Post by eric on Oct 8, 2015 12:35:59 GMT -6
Short version: SHOT BLOCKING!!!!!
Longer version: here are the values from the previous methodology (to the left) and the current one (in the middle)...
Inside Scoring, Shot Blocking, and Strength are still by far and away the best. (Remember that the base player is a terrible scorer and Inside Scoring gives the biggest bump to usage, so even though it makes him a better scorer he's still pretty terrible and thus the wins added to the team are deflated.) A lot of the other middling attributes look a lot better now though, but Handling has not kept pace.
Much longer version:
Okay. So minutes played got pretty weird. The attributes that caused increase in playing time for small forward were all around 15% increases, and the distance from top to bottom for each group was 0.5% and 0.8% respectively; that's why I was so comfortable assigning them discrete groups. For centers, Inside Scoring and Strength were bang on next to each other at 13% above base, and everybody else was in a continuum that's really hard to divide but whose highest member (Shot Blocking) was over 2% higher than its lowest (Passing). Even more troubling, the highest five members are Shot Blocking and both Reboundings (makes sense!) and Drive Defense and Quickness (not so much). My guess is there are three tiers, but the one that's very slightly higher than base is too close to call with this amount of data. Inside Scoring and Strength are definitely good (what else is new).
.
Usage went mostly like I saw before with a few crucial differences:
Numbers are given to the nearest half a use per 36 minutes. It's important to look at the particulars of usage here, though.
-Inside Scoring dramatically increased FGr (the portion of uses that were FGAs) and FTr (the portion that were FTAs) while dramatically decreasing TOr (the portion that were TOVs) and 3Pr (the portion that were 3PAs), even though raw turnovers increased. It's a little surprising that TOr went down so much, but you can make up an explanation for that pretty easily when you notice that assists went up quite a bit: instead of getting trapped and trying to force a pass into double coverage, the player is throwing passes out of double coverage and those are pretty easy to get assists on.
-Jump Shot slightly increased FGr at the cost of FTr and TOr, but did not change 3Pr. It's the only way to increase FT% but it also means your guy will be taking more long twos, which are bad for everyone but especially big men who you want in the paint.
-Handling as always was the only attribute to cut raw turnovers, and did so by replacing them with shots and free throw attempts. Counter-intuitively, the attribute that decreases usage suffers the same deflationary effect as those that increase it, but Handling still needs to be downgraded in my grade scheme.
-Passing has a really interesting dynamic. It increases the number of assists and turnovers, but you get more assists than turnovers and so the ast:tov ratio increases. That's not the interesting part. The interesting part is that it's free throw rate that drops dramatically, while three point rate increases almost as much as it does for Three Point attribute. Clearly Passing has some mechanism whereby elite passers will take more jump shots including (and presumably not limited to) three pointers. This makes it a really hard attribute to extrapolate Win-wise: it's reasonable to assume a low Handling player won't get as much out of Passing upgrades, but having to include Outside Scoring as well is a surprise. To me this makes Passing an even worse attribute to increase for big men, who traditionally have poor Outside Scoring grades, but a much better attribute to increase for guards, who traditionally have excellent Outside Scoring grades.
-Offensive Rebounding gives a little bump to FTr and a little ding to 3Pr. Obviously.
-Defensive Rebounding is the only attribute to significantly increase TOr, which is where all its USG increase comes from. It could be a fluke, but you can paint a picture where it makes sense: ORBs go right back at the basket, DRBs go to a teammate, passes can get picked off...? Maybe. I went back and checked SF and they have a similar increase though not as big. We'll see if PF have the same deal down the road.
-Strength and Jumping have the same usage profile as Inside Scoring, which makes sense since they all factor into Inside Scoring grade.
.
Finally, I realized I messed up the SF numbers so here they are corrected:
Rebounding (including Strength) turned out to be a lot more important, otherwise it basically spread the numbers out more, making the best more pronounced. I'm going to condense this and update the New Guys post right.... now!
Longer version: here are the values from the previous methodology (to the left) and the current one (in the middle)...
ws w att
1.0 1.7 inside scoring
0.3 0.8 jump shot
0.0 0.0 three shot
0.7 0.4 handling
-0.4 0.2 quickness
-0.2 0.2 passing
0.6 1.2 stealing
2.1 3.6 shot blocking
0.5 1.6 post defense
0.1 0.7 perimeter defense
-0.2 0.2 drive defense
0.9 2.1 offensive rebounding
0.6 1.6 defensive rebounding
0.9 1.6 strength
-0.3 -0.2 jumping
Inside Scoring, Shot Blocking, and Strength are still by far and away the best. (Remember that the base player is a terrible scorer and Inside Scoring gives the biggest bump to usage, so even though it makes him a better scorer he's still pretty terrible and thus the wins added to the team are deflated.) A lot of the other middling attributes look a lot better now though, but Handling has not kept pace.
