Latrinsorm
07-28-2013, 09:29 PM
Synergy Sports breaks down every NBA possession finish* into one of 11 categories:
Isolation
Pick and Roll (Handler)
Post-Up
Pick and Roll (Roll)
Spot-Up
Off Screen
Hand Off
Cut
Offensive Rebound
Transition
Other (anything that won't fit in the above, mostly half-court heaves and deflections)
*Of course, almost no NBA play is just a spot-up, just a cut, etc. You might hand off, run a pick and roll, back out, isolate, post-up, and then pass to an open shooter for a spot-up look. That would count as a spot-up because that's how the play finished.
And I wondered, how might these plays shake out in terms of points per possession and % of plays run? Well I added up every team in the NBA and got...
ppp %Run Type
1.17 8.36% Cut
0.88 6.50% Hand Off
0.81 19.85% Isolation
1.08 13.14% O Rebound
0.88 4.95% Off Screen
0.42 13.17% Other
0.78 9.42% P&R Handler
0.99 6.30% P&R Roll Man
0.82 2.58% Post-Up
0.97 9.82% Spot-Up
1.12 5.91% Transition
...or, ignoring transition/OReb/other (because those aren't really plays you can draw up) and sorting by ppp, we have...
ppp %Run Type
1.17 11.24% Cut
0.99 8.68% P&R Roll Man
0.97 26.62% Spot-Up
0.88 6.64% Off Screen
0.88 3.40% Hand Off
0.82 12.67% Post-Up
0.81 13.18% Isolation
0.78 17.58% P&R Handler
...which makes sense, mostly. I was surprised at how abysmal the points generated by P&R handler and post-up were, and while error bars are always an issue they're only about .01, which isn't going to change things very dramatically. Cuts are clear #1s, roll and spot-up in the second tier, off screen and hand-off in the third tier, then the last three bringing up the rear. Another handy thing Synergy tracks is the % of times any particular play draws a foul, but that breaks down almost exactly as you expect: nobody fouls shooters (spot-up, screen, hand off); everybody fouls the post, the roll, and the cut; isolation and P&R handler are in the middle.
.
How about 3s? Synergy doesn't count any 3s as coming from cuts or posting up, which leaves us with:
http://img.photobucket.com/albums/v456/johnnyoldschool/NBASynergy3Types_zps58660449.jpg
Obviously it's rare for an offensive rebound to bounce out all the way to an open three point shooter, which is why the error bars are so ginormous on that category and we can't tell anything about it. The only statistically significant conclusions we can draw are...
1. Spotting up (very small error bar) is better than hand off, P&R handler, and isolation.
2. Isolation (very low value) is worse than everything.
3. Transition is better than P&R handler.
4. Everything else is indistinguishable, and it's a little surprising to me that pick and pop 3s stood up so well.
Ok, but that's NBA league average, surely my team/player is actually good at isolation 3s? I haven't done every player yet, but as it turns out the 30 NBA teams break down as 26 indistinguishable and 4 significantly worse at isolation 3s than their overall values: Boston, Detroit, Milwaukee, Oklahoma City. In Detroit's defense, they obviously realized this was a terrible play for them and took the second least isolation 3s of any team. OKC took the second most, so... I guess shock collars are in order? Seriously though you can guess why this would happen: Westbrook can't shoot, Westbrook has terrifying quickness so teams play 15 feet off him, Westbrook has no sense so he takes the shot. We'll see if that's what actually happened, though getting 450 some odd players done is going to take a lot longer than 30 teams.
Isolation
Pick and Roll (Handler)
Post-Up
Pick and Roll (Roll)
Spot-Up
Off Screen
Hand Off
Cut
Offensive Rebound
Transition
Other (anything that won't fit in the above, mostly half-court heaves and deflections)
*Of course, almost no NBA play is just a spot-up, just a cut, etc. You might hand off, run a pick and roll, back out, isolate, post-up, and then pass to an open shooter for a spot-up look. That would count as a spot-up because that's how the play finished.
And I wondered, how might these plays shake out in terms of points per possession and % of plays run? Well I added up every team in the NBA and got...
ppp %Run Type
1.17 8.36% Cut
0.88 6.50% Hand Off
0.81 19.85% Isolation
1.08 13.14% O Rebound
0.88 4.95% Off Screen
0.42 13.17% Other
0.78 9.42% P&R Handler
0.99 6.30% P&R Roll Man
0.82 2.58% Post-Up
0.97 9.82% Spot-Up
1.12 5.91% Transition
...or, ignoring transition/OReb/other (because those aren't really plays you can draw up) and sorting by ppp, we have...
ppp %Run Type
1.17 11.24% Cut
0.99 8.68% P&R Roll Man
0.97 26.62% Spot-Up
0.88 6.64% Off Screen
0.88 3.40% Hand Off
0.82 12.67% Post-Up
0.81 13.18% Isolation
0.78 17.58% P&R Handler
...which makes sense, mostly. I was surprised at how abysmal the points generated by P&R handler and post-up were, and while error bars are always an issue they're only about .01, which isn't going to change things very dramatically. Cuts are clear #1s, roll and spot-up in the second tier, off screen and hand-off in the third tier, then the last three bringing up the rear. Another handy thing Synergy tracks is the % of times any particular play draws a foul, but that breaks down almost exactly as you expect: nobody fouls shooters (spot-up, screen, hand off); everybody fouls the post, the roll, and the cut; isolation and P&R handler are in the middle.
.
How about 3s? Synergy doesn't count any 3s as coming from cuts or posting up, which leaves us with:
http://img.photobucket.com/albums/v456/johnnyoldschool/NBASynergy3Types_zps58660449.jpg
Obviously it's rare for an offensive rebound to bounce out all the way to an open three point shooter, which is why the error bars are so ginormous on that category and we can't tell anything about it. The only statistically significant conclusions we can draw are...
1. Spotting up (very small error bar) is better than hand off, P&R handler, and isolation.
2. Isolation (very low value) is worse than everything.
3. Transition is better than P&R handler.
4. Everything else is indistinguishable, and it's a little surprising to me that pick and pop 3s stood up so well.
Ok, but that's NBA league average, surely my team/player is actually good at isolation 3s? I haven't done every player yet, but as it turns out the 30 NBA teams break down as 26 indistinguishable and 4 significantly worse at isolation 3s than their overall values: Boston, Detroit, Milwaukee, Oklahoma City. In Detroit's defense, they obviously realized this was a terrible play for them and took the second least isolation 3s of any team. OKC took the second most, so... I guess shock collars are in order? Seriously though you can guess why this would happen: Westbrook can't shoot, Westbrook has terrifying quickness so teams play 15 feet off him, Westbrook has no sense so he takes the shot. We'll see if that's what actually happened, though getting 450 some odd players done is going to take a lot longer than 30 teams.