APBRmetrics

This is an introductory level question--what is the best way to calculate usage for a player? I used the formula from basketball reference and then the formula from another post on this site and ended up with an 8% difference. According to bbreference the total was 27% and using the other formula it was 35.

thanks

Basketball reference's usage rate is not pace adjusted, while hoopdata's is.

I do not know if that is causing the difference you saw or not though without the citation of your other usage rate source.

I found this formula on thecity2.com--it had been referenced before in this forum

■USG% is the percentage of team possessions (POSS) that a player uses either through shots attempts, turnovers, or getting fouled (FGA+TOV+0.44*FTA). So, it’s USG%=100*(FGA+TOV+0.44*FTA)/POSS. Some also count a fraction of the assists that a player gets (e.g. 0.3*AST) in the numerator.

This is the formula cited on basketball reference

Usage Percentage -- the formula is 100 * ((FGA + 0.44 * FTA + TOV) * (Tm MP / 5)) / (MP * (Tm FGA + 0.44 * Tm FTA + Tm TOV)). Usage percentage is an estimate of the percentage of team plays used by a player while he was on the floor.

**My question is why does the reference formula include minutes played?

100 * ((FGA + 0.44 * FTA + TOV) * (Tm MP / 5)) / (MP * (Tm FGA + 0.44 * Tm FTA + Tm TOV))

can be re-arranged into

100 * ((FGA + 0.44 * FTA + TOV) / (MP) * (Tm MP / 5)) / (Tm FGA + 0.44 * Tm FTA + Tm TOV))=

or 100 * individual possessions used per minute * minutes per team possession=

100 * individual possessions used per minute * a pace factor for that team

The more minutes per team possession for a particular team (the slower the team), the higher the usage rate will be for a player using posessions at a certain per minute possession usage per minute.

If a team uses possessions slower than league average (more minutes per possession) then a player on that team's usage will higher than the usage for a player's with the same individual possessions used per minute but who plays on a team that plays faster than the first team.

Crow wrote:Basketball reference's usage rate is not pace adjusted, while hoopdata's is.

I do not know if that is causing the difference you saw or not though without the citation of your other usage rate source.

Ours is, by definition, pace-adjusted. It's the percentage of team plays used by the player while on the court, so everything is relative to the team's pace.

Minutes are in the formula to estimate the number of plays that took place while the player was in the game. We have to estimate it because we don't know what the exact # is.

Ok, I didn't immediately pick up on your formula having pace included, the way it was organized.

But it, as with other versions, is only estimated pace based on team average and not player specific actual pace as you note.

I personally like Hollinger's formula. It adjusts for assists which I've always thought was important. This is an underrated component in usage rate.

I think I agree.

True, but it moves away from the typical assumption of Usage being a number that represents how many possessions a player ends.

Think of it in these terms:

A 1st degree possession is a possession that a player ends (FTA, FGA, TO)
A 2nd degree possession is one that leads to a first degree possession (Assist, pass that leads to a FTA, or pass that leads to a TO)
A 3rd degree possession is one that leads to a second degree possession (Hockey Assist, etc.)
And so forth, and so on.

Including assists would be adding 1st degree possessions with only partial 2nd degree possessions, and therefore, it's an incomplete representation. Because not all 2nd degree possessions are being accounted for, Hollinger's USG formula is misleading IMO.

Doesn't an offensive foul without the ball (illegal screen, foul going for the OReb,...) also charge a player with a turnover -- and a possession used, even if he has not touched the ball?

At b-r.com, Reggie Evans has a Usg% of 7.5 and a TO% of 30.

sideshowbob wrote:True, but it moves away from the typical assumption of Usage being a number that represents how many possessions a player ends.

Think of it in these terms:

A 1st degree possession is a possession that a player ends (FTA, FGA, TO)
A 2nd degree possession is one that leads to a first degree possession (Assist, pass that leads to a FTA, or pass that leads to a TO)
A 3rd degree possession is one that leads to a second degree possession (Hockey Assist, etc.)
And so forth, and so on.

Including assists would be adding 1st degree possessions with only partial 2nd degree possessions, and therefore, it's an incomplete representation. Because not all 2nd degree possessions are being accounted for, Hollinger's USG formula is misleading IMO.

I hear you on the "pass that leads to FTA", but most forms of usage rate are incomplete.

I think Hollinger's goal is to have the most complete picture possible, therefore assists must be taken into account.

Would somebody be willing to create an excel spreadsheet for me that will clarify this usage issue--something i can just plug my teams stats into and get a clear look?

I think this is right,



but check it and run it and see.

Mike G wrote:Doesn't an offensive foul without the ball (illegal screen, foul going for the OReb,...) also charge a player with a turnover -- and a possession used, ...

I wrote this, and I don't actually know, regarding the highlighted part. After a missed shot, the possession hasn't ended until the DReb is secured.
Except that the shot clock has restarted.
So it doesn't seem that this should be a TO.
And if it's not, some guys seem to get lots of TO, relative to their shots and assists.

Mike G wrote:Doesn't an offensive foul without the ball (illegal screen, foul going for the OReb,...) also charge a player with a turnover -- and a possession used, ...?

I wrote this, and I don't actually know, regarding the highlighted part. After a missed shot, the possession hasn't ended until the DReb is secured.
Except that the shot clock has restarted.
So it doesn't seem that this should be a TO.
And if it's not, some guys seem to get lots of TO, relative to their shots and assists.