Much longer version:
Okay. So minutes played got pretty weird. The attributes that caused increase in playing time for small forward were all around 15% increases, and the distance from top to bottom for each group was 0.5% and 0.8% respectively; that's why I was so comfortable assigning them discrete groups. For centers, Inside Scoring and Strength were bang on next to each other at 13% above base, and everybody else was in a continuum that's really hard to divide but whose highest member (Shot Blocking) was over 2% higher than its lowest (Passing). Even more troubling, the highest five members are Shot Blocking and both Reboundings (makes sense!) and Drive Defense and Quickness (not so much). My guess is there are three tiers, but the one that's very slightly higher than base is too close to call with this amount of data. Inside Scoring and Strength are definitely good (what else is new).
.
Usage went mostly like I saw before with a few crucial differences:
c usg
100 inside scoring
5 jump shot
0 three shot
-5 handling
0 quickness
30 passing
0 stealing
0 shot blocking
0 post defense
0 perimeter defense
0 drive defense
5 offensive rebounding
5 defensive rebounding
85 strength
35 jumpingNumbers are given to the nearest half a use per 36 minutes. It's important to look at the particulars of usage here, though.
-Inside Scoring dramatically increased FGr (the portion of uses that were FGAs) and FTr (the portion that were FTAs) while dramatically decreasing TOr (the portion that were TOVs) and 3Pr (the portion that were 3PAs), even though raw turnovers increased. It's a little surprising that TOr went down so much, but you can make up an explanation for that pretty easily when you notice that assists went up quite a bit: instead of getting trapped and trying to force a pass into double coverage, the player is throwing passes out of double coverage and those are pretty easy to get assists on.
-Jump Shot slightly increased FGr at the cost of FTr and TOr, but did not change 3Pr. It's the only way to increase FT% but it also means your guy will be taking more long twos, which are bad for everyone but especially big men who you want in the paint.
-Handling as always was the only attribute to cut raw turnovers, and did so by replacing them with shots and free throw attempts. Counter-intuitively, the attribute that decreases usage suffers the same deflationary effect as those that increase it, but Handling still needs to be downgraded in my grade scheme.
-Passing has a really interesting dynamic. It increases the number of assists and turnovers, but you get more assists than turnovers and so the ast:tov ratio increases. That's not the interesting part. The interesting part is that it's free throw rate that drops dramatically, while three point rate increases almost as much as it does for Three Point attribute. Clearly Passing has some mechanism whereby elite passers will take more jump shots including (and presumably not limited to) three pointers. This makes it a really hard attribute to extrapolate Win-wise: it's reasonable to assume a low Handling player won't get as much out of Passing upgrades, but having to include Outside Scoring as well is a surprise. To me this makes Passing an even worse attribute to increase for big men, who traditionally have poor Outside Scoring grades, but a much better attribute to increase for guards, who traditionally have excellent Outside Scoring grades.
-Offensive Rebounding gives a little bump to FTr and a little ding to 3Pr. Obviously.
-Defensive Rebounding is the only attribute to significantly increase TOr, which is where all its USG increase comes from. It could be a fluke, but you can paint a picture where it makes sense: ORBs go right back at the basket, DRBs go to a teammate, passes can get picked off...? Maybe. I went back and checked SF and they have a similar increase though not as big. We'll see if PF have the same deal down the road.
-Strength and Jumping have the same usage profile as Inside Scoring, which makes sense since they all factor into Inside Scoring grade.
.
Finally, I realized I messed up the SF numbers so here they are corrected:
sf attribute
2.9 inside scoring
2.4 jump shot
1.4 three shot
1.1 handling
1.2 quickness
-0.2 passing
1.0 stealing
2.6 shot blocking
1.0 post defense
1.8 perimeter defense
0.4 drive defense
2.1 offensive rebounding
1.5 defensive rebounding
2.0 strength
0.1 jumpingRebounding (including Strength) turned out to be a lot more important, otherwise it basically spread the numbers out more, making the best more pronounced. I'm going to condense this and update the New Guys post right.... now!
