts on what the numbers show (and don't show):

• This is definitely not a be-all, end-all stat to measure a player's value...no such number exists, although UMTerps on insidemdsports gets pretty close.  Instea d, this provides another statistical angle that can be useful for comparing players against each other.  A margin of error (I would go with about +/- 2) can be assumed.
• I find it very interesting that several trends many have observed in traditional stats (pts, rebs, etc) as well as the "eye test" have also shown up in plus/minus numbers.  Most notable of these is the regression of Mosley from a consistent positive force to more of a role player.  Another one that GW apparently saw before many of us - Padgett may not be ready for ACC-level play yet.  It's hard to quantify this because of the small sample size, but when Padgett has played (not just in garbage time but also spot situations mid-game) the offense has ground to a halt and the overall play has degraded.  Conversely, Gregory has most consistently correlated to improved team play since the conference season started.  He is playing fewer minutes now, with Williams and Milbourne generally getting 60+ minutes at the 4 and 5, and seemingly making better use of those minutes.  The numbers for Tucker and Bowie have both been inconsistent in conference play, but much more so for Tucker.
• As for the seniors, I think you can more or less throw out any plus/minus numbers for Vasquez.  This is his team, he plays 35+ minutes per game, and more or less the team goes as he goes.  On the other hand, Hayes and Milbourne have both been steady for most of the season in terms of plus/minus.  I find it interesting that Hayes has the best overall DEF score (just ahead of Bowie) in conference play.  After the game at FSU where Kitchen dropped 29 on us, there was a flame fest over on IMS about how bad of a defender Hayes allegedly is.  IMO it was a bit overblown, and the +/- numbers show that there is little reason to think Hayes is an Achilles heel for us defensively.  I personally feel he is an average defender with a few rather pronounced weaknesses.  But it's hard to believe that we could be the 22nd-best defensive team in the country (see kenpom.com) while playing a bad defensive player 30-35 minutes per game.

Please discuss, and I welcome any questions or feedback on methodology or what the numbers show.