Re: Gary Payton's D
- --- In APBR_analysis@y..., "Dean Oliver" <deano@t...> wrote:
>Good question, Dean; I have long thought that this defensive award
> Payton won election to the first team All-D for the 8th straight
> year. Does he deserve it?
is a mixed bag of bones thrown to players who seem to
deserve "something" for their troubles; whether laboring in the
shadows for a champion, or for carrying a poor team to respectability.
Does Doug Steele's (or your) defensive index make the assumption
that a player generally guards his counterpart at the same position?
That is my impression, and I am sure you agree this is a rough
approximation at best. Players go in and out at different times,
many guys play more than one position, etc.
There are certainly many cases of good defensive centers guarding
power forwards who are scoring threats. Point guards who are good
defenders often check the opponent's shooting guard, etc.
Has anyone got an approximation of the pct. of time a player
actually is guarding the same position on the other team? Not sure
if such an index would help anything, but it might lead to this
number having a estimable degree of credibility.
I am still looking for wide availability of individual +/- (is
that the accepted term?) ratings. I have no doubt than we can
effectively measure offense, optimistically; and that we can factor
out the total team +/- to reach an individual's worth; and finally,
that we can factor the player's offense out of his total worth, and
that what remains could be called his defensive value.
I propose that our mission in here be to uncover a source of the
holy +/- grail. I have read objections to its significance, but I
haven't found one that holds water. If you can point me in the right
direction, I will get right on it.
> One of my points here also is that the defensive awards are allhard
> pretty poorly determined. Payton shouldn't have made the team, I'm
> pretty convinced. It takes usually a year or two for voters to
> figure out who really is good and bad defensively. It's simply
> to do.
> Dean Oliver
> Journal of Basketball Studies
- --- In APBR_analysis@y..., msg_53@h... wrote:
> --- In APBR_analysis@y..., "Dean Oliver" <deano@t...> wrote:position?
> Does Doug Steele's (or your) defensive index make the assumption
> that a player generally guards his counterpart at the same
Doug's method looks at matchups, yes. Mine doesn't. The methods
have different strengths and weaknesses. That's why I look at them
A key thing I've observed (and recorded) is that there is a lot of
help in defense. Big men are involved in defense a lot more. Every
drive to the middle brings a big man to help whereas not every post
move brings a perimeter guy to help. Big men then generally are keys
to defense, both positively and negatively. This also means that,
yes, players don't guard the same person all the time -- both because
they go in and out at different times and because of switching/help.
There is also the problem of fast breaks, where no one is guarding
> Has anyone got an approximation of the pct. of time a playerI have a few game scores where I've tracked how often different
> actually is guarding the same position on the other team? Not sure
> if such an index would help anything, but it might lead to this
> number having a estimable degree of credibility.
players were involved in stopping the opponent and in allowing a
> I am still looking for wide availability of individual +/- (isThis may be a way to start getting at defense. On its own, I am one
> that the accepted term?) ratings. I have no doubt than we can
> effectively measure offense, optimistically; and that we can factor
> out the total team +/- to reach an individual's worth; and finally,
> that we can factor the player's offense out of his total worth, and
> that what remains could be called his defensive value.
of those people who doesn't see too much value in the +/- #. But
perhaps it can be used to get at defense. There is still strong
correlation problems (always playing with the same people) and the
fact that it often reflects substitution pattern.
The +/- number is often in Harvey Pollack's stat guide, which Bob
Chaikin has for several years.
Journal of Basketball Studies
i agree with dean, the +/- "ratio" (as "they" call it) is in my opinion
useless. yet a number of nba teams use it (IBM "data mining" techniques
extract most of it). the bottom line is what good is a part time player with
a very positive +/- on a team where most of the starters have negative +/-
(or vice versa). the bottom line is who plays at the same time with whom, and
as far as i know that info is not closely tracked by nba teams. obviously the
teams with high W-L % will have starters with a high +/- ratio and those
teams with a low W-L % will have a low +/-...
in hockey its one thing because players tend to play with specific "lines",
but not so in the nba...
i do have pollack's guides from 88-89 to 96-97, some of which have the +/-,
and i've looked closely at the listings of +/- without being able to discern
any appreciable pattern...
as for how often players guard their "counterparts" the answer is well over
50% of the time (meaning in seconds and minutes), and my guess is 80% of the
time if not more (i used to watch hundreds of nba games on tape over a decade
ago). however, the problem is that those defensive statistics that the league
keeps track of (steals, blocks, charging, fouls) occur much more often when
an offensive player loses his man and is then guarded by one of the other 4
opponents. players who block 200-300 or more shots a year block all 5
opponents shots, not just the man he is directly guarding, many many fouls
occur on switches, and i'd guess most steals are not by the man directly
guarding the player with the ball but made on pass attempts...