Sunteți pe pagina 1din 7

Individual Offensive and Defensive Rating:

A Guideline for Evaluating and Improving Player Talent


Steven Jacobson
In the NBA today, many assessments of player performance is governed by
arbitrary metrics, flashy stats, and public perception. A defender who plays
good perimeter defense and forces the shooter to take and miss an
inefficient shot does not show up on the stat sheet with a block or a steal.
However, he was the catalyst for a result that was nearly as good as either of
those stats. A point guard who puts up 25 points per game would almost
certainly receive top dollar, but if he cant create opportunities away from
the ball and make room for his teammates to contribute, is he really as
valuable as the 25 PPG suggests? Individual offensive and defensive ratings
help to alleviate this problem, measuring points produced as a direct result
of the player being on the court, per 100 possessions, and number of points
that were scored against a players team, as a direct result of the player
being on the court, per 100 possessions. Evaluating players by this metric, in
combination with position specific statistics and player salary is necessary to
more accurately evaluate player talent.
First to represent how each of these metrics measure performance, it is
necessary to see how offensive (and defensive) ratings correlate with a

teams wins (and losses).

The correlation coefficients for the above graphs top left, bottom left, top
right, bottom right are .82, .27, .70, .62, respectively. The difference between
correlation coefficients for the defensive ratings is significantly smaller than
those for the offensive ratings. This difference can be attributed to the fact
that very often good defense is described by a lack of statistics (forcing a
change of possession as in the example above), and further, by total team
success on defense. Their small difference in correlation to losing (higher
defensive rating means more points are scored against that defender and is
less favorable) shows that when one defender has a low defensive rating, it
may be likely that other defenders on the court also have low defensive
ratings, which will, of course, have positive results for the team. On the other
hand, team offensive rating has a high correlation with winning while
individual offensive rating only has a small correlation to winning. This makes
sense because a player can put up great stats, but if the rest of his team
isnt doing much, they still may lose. However, there still is a positive
correlation, suggesting that individual offensive rating does affect winning. To

more tangibly represent how offensive rating affects winning, four of the top
five teams (GSW, ATL, LAC, CLE) in the 2014-2015 regular season standings
also are among the top six teams in offensive rating, while the bottom five
teams (MIN, NYK, PHI, LAL, ORL) in the same seasons standings are among
the bottom seven in offensive rating.
The top 20 individually rated offensive players and top 20 individually rated
defensive players are shown below. A minimum qualification of 58 games
played (NBA minimum for most standard statistics) was necessary to be
included in this calculation. The rest of the ratings can be seen here.
Name
Brandan
Wright
Tyson
Chandler
Tyler
Hansbrough
Chris Paul
Stephen Curry
DeAndre
Jordan
Anthony
Morrow

Tea
m
PHX
DAL
TOR
LAC
GS
W
LAC
OKC

Ed Davis

LAL

Anthony Davis

NOP

Jimmy Butler

CHI

Rudy Gobert
Patrick
Patterson

UTA
TOR
HO

Ind. Off.
Rating
133.40677
26
132.69833
27
131.60102
87
129.48346
33
124.92474
88
124.72239
87
123.21332
26
123.10381
17
122.57699
68
122.43701
85
121.56607
18
121.38684
40
121.19622

Team Off.
Rating
105.292616
5
109.454624
7
111.030219
0
112.403900
4
111.603607
3
112.403900
4
107.766782
3
103.428708
6
108.201314
9
107.544086
8
105.111443
9
111.030219
0
106.963291

Name

Tea
m

Kawhi Leonard

SAS

Tim Duncan

SAS

Andrew Bogut
Draymond
Green

GSW

Danny Green

SAS

Tony Allen

MEM

Ian Mahinmi

IND

Kosta Koufos

MEM

Justin Holiday

GSW

Nick Calathes

MEM

Stephen Curry

GSW

Manu Ginobili
Bismack

SAS

GSW

Ind. Def.
Rating
108.458836
7
108.568211
7
108.743104
7
108.827883
1
109.278545
6
109.398562
3
109.478877
4
109.554811
2
109.626283
3
109.626327
6
109.627767
3
109.662551
4
109.663240

Team Def.
Rating
102.0052512
102.0052512
101.3542965
101.3542965
102.0052512
102.2217926
103.2400066
102.2217926
101.3542965
102.2217926
101.3542965
102.0052512

Both individual offensive and defensive ratings can vary for each position in
the league. For example, a centers offensive rating is most heavily reliant on
his ability to grab offensive rebounds, while a shooting guards offensive
rating is most greatly dependent on successful
3 point shots. These
variations are shown in the table below, which states the correlation
coefficient between each listed statistic and offensive and defensive rating
for each position (note that there are negative correlations for the last three
statistics because the lower the defensive rating, the better the defender).

