APBRmetrics

maybe a silly question but a important one for me. I'm tracking possessions by hand and was giving each the player with the assist and the player who made the shot with a possession. This way the total individual possessions would be more than the total team possession. I've been ignoring offensive rebounds but maybe shouldn't. I do credit the points to the passer/scorer accordingly. Should I do the same with possession used? can each guy use half a possession? I do have Dean Olivers book but just wanted to get a simpler answer. Thanks!!!

PS what formula would you all use for a league that uses FIBA rules. I see that most of you use 0.44 to calculate possessions. using the formula for possessions gives me a big difference for both teams, I want to average it but sometimes its off by 10, does this mean whoever is recording the data is just bad or am I doing something wrong. Thanks again.

The general equation for possessions is field goal attempts + turnovers + .44*free throw attempts - offensive rebounds. The .44 comes from how often a free throw is the last one (due to the combination of and-ones, two shot fouls, and three shot fouls) and thus is the potential end to a possession. So assists are not counted as using a possession. And the multiplier for free throws might be a little different if FIBA has a different proportion of foul types, but I don't have any info on that. Of course, if you're scoring games by hand you can just count the number of possessions directly. The formula is used to estimate possessions from boxscore data after the fact.

I've personally wondered if that's the best way to do it; certain assists are certainly a player's attempt to use the possession (an obvious one would be an alley-oop) even if someone else technically takes the shot. But the simple answer to your question is that if you include assists and shots as possessions, your numbers will not square up with what you find elsewhere. You could try a different system, like splitting a possession between the passer and the shooter, but then your data will not line up with the typical calculation at the player level (although it should at the team level).

For individuals, Ed Kupfer found this linear-weights style possession estimator:

POSS = 0.74 * FGA + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

Points produced could also be estimated:

PtsProd = 1.45 * 2Made + 2.2 * 3Made + FTMade + 0.6 * OR + 0.6 * AST

Btw, it's been 7 years and that whole thread I linked is still a must read.

Neil Paine wrote:For individuals, Ed Kupfer found this linear-weights style possession estimator:

POSS = 0.74 * FGA + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

Points produced could also be estimated:

PtsProd = 1.45 * 2Made + 2.2 * 3Made + FTMade + 0.6 * OR + 0.6 * AST

Btw, it's been 7 years and that whole thread I linked is still a must read.

A made field goal definitely ends a possession, so why isn't it:

POSS = FG + 0.74 * FGM + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

where FG = made field goals, and FGM = missed field goals

Or was Ed taking into account assisted vs. unassisted field goals somehow?

btw, I love this stuff - thanks for posting that link, Neil

EvanZ wrote: A made field goal definitely ends a possession, so why isn't it:

POSS = FG + 0.74 * FGM + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

where FG = made field goals, and FGM = missed field goals

Evan -- I beg you -- PLEASE don't use FGM to mean missed FG.
Some use that to mean FG made.
See if FGX feels right, for FG missed.

Mike G wrote:
Evan -- I beg you -- PLEASE don't use FGM to mean missed FG.
Some use that to mean FG made.
See if FGX feels right, for FG missed.

You don't have to beg. I've used FGX before, too. In fact, if you scroll to the right on my PSAMS spreadsheet, that's how I normally do it, too:

http://thecity2.com/psams-ratings/

Therefore a player can have points produced without using a possession correct? For example if he grabs a offensive rebound and passes to a shooter who makes the shot he will obviously get contributions to the made shot but not use a possession. Is this correct? Because then a player who doesnt take shots but distributes a lot would have a flawed high Individual Offensive Rating. Am I wrong?

EvanZ wrote:

Neil Paine wrote:For individuals, Ed Kupfer found this linear-weights style possession estimator:

POSS = 0.74 * FGA + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

Points produced could also be estimated:

PtsProd = 1.45 * 2Made + 2.2 * 3Made + FTMade + 0.6 * OR + 0.6 * AST

Btw, it's been 7 years and that whole thread I linked is still a must read.

