Friday, January 7, 2011

Plus/Minus Versus Hope

Here's the game chart:

Player(+)(-)(+/-)MIN(+)/min(-)/min(+/-)/min
Brink65605242.712.500.21
DeBoer1214-252.402.80-0.40
DeYoung1814492.001.560.44
Fabiyi00000.000.000.00
Greene00000.000.000.00
Haverdink47416162.942.560.38
Kruis42411192.212.160.05
Powell56497262.151.880.27
Rietema00000.000.000.00
Rodts755718302.501.900.60
Salo00000.000.000.00
Schnyders41347192.161.790.37
Schuster654223262.501.620.88
Snikkers54531262.082.040.04
Vallie00000.000.000.00
Team475405702002.382.030.35

The most interesting number here, to me, is Brent Schuster's +23. Calvin won the game by 14, so that means they were outscored by nine points when Brent was on the bench.

Season Totals: (sorted by overall +/-)

Player(+)(-)(+/-)MIN(+)/min(-)/min(+/-)/min
Brink465381842112.201.810.40
Schuster450372782122.121.750.37
Snikkers505437682452.061.780.28
Powell464406582302.021.770.25
Rodts540489512682.011.820.19
DeBoer230202281112.071.820.25
Schnyders329312171642.011.900.10
Haverdink312298141641.901.820.09
DeYoung1771698882.011.920.09
Fabiyi81742.000.251.75
Greene00000.000.000.00
Rietema00000.000.000.00
Salo00000.000.000.00
Vallie2324-1151.531.60-0.07
Kruis147149-2901.631.66-0.02
Team3650324041018022.031.800.23

I'm kind of surprised that Tyler Kruis isn't scoring better on offense here, but if we look at the numbers, this chart appears to be validated. The average (+/min) for the ten players that make up the Calvin rotation is 2.00, and Tyler has been credited with a 1.63, that's good for a 0.815 ratio (player to team). This is very similar to the 0.830 ratio we find between his personal points per weighted shot average (0.94) and the average of the same ten player rotation (1.13).

This sort of calculation doesn't follow exactly for all of the players, but it's getting me close in most cases. Using this calculation is grossly underestimating Danny Rodts' scoring efficiency, but mostly we're in the ballpark. This either means that we're only ever going to get in the ballpark or we just need more data. I'm guessing it's data. This method has to take A LOT longer to settle down than using actual point totals; there's still a decent amount of noise in the data.

I wouldn't guess that +/- would be the best way (or even a particularly good way) to judge individual offensive abilities given the ease with which we can find traditional offensive stats, but the fact that it's even approximating individual offensive efficiency (even if rather loosely), give me confidence that the defensive numbers also mean something (even if they're not perfect).