Post by eric on Jul 22, 2015 18:08:20 GMT -6
Building off the methodology described here, I tested a point guard to see what his attributes were worth when it came to team Win Shares. One flaw I discovered is that the methodology is based on comparing against a base sample of 20 seasons, so if the base happened to be good or bad (due to random noise) it would skew the rest of the results. It turned out the center base was bad and the point guard base was good, so I adjusted for the position-to-position comparison at the end of the article.
Here's the meat:
All of the following reflects the center case. The point guard was lousy, so anything that increases usage will appear worse than it really is. The big boys for increasing usage were Inside Scoring (157), Strength (105), and Jumping (56): compare 234/219/103. Only Inside Scoring increased team pace, so created more uses rather than just taking them from someone else on the team. Passing increased usage a little at 27 for the point guard to 40 for the center. Handling was the only attribute to decrease usage, but it decreased turnovers more than usage so it was in a good way. Almost every change in the box score was the same for point guard or center, it was just more or less pronounced along obvious lines: Shot Blocking produced 57 for the center to 20 for the point guard, Passing made 61 assists for the point guard to 42 for the center. Height had the same weird effect of increasing rebounds while decreasing blocks.
Here are the tws to attributes side by side, both corrected for the base deviation:
I think a lot of these values are the same, but the error bars are just a little too big. It's going to be a long time until I do another one of these attribute analyses, but I think the general idea is clear. Inside Scoring is awesome for everyone, Handling, Shot Blocking, and Strength are really good, everything else is situational.
Here's the meat:
tws characteristic
0.3 pound
2.0 inch
-0.5 experience
1.2 inside scoring
0.8 jump shot
0.8 three shot
1.7 handling
1.2 quickness
0.4 passing
-0.5 stealing
1.0 shot blocking
0.5 post defense
1.1 perimeter defense
0.3 drive defense
-0.2 offensive rebounding
0.3 defensive rebounding
0.5 strength
0.1 jumping
tws characteristic
1.7 handling
1.2 quickness
1.2 inside scoring
1.1 perimeter defense
1.0 shot blocking
0.8 three shot
0.8 jump shot
0.5 strength
0.5 post defense
0.4 passing
0.3 drive defense
0.3 defensive rebounding
0.1 jumping
-0.2 offensive rebounding
-0.5 stealingAll of the following reflects the center case. The point guard was lousy, so anything that increases usage will appear worse than it really is. The big boys for increasing usage were Inside Scoring (157), Strength (105), and Jumping (56): compare 234/219/103. Only Inside Scoring increased team pace, so created more uses rather than just taking them from someone else on the team. Passing increased usage a little at 27 for the point guard to 40 for the center. Handling was the only attribute to decrease usage, but it decreased turnovers more than usage so it was in a good way. Almost every change in the box score was the same for point guard or center, it was just more or less pronounced along obvious lines: Shot Blocking produced 57 for the center to 20 for the point guard, Passing made 61 assists for the point guard to 42 for the center. Height had the same weird effect of increasing rebounds while decreasing blocks.
Here are the tws to attributes side by side, both corrected for the base deviation:
pg c_ characteristic
0.3 -0.4 pound
2.0 2.1 inch
-0.5 0.5 experience
1.2 1.0 inside scoring
0.8 0.3 jump shot
0.8 0.0 three shot
1.7 0.7 handling
1.2 -0.4 quickness
0.4 -0.2 passing
-0.5 0.6 stealing
1.0 2.1 shot blocking
0.5 0.5 post defense
1.1 0.1 perimeter defense
0.3 -0.2 drive defense
-0.2 0.9 offensive rebounding
0.3 0.6 defensive rebounding
0.5 0.9 strength
0.1 -0.3 jumpingI think a lot of these values are the same, but the error bars are just a little too big. It's going to be a long time until I do another one of these attribute analyses, but I think the general idea is clear. Inside Scoring is awesome for everyone, Handling, Shot Blocking, and Strength are really good, everything else is situational.
