Team by Team ORating after an initial offensive rebound?

[YaoPau]

I'm looking for each team's offensive rating after they pull down an initial offensive rebound. So for example, if Joakim Noah grabs the offensive board, shoots and misses, then tips it back in, that counts as a single possession, just like how normal orating is calculated.

Does someone have this info?

[YaoPau]

Just to continue where I'm headed with this...

I've always heard the long 2pt shot is the worst shot in basketball. But using Synergy, we can estimate expected points given a regular, half-court possession.

The Chicago Bulls are at 0.94 PPP overall, putting them just above league average, but that includes some non-halfcourt data, like possessions occurring directly after an offensive rebound, in transition, in the penalty, end of half heaves, and some other weird ones like quick shots after a long offensive rebound.

I removed those as best I could (there's no FT/FTA info for free throws that follow a non-shooting foul, when the opponent is in the penalty), and the Bulls had about a 0.87 PPP in the halfcourt this season. That includes all pick and rolls, spot ups, screens, iso plays, hand offs, and cuts, and does not include free throws that occur on non-shooting fouls when the opponent is in the penalty. Maybe it's arguable that some of those free throws should be included, in which case the PPP would tick up a point or two. So...0.9ish.

Anyway, that means if a player can hit an open 2 around 45% of the time (I'd guess that's a decent chunk of NBA players, since hoopdata numbers include all shots from 16-23 feet away and not just open ones, and few dozen players are near that 45% mark overall), and that player is open 18ft from the hoop with the ball in his hands, it's a decent play to take that jumpshot. That might be negated somewhat by the low probability of getting an offensive rebound off a miss, especially if it's a big man shooting the 18 footer.

Not sure if that conclusion's been discussed here, and maybe that's obvious to every NBA coach ever, but whenever I'd see a long open 2 I think "bad shot", and I'm changing my opinion on that.

[EvanZ]

This doesn't directly answer your question, but I did write a couple of posts recently on the 16-23 ft jumper you may find interesting. I share some of your same questions.

http://thecity2.com/2011/03/20/monta-el ... long-shot/

http://thecity2.com/2011/03/27/long-shots-part-deux/

[Crow]

Your thinking about the value of a shot and the possibility of an offensive rebound as a sequence (from a team perspective) is more compatible with EZPM and PER stat weights for missed shots than some other metrics.

The appropriate stat weights can vary depending on your perspective. Either every action is distinct and valued separately or you recognize context and assign value conditioned by the act before and / or after it. Some systems are all one, some have a few conditional values, some may have more, or I guess SPM from regression or APM is all derived conditional / contextual values. Systems with just a few conditional / contextual values may be both popular and controversial. Should one use conditional / contextual values for defensive rebounds, missed shots and assisted FGs and that is it? It could be done in more places or less.

[YaoPau]

Cool post Evan. I like seeing those assisted FG% numbers around 50% for the top guys, kinda syncs with what I was hoping for.

As for it matching PER or EZPM, I don't think it's particularly useful in a player rater. Maybe, I dunno. But just looking through Synergy I've been focusing on sources of sustainable offense, ways that teams are able to consistently generate an above average PPP. Obviously for the Jazz for years it was through PnR, for the Bulls this year it was Rose in isolation, but I wouldn't have guessed that running a play to generate an open 18footer for, say, Pau Gasol, would be similarly efficient. But I'm pretty sure it is. (Gasol, for example, is generating 1.11 PPP on spot-up attempts, almost all from 2pters this year.)

[EvanZ]

Pau is very efficient, so the question is why doesn't he take more of those and Kobe less? Or is Pau so efficient because he takes so few, and only when he's wide open, early in the shot clock, etc...

[DSMok1]

From an old thread ("Efficiency in different possession start cases")

I've done a study on the average offensive efficiency in different possession start cases, using data from play-by-plays of the Spanish basketball league. You can find it at
http://www.muevetebasket.es/cont.php?id=500

It's in Spanish, so I'll translate the main table here

Code: Select all

Possesion start             Efficiency
Start of quarter            104.7
2p FG made                  107.0
3p FG made                  106.9
FT made                     108.4
2p FG missed, def. reb.     112.5
3p FG missed, def. reb.     107.3
FT missed, def. reb.        103.2
2p FG missed, off. reb.     124.5
3p FG missed, off. reb.     115.5
Steal                       126.8
Non-steal turnover          111.6
Average                     110.4

This is based on 280 games from last season (18 teams, 31 games per team). In all cases there was at least 1000 possessions observed. I know there's some bias in the data, because the distribution of cases is not the same for all teams, but I don't think it has much impact on the results. The figures for offensive rebounds are the expected offensive efficiency after the rebound was taken. This means those possessions are taken into account both for the OR case and whatever was the true start of the possession.

