APBRmetrics

[t50fox]

"The first thing you have to do is curtail the individual desire of the boy in the interest of the team. Then you have to correct two deficiencies every boy has-in playing defense and recognizing the value of ball possession".-Adolph Rupp
"Possessions are the currency of basketball"-John Hollinger Pro Basketball Prospectus 2002.
The question is what type of possession are these two refering to ?
The Hollinger/Oliver method of measureing possessions is :
Possessions=FGA- O Reb + TO + (FTA * .44).
I refer to this as a continuation possession. I believe the inclusion of offensive rebounds into the equation really masks offensive efficiency for a team. As a hobby, for many years I've used the following equation to measure possession stats and produce a points per possession percentage:
Offensive Possessions= Opp FGM + Opp FTM*.5 + Opp TO + Reb +164
Defensive Possessions= FGM + FTM*.5 + TO + Opp Reb + 164
The 164 are the possessions that each team recieves to start a period of an NBA game times 82.
I felt the possession begins whenever the team reclaims the ball and ends when they shoot or turn it over.
To illustrate the difference lets look at a game scenario. My team has the ball and sets the offense to run a play. We get off a shot but it misses, and my forward rebounds the ball and kicks it out to the guard. He again resets the offense and we shoot and miss a second time. My center gets the rebound and puts it in. The continuation possession says that my offense produced a rating of 2.00 (one possession = two points). My rating shows an offense with an efficiency of .67 (three possessions= two points). In reviewing the game scenario which rating more accurately reflects that offenses efficency in scoring ?
There are numerous season examples as well regarding the effects of offensive rebounds to efficency rating:
In 1998 the Hollinger offensive efficency rating showed that New Jersey had an offensive effeciency of 104.76 to rank 7th best in the league. Phoenix was rated 9th best with an efficency of 104.53. Using my rating I had New Jersey (91. as 15th best and Phoenix (94.5) ranked 6th best.
The following are the actual number comparisons:
New Jersey Phoenix
3pt Adj FG % 46.3% 49.9%
FT % .744 .749
Assists 1685 2124
Turnovers 1180 1236
Rebounds 3481 3441
Off Rebounds 1344 991
Points 8170 8170
How can a team with lower shooting perentages, fewer assists and equivelent rebounds be rated so much higher then another ? It's the removal of the offensive rebounds from the equation.
The Hollinger possessions by team are Jersey 7798 and Phoenix 7812.
I calculated the possessions as New Jersey 8970 and Phoenix 8643.
There lies the fundemental difference in the two rating systems. New Jersey was able to score as many points as Phoenix by dominating the offensive glass. I would call that a energetic offense but hardly a effiecent one.
I could make several more comparions (Sacremento-Dallas in 04) and Golden State-Sacremento in 03), but I'll let you run the numbers instead.
Finally, if the most important stat in winning basketball is FG % then should'nt a offensive effiecency rating reflect that ? In many cases with the continuation possession that fact is clearly distorted.

[WizardsKev]

I think there's value in looking at possessions (what you'd call continuation possessions) as well as plays (what you call "possessions"). To me, the value in "continuation possessions" is that they're approximately equal for a game. Teams take turns with the ball until the clock runs out. The team with the better offensive rating wins. In this evaluation, offensive rebounds boost offensive efficiency by giving a team more shots at the basket on some possessions.

Looking at "plays" also has value, in my opinion, because it can tell us how efficiently a team's offense functions if it never got an offensive rebound. That's useful information as well.

In my opinion, the "continuation" possession is a little more useful because they're approximately equal in a game.

To me, a possession starts when Team A gets the ball and ends when Team B gets it back from them. The other thing you're describing -- where offensive rebounds are not subtracted -- is (in Dean's parlance, which I agree with) a "play."

Both useful, though I think the "possession" data is more valuable for evaluating relative team strength.

[HoopStudies]

The concept of the same number of possessions for both teams is so fundamental to what we have accomplished, there is no way we're going to change it now. It is an extremely valuable way to look at the game, as fundamental as outs are to baseball. The concept has been around for so long because it is so valuable. Being efficient with outs is what wins in baseball. Being efficient with possessions is what wins in basketball. Period.

We have given the name "plays" to what I think you're calling possessions. Feel free to estimate those in a better way. Feel free to improve the estimation of possessions (as I posted some data to help a few days ago). But don't try to change the name. And points per possession, points per 100 possessions, rating, efficiency rating -- among the many labels for the same thing -- it will remain the standard for evaluating teams because it defines what wins.
_________________
Dean Oliver
Consultant to the Seattle Supersonics
Author, Basketball on Paper
http://www.basketballonpaper.com

[Mike G]

[HoopStudies]

[hpanic7342]

t50fox, why don't you think that dominating the offensive glass is part of an efficient offense? If an offense can squeeze in an extra shot more often than another team before giving up the ball, that sounds to me like it contributes tremendously to anyone's notion of "efficiency."

Offensive rebounding is an offensive skill, just like shooting, passing, and not turning the ball over. Any attempts to measure offense should take this into account.

[bchaikin]

i've always used -

team possession - anytime a team has the ball and either scores (so the other team gets the ball) or looses the ball to the other team, regardless of what happens in between. a steal in the backcourt and immediate shot and score for a 1 sec team possession is equivalent to a team possession that has several non-shooting fouls called and several out of bounds played, followed by jump ball situations, etc. as long as the opponent does not clearly get possession of the ball, it's still the same team possession...

player possession - anytime a player has the ball and then either shoots, passes (a pass that has an opportunity to be an assist), gets fouled, or turns the ball over....

[Mike G]

[admin]

My conversion is complete, in that I'm now thinking of ways to point out that the possession definition is better than the play one. Congrats to Dean and company.

I still do think that both definitions have their merits, but the important thing to me is we've had this discussion before, we've voted on it, and the play defintion supporters were rather badly outvoted. I don't see that changing any time in the near future.

If we as a group are going to make sense to the world at large, we have to be speaking the same language. One new language is difficult enough to teach.