APBRmetrics

supersub15 · Joined: 21 Sep 2006 Posts: 78

Basketball-Reference uses PSA:
PSA = PTS / (FGA + 0.44*FTA)

ESPN use PPS:
PPS = PTS / FGA

What is the proper name and which formula is a better indicator of scoring efficiency?

Thanks.

Eli W · Joined: 01 Feb 2005 Posts: 285

Points per shot attempt (PSA) is used more frequently by APBR-types. However, the trend has been to present it as true shooting percentage (TS%), which is just PSA/2. Personally, PSA makes a lot more sense to me, but most are using TS% now.

As far as which of PSA and PPS is a better indicator of scoring efficiency, probably PSA. Both PSA and PPS include points scored from free throws in the numerator, so it's preferable to have some measure of free throw attempts in the denominator. But it should be noted that 0.44*FTA, which is meant to approximate times fouled while shooting, is fairly rough on an individual player level.

admin · Posted: Tue Aug 21, 2007 11:42 am Post subject:

DLew · Joined: 13 Nov 2006 Posts: 57

I would argue that Points Per Shot (PPS) is not that bad of an approximation of scoring contribution at the individual level because of the positive externalities getting to the line creates. PPS does not use free throws in the denominator, and therefore rewards them. Drawing a foul is and making two shots is worth more than making a field goal because drawing a foul puts the other team closer to the bonus, and it might force the opposing team to remove one of their better players.

At the team level Kevin has clearly shown the superiority of TS%, but at the individual level it slightly undervalues the ability to draw fouls because of positive externalities associated with free throws. TS% is a better stat than PPS, but not by as much as it might first appear.

Eli W · Joined: 01 Feb 2005 Posts: 285

Thanks for the link to the free throw multiplier thread, I had forgot about that.

As for your HoopsWorld article, I think it's important to remember that different stats have different purposes, which makes it hard to say that one is "better" than another. Is TS% "better" than eFG% because on the team level it correlates better with ORtg? I don't think so. eFG% is just trying to get at something different than TS%. For some purposes, TS% is more useful, for others eFG% is more useful. When it comes to TS% and PPS though, I do agree that for most purposes TS% is more useful.

Statman · Joined: 20 Feb 2005 Posts: 79

DLew · Joined: 13 Nov 2006 Posts: 57

The .44 coefficient is there simply to most accurately estimate possessions because there are times when 2 free throws do not end a possession (3 pt foul, technical, or flagrant). The .44 does not in any way shape or form account for the externalities associated with drawing fouls.

I would never argue that PPS is better than TS%, my main point is that I consider PPS to be better than FG% for individuals in spite of what Kevin showed at the team level, due to the reasons I listed before.

Statman · Joined: 20 Feb 2005 Posts: 79

DLew · Joined: 13 Nov 2006 Posts: 57

It isn't intended to, and it doesn't. I am not saying the externalities are extremely large. However using .44*fta in the denominator is only an accurate valuation of FTs if there is no externality. If FTs have any value beyond the points they generate then this estimator fails to account for it.

Whether or not this externality exists in any sense large enough to matter is up for debate.