First thing is, is there any similar study on the NBA? What I've found at 82games.com is only partial data for some cases and they tend to focus on eFG% instead of efficiency.

Now, my comments on the results:
- Not surprisingly, the way your opponent scores (2p, 3p or FT) does not affect your efficiency much.
- But it's a different story if the possession starts with a defensive rebound. Most teams did better after an opponent missed 2p FG. There was a significant exception, though. Sergio Rodriguez's team managed an efficiency of 119 after missed 3-pointers by taking advantage of the excellent rebounding ability of their wing players and their love for the fastbreak. I have to say that in the season before, 2003-04, the efficiency in the 3p case was slightly better than in the 2p case.
- It's easier to put back the ball when an offensive rebound follows a 2 point shot than when the shot is made behind the arc.
- Good defense improves your offense by around 5%. Same in the other way around, good offense improves defense. I love it.
- A steal is a great offensive weapon, more than an offensive rebound.

That should give a pretty good relative value for ORB, though it's obviously from a different context.

In that same thread, Kevin Pelton refers to some other numbers compiled by Roland Beech, but the link (to another thread) is dead.

EvanZ, you could compile these same numbers fairly easily from your database, couldn't you? Perhaps even by team? Have they been compiled elsewhere (say, Synergy?)

[Mike G]

I can't help but want to rank these by their efficiency.

Code: Select all

Effic.   Possession start
126.8   Steal
124.5   (2p FG missed, off. reb.)

115.5   (3p FG missed, off. reb.)
112.5   2p FG missed, def. reb.
111.6   Non-steal turnover

110.4   Average

108.4   FT made
107.3   3p FG missed, def. reb.
107.0   2p FG made
106.9   3p FG made

104.7   Start of quarter
103.2   FT missed, def. reb.

Of course, we don't generally refer to on OReb as the start of the possession, but that's what the opening post was about.
I don't see "after a timeout" here. It's appeared in some such breakdown before.

[EvanZ]

I wouldn't say "easily" (otherwise, I would do it immediately), but it's something I could do, if it was considered very useful.

[EvanZ]

I wonder if you are on a team with a prolific offensive rebounder, should that be reflected in your own offensive efficiency. For example, in ezPM I could vary the baseline PPP according to these efficiency numbers (well, their NBA equivalents). Interesting...

Example 2: The marginal value of a 2-pt FG after a steal would be 2-1.27, instead of 2-1.08.

[DSMok1]

That's certainly something to consider. I don't know who you'd give the credit to, though. Probably all of the credit should be to the offensive rebounder/whoever puts it back in, since they are the ones creating the marginal value above league average, not the one shooting it.

Of course, if you've got really good offensive rebounders, you might be inclined to take more low percentage shots from the outside?

On the other, hand possession START numbers should definitely be adjusting the value of the offensive production after it, in my opinion. So--after a steal, the expected points are 1.27. So whoever got the steal (and some credit to other teammates) should definitely get credit for that 1.27 - 1.0X, whatever the average offensive possession is.

[Bobbofitos]

I can't help but want to rank these by their efficiency.

Code: Select all

Effic.   Possession start
126.8   Steal
124.5   (2p FG missed, off. reb.)

115.5   (3p FG missed, off. reb.)
112.5   2p FG missed, def. reb.
111.6   Non-steal turnover

110.4   Average

108.4   FT made
107.3   3p FG missed, def. reb.
107.0   2p FG made
106.9   3p FG made

104.7   Start of quarter
103.2   FT missed, def. reb.

Of course, we don't generally refer to on OReb as the start of the possession, but that's what the opening post was about.
I don't see "after a timeout" here. It's appeared in some such breakdown before.

Possessions after timeouts are dead-ball situations, which I believe are the lowest EV possessions. (Similar to start of quarters)

[Bobbofitos]

I wonder if you are on a team with a prolific offensive rebounder, should that be reflected in your own offensive efficiency. For example, in ezPM I could vary the baseline PPP according to these efficiency numbers (well, their NBA equivalents). Interesting...

Example 2: The marginal value of a 2-pt FG after a steal would be 2-1.27, instead of 2-1.08.

This is a good way of looking at things, since every action is technically the EV gained or lost from the individual possession value, rather than the overall average value of a possession. So players that have a line that goes like:
Steal
Layup
Shouldn't get the credit for the full FG after the layup, since their positive value - the difference between the possession type and the 2pt FGM - is not as much as say, in a halfcourt set where the 2 pt FGM is now relative to the avg value of that possession. (To further confuse matters, you can weigh in the bonus/out of the bonus, shot clock time remaining, etc.)

[EvanZ]

To make it more complicated, we have events like "steals after defensive rebounds"!

[EvanZ]

I would hate to spend all the time implementing such a system only to find the differences are in the range of the noise.

I wish I knew the answer a priori.