A made field goal definitely ends a possession, so why isn't it:

POSS = FG + 0.74 * FGM + 0.44 * FTA + 0.25 * OR + 0.25 * AST + TO

where FG = made field goals, and FGM = missed field goals

Or was Ed taking into account assisted vs. unassisted field goals somehow?

btw, I love this stuff - thanks for posting that link, Neil

A made FG does not necessarily end a possession. It could be a made FG and a shooting foul and then a FTA. Furthermore, that FTA could be missed, and then the same team could get an offensive rebound.

Gabe, that is a rare event. Surely, you don't think that is accounting for the difference between Ed's formula and the one I suggested. Are you?

EvanZ wrote:Gabe, that is a rare event. Surely, you don't think that is accounting for the difference between Ed's formula and the one I suggested. Are you?

A made FG along with an And 1 FTA is a "rare event"?

gfarkas wrote:

EvanZ wrote:Gabe, that is a rare event. Surely, you don't think that is accounting for the difference between Ed's formula and the one I suggested. Are you?

A made FG along with an And 1 FTA is a "rare event"?

By my estimate, there have been about 200 missed AND1 this season. There have been 18,326 made field goals. Is that not rare?

Edit: I should also add that of those 200 missed shots we should only expect around 10% of them to be rebounded by the offense, if I recall correctly studies that have been done on rebounding off missed FTA. So, that's about 20 made field goals out of 18,000 that did not end a possession for the offensive team.

It's still not clear to me where Ed's equation came from.

EvanZ wrote: It's still not clear to me where Ed's equation came from.

How much of a difference does it make? The and1 thing should also be captured by the 0.44 coefficient for FTA.

If you want it to be more accurate, using the average ORB% on the missed field goals is pretty clearly the better way. We can account for the "and1", if we think it is not accurate represented by 0.44, by using 0.97 on field goals made (FGM, something like 2.5% to 3.x% is my suspected rate of fouls when a field goal is made, Evan's numbers of and1s seems to agree with that assuming league average conversation rate of those free throws).

I don't think that this is so different in FIBA ball, but in the end the equation to estimate possessions can be adjusted by using league average values.

EvanZ wrote:

gfarkas wrote:A made FG along with an And 1 FTA is a "rare event"?

By my estimate, there have been about 200 missed AND1 this season. There have been 18,326 made field goals. Is that not rare?

Edit: I should also add that of those 200 missed shots we should only expect around 10% of them to be rebounded by the offense, if I recall correctly studies that have been done on rebounding off missed FTA. So, that's about 20 made field goals out of 18,000 that did not end a possession for the offensive team.

It's still not clear to me where Ed's equation came from.

You're conflating 2 different things I wrote. The first is the occurrence of a made FG along with an And 1 FTA. The second is that same And 1 FTA being missed and then subsequently rebounded by the offense.

The second idea is, of course, a rare event, both relatively and absolutely.

However, the first is not that rare, in my opinion.

gfarkas wrote: However, the first is not that rare, in my opinion.

If we assume league average FT% for possible And1s, we end up having around 270 possible and1s out of 18,326 made field goals. No idea, but that sounds still rather rare to me.

Evan, another idea, changed ORB% over the years. Over the years the ORB% became lower. Ed's formula has 26% in it, while it was 31% from 1981 to 2005 in average.

gfarkas wrote:

EvanZ wrote:

gfarkas wrote:A made FG along with an And 1 FTA is a "rare event"?

By my estimate, there have been about 200 missed AND1 this season. There have been 18,326 made field goals. Is that not rare?

Edit: I should also add that of those 200 missed shots we should only expect around 10% of them to be rebounded by the offense, if I recall correctly studies that have been done on rebounding off missed FTA. So, that's about 20 made field goals out of 18,000 that did not end a possession for the offensive team.

It's still not clear to me where Ed's equation came from.

You're conflating 2 different things I wrote. The first is the occurrence of a made FG along with an And 1 FTA. The second is that same And 1 FTA being missed and then subsequently rebounded by the offense.

The second idea is, of course, a rare event, both relatively and absolutely.

However, the first is not that rare, in my opinion.

I just re-calculated totals for this season. There have been 17,389 FGA (my total was off before). There have been 641 made And1 attempts (assuming 75% FT%, that equates to roughly 200 missed And1 attempts).

If an AND1 shot is made, the possession is over, so I don't care about those. If it is missed, then we have 200/17,000 to worry about (~1%).

Do you want to keep going on about this?