## Jerry's 57: Example net pts

Expand Messages
• Just because Jerry Stackhouse had a big game, I figured I d look at my former Tar Heel mate to show the difference between the Net Point calculations I make.
Message 1 of 3 , Apr 3 10:05 PM
Just because Jerry Stackhouse had a big game, I figured I'd look at my
former Tar Heel mate to show the difference between the Net Point
calculations I make.

Version 1 is

Points Produced - Poss*DRtg/100

Version 2 is

Points Produced - MIN*TmPoss/TMMIN*DRtg/100

Basically, I can calculate the number of points produced by a player, but
then what is the number of points allowed??? In Version 1, a player is
assumed to allow points only on possessions he uses at an efficiency given
by his defensive rating. In Version 2, a player is assumed to allow
points at the same efficiency (his defensive rating), but the number of
possessions he faces is now estimated as the average number of possessions
during the minutes he plays.

What is better? Neither. There is no real physical meaning to either
because no one player is a team (even Jerry). In this game, I estimate
Jerry to be either +16 or +33 -- a big difference (though either number is
awesome). I have no way to reality-check the numbers other than making
sure that the team totals add up -- and they do. Detroit won by 27 and

But that is the difference between the 2 NetPt calcs I presented for the
season for "Top Players in 2001".

Dean Oliver
www.tsoft.com/~deano/index.html
deano@...
• Deano: do rebounds count at all in your points produced (not ratings )lists ? The reason I ask is, I tried like heck to reconcile your lists with mine, and
Message 2 of 3 , Apr 11 5:03 PM
Deano: do rebounds count at all in your "points produced"
(not "ratings")lists ?
The reason I ask is, I tried like heck to reconcile your lists with
mine, and the big obstacle seems to be that I rank rebounders higher
than you do. Rebounds do not produce points directly, but they yield
possessions, which may lead to points.
Pardon my ignorance.
• ... with ... higher ... yield ... No problem. I ve only partially explained what the methods are and, given my current 70 hr/week schedule, that s all it s
Message 3 of 3 , Apr 11 6:53 PM
--- In APBR_analysis@y..., msg_53@h... wrote:
> Deano: do rebounds count at all in your "points produced"
> (not "ratings")lists ?
> The reason I ask is, I tried like heck to reconcile your lists
with
> mine, and the big obstacle seems to be that I rank rebounders
higher
> than you do. Rebounds do not produce points directly, but they
yield
> possessions, which may lead to points.
> Pardon my ignorance.

No problem. I've only partially explained what the methods are and,
given my current 70 hr/week schedule, that's all it's gonna be for a
while.

Rebounds definitely go into points produced, but only offensive
rebounds. Defensive rebounds go into the defensive rating.
Offensive rebounds are credited for what they are worth on a team.
On a team that shoots horribly, they are worth more.

Basically, I calculate the team's floor percentage with offensive
rebounds and subtract off their floor percentage without them. The
weight I generally assign to ORs is

ScPoss/(Poss + OR)

where

ScPoss = FG + (1-(1-FT%)^2)*0.4*FTA
Poss = FGA - OR + 0.4*FTA + TO

This is only partially accurate. I have noticed that ORs tend to
increase the likelihood of scoring, whereas this assumes that they do
not.

You then also have to take away credit from guys who have benefitted
from the ORs. That's a bit more complicated.

Generally, though, I agree that a big difference between what I do
and what others do is in rebounds. Other people effectively weight
them more heavily than I do. Mine actually have a pretty big weight
defensively, but few people have had much credit coming from
offensive rebounds. (This is quite different from HS ball, where
offensive rebounds end up much more important by my theory/formula.)

I would, however, submit that defensive rebounds really don't produce
points. Nor do steals or blocks. They end an opponent's possession.
(They can create points, but this is almost negligible
statistically.) They are defensive stats. Including them in the
offense makes what you're calculating not measurable or confirmable.
That's why I keep them separate. It still makes for an approximate
value technique, which is what you do.

Does that help?

Dean Oliver