Correlation coefficient values for specific statistics and offensive rating and defensive
rating for each position

Point guards should be able to not only score, but assist, shooting guards
should be able to shoot the three, small forwards need to be versatile, and
power forwards and centers should be able to draw fouls and rebound. This
isnt necessarily news to anyone evaluating talent (although it does reinforce
prior knowledge). However, being able to see how each stat impacts a
players efficiency for his position relative to other stats is significant. For
example, shooting guards shouldnt bother with shooting a two point shot
unless it is a highly probable shot, and should instead focus on shooting
threes. It is also interesting that turnovers have a positive correlation,
although not very large, with point guard offensive rating. This may suggest
that turnovers, while of course not good for a team, may be a predictor that
a good point guard is trying to spread the ball around more and make good
plays that may result in not only more turnovers, but also more points. This
is, of course, not to say that a point guard should create more turnovers, but
rather that he shouldnt be hesitant to make a play or difficult pass because
he fears turning the ball over. As an example, Chris Paul, who averages .3
more turnovers than the league average at point guard per game, also
boasts the leagues most assists per game at 10.2 and the best assist to
turnover ratio. The above table can also be used in conjunction with statistic
differentials for each player at his respective position to evaluate whether a
players offensive rating is a result of his team, or his individual stats. Player
stat differentials, compared to the league average at his position, can be
found here. For example, using the table, it is clear that a centers offensive
rating is most predicted by his free throws and offensive rebounds. Sorting
by free throw and offensive rebounding differentials, one can see which
centers are above the position average, and subsequently see whether or
not their offensive rating agrees with the differentials. Andre Drummond, for
example, has 3.02 more rebounds than the average center. He also shoots
4

1.91 more free throws and makes .03 more free throws than the average
center and does not have any other stats that are significantly below the
average for centers. Yet, his offensive rating is 1.81 less than the average
center. This may mean that Drummond is in fact being dragged down by his
team, or at least by a system that does not utilize his full potential. A team
with a system that values big men then may want to look into acquiring
Drummond, possibly even at a price that is significantly below his true value.
Often when teams are rebuilding, their first question is what player they
should acquire (or draft) to build a team around. The first step in answering
this question is to decide what position to start building around. The
following table displays the correlation coefficient between individual
offensive rating and wins at each position.

Correlation coefficient values for individual offensive rating and wins at each position.

The above table tells us, for each position, how good of a predictor individual
offensive rating is for wins. It is reasonable to assume that any player a team
would want to build around would have an individual offensive rating that is
relatively very good. Thus, according to these values, a team would be most
wise to build around a small forward or point guard because their ratings
correlate most significantly with team wins.
While players that a team would want to build around most likely cost top
dollar (unless they are drafted), it is also necessary to add other players to
the roster that wont cost as much. The plots below show that there is
currently no strong relationship between individual offensive rating and
salary (r=.23) and individual defensive rating and salary (r=-.06)

This relationship suggests that there may be an inaccurate method currently


being used to evaluate talent and establish salaries. Often, teams pay a high
salary for flashy players who can score, but may not be focusing enough on
players who play solid defense or who contribute in ways other than scoring
for the offense. Teams may also feel public pressure to re-sign a player or let
a player go. Even more, players tend to get paid more later in their careers,
especially if they used to be an integral part of the team or have played for
the team for a long time (or both). While he did not play the minimum
number of games to be included in this study (58 games), Kobe Bryant
serves as a great example here. He will be paid $25 million dollars in the
2015-2016 season, but is nowhere near as productive as he once was.
In order to see who the most offensively and defensively efficient and
productive players are in the NBA while also considering their salary, it is
important to come up with a metric to grade each player per dollar spent in
order to assist general managers in getting the most for their money.
However, because salaries vary by a significant amount more than player
ratings, it would make sense to compare players of similar salaries (If all
players were compared by this metric at once, these results would state that
Luke Babbitt is more valuable than LeBron James). Players were rated by
their individual offensive rating divided by their salary, resulting in points
produced by a player, per 100 possessions, per dollar. Also, note that
because a lower defensive rating is better, it did not make sense to divide by
the salary. As a result, the individual defensive rating and salary were
multiplied. While the actual resulting number does not have a practical
application by itself, the values serve to compare players relative to each
other on defense (thus, a smaller number is still the most favorable). The
tables can be seen here.

Evaluating player talent using individual offensive and defensive ratings is a


powerful tool for any GM looking to analyze player performance more
holistically than just looking at flashy stats such as PPG. While an
organizations opinions on players are commonly influenced by flashy stats
and public perception, looking at each players individual ratings in
conjunction with the stats that are important to each position adds
significant insight into why a player is performing the way he is and helps to
more accurately define each players contributions to his team. Even more,
when used comparatively with salary, this tool can find hidden talent and
expose talent that is being over-compensated. Ultimately, this is a tool that
can help to find the most efficient players, highest potential players, and
most under-valued players at each position.
Notes:
All data was obtained from stats.nba.com and basketballreference.com
Equations used to calculate individual offensive and defensive ratings
are from http://www.basketball-reference.com/about/ratings.html, but
originated from Dean Olivers Basketball on Paper.

S-ar putea să vă placă și