[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Celtics' Stuff ] Do Point Guards Develop Differently?
Excellent, excellent analysis and research. This guy obviously doesn't have a
job and has too much time on his hands. It is a shame because he looks like a
talented individual who could make some money in the real world. I enjoyed
the article....
DJessen33
<< From HOOPSWORLD.com
Do Point Guards Develop Differently?
By Kevin Pelton
Feb 27, 2004, 00:30
Two seasons ago, the Spurs fan Web site SpursReport.com had a regular
feature comparing the statistics of San Antonio rookie point guard Tony Parker with
the rookie statistics of several of the league's top point guards, players
like John Stockton and Gary Payton. Surprisingly, the site found that Parker's
numbers compared quite favorably indeed with these legends. (I later threw
together a similar list of the mediocre rookie statistics of some All-Star point
guards for this Supersonics.com article cautioning fans against expecting too
much too soon from the Sonics' Luke Ridnour.)
Or was it surprising?
Point guard, after all, is reputed to be the toughest position for young
players to learn in the NBA. At the same time, while several All-Star point
guards improved dramatically over their rookie numbers, many players who went on to
do little of anything had similar rookie statistics. So maybe, my thinking
has recently gone, rookie statistics just don't tell us very much about what a
point guard is going to end up becoming.
Alas, conventional wisdom tends not to hold much sway in "Page 23", so I
decided to study the issue of point guard development in more depth.
The first question to be asked is, "Do rookie point guards really struggle
relative to other rookies?"
This is not necessarily an easy question to answer. My primary method was to
look at how many point guards were amongst the top ten rookies each season
dating back to 1995-96 by my VORP (Value Over Replacement Player) rating. I
compiled these lists for a column last season comparing last year's rookie class
to its predecessors.
By far, point guard is the least-represented position amongst this group,
with just 11 point guards in eight seasons (the expectation would be 16). The
next lowest position was centers, with 15 players.
However, this could represent a bias in my rating system. Veteran point
guards have, in my opinion, been rated fairly, but I'm a biased observer. For a
control, I went to the All-Rookie teams over that same span -- which also
feature ten players, chosen without regard to position. 17 point guards have made
either first- or second-team All-Rookie, slightly above the expectation and tied
for third with small forwards. Some of these selections were dubious (unless
you're a big Tyus Edney or J.R. Bremer fan), but that's a point against
conventional wisdom.
The evidence, however, is pretty overwhelming when we look ahead of time at
the next portion of my study and evaluate the relative per-minute performance
(again, by my formula) of rookie point guards versus all rookies. Limiting the
sample to players who played at least 500 minutes as rookies and played in
the NBA at least one other season (the second part becomes important later),
rookie PGs have an average efficiency of .451, as compared to .466 for all
rookies. By my formula, that's about the difference this season between David
Wesley, and his backup, Steve Smith.
(As an aside, I should mention that starting with that study, I'm not using
players "listed" as point guards, as my relevant spreadsheet has no such
designation. Instead, I defined point guards as players 6-2 or under and those with
ratings of 2.50 or greater in my passing rating. Though there were some
mistakes, the vast majority of players were correctly placed and I'm not sure this
method is any worse than guessing which position a player "really" played,
often a questionable process. 80 of the 395 players who meet the initial
qualifications were deemed point guards.)
Despite the All-Rookie voting -- which occasionally is tough to justify
statistically -- I think it is safe to conclude that point guards struggle more
than other rookies, which allows us to move on to asking, "Do point guards
develop more than other rookies?"
Let us return, then, to the sample of 395 players who played at least 500
minutes as rookies, were rookies in the 1989-90 season or more recently, and
played at least one other NBA season. Here, their rookie and sophomore seasons
are compared (again using the per-minute efficiency rating):
Group Year1 Year2
Overall .466 .472
PGs .451 .464The point guards, by this method, improve more than
twice as much as all rookies (despite the fact that they are included in this
measure). While their lower initial efficiency explains some of this -- my studies
have confirmed the conventional wisdom that regression to the mean causes bad
players to improve more than better ones -- this is strong evidence for point
guard development.
Another way to look at the issue is to see what percentage of players
improved from their first season to their second season. This reduces the impact any
one player can have. 66.3% of the point guards improved, as compared to 57.7%
of all rookies.
In addition to looking at the development from year one to year two, I also
wanted to take a longer-term look. I chose to use season five, as early
entrants should be developed by this point and most players should be entering their
prime. Let's repeat the calculations for players with at least five years of
NBA experience:
Group Year1 Year5
Overall .473 .468
PGs .455 .463Point guards don't improve very much -- in fact, the 53
players are, on average, worse than the second-year point guards. However, the
250 total players actually decline on a per-minute basis from their rookie
seasons to their fifth years. In part, this is because of anomalies like
Gheorghe Muresan, who posted an efficiency of zero, his excuse being that he only
played one minute during his fifth season. At the same time, it also reflects the
point I made last week when discussing this year's sophomore class -- that
rookies just don't improve as much on a per-minute basis as we tend to believe.
There's another piece of information here, I think. You'll notice a larger
difference between the rookie performance of all players who last five years
and the total population than there is for point guards. I think this may
indicate that rookie performance is a better indicator of long-term success for
players other than point guards.
To negate the impact of the Muresan-esque flukes, let's look at the
percentage of players who improved. Despite the average decline, 51.8% of all rookies
improved by their fifth years. But here the point guards dominate -- a full
64.2% of them were better in year five than in their rookie seasons.
Overall, I feel confident in concluding that point guards develop more than
rookies at other positions. There is another part of this question, however,
and that is whether rookie performance indeed means less than point guards than
for other rookies.
To answer this aspect of the question, I decided to look at the variability
of the relationship between the player's rookie performance and that in years
two and five. I did that in two ways, by calculating the correlations and the
average squared errors. Here is that table:
Year 2 Year 5
Group r MSE r MSE
Overall .592 1477.5 .320 4665.1
PGs .591 1244.5 .150 6332.3What do these numbers mean?
Correlation(r) measures the strength of the relationship between two sets of numbers; a
high correlation means that the two sets are closely related, a small correlation
means little or no relationship at all. As would be expected, the
correlations with second-year performance are reasonably strong; the year-five
correlations not very strong at all. The relevant information is that the correlation in
year five for point guards is less than half that of all players. In other
words, knowledge of how a point guard performed as a rookie doesn't tell us very
much about what he's going to be like four years later.
Mean squared error is another way of looking at the issue that provides a
harsher penalty for big changes by squaring the difference between year two or
year five efficiency and that in year one. (Technical note: I multiplied the
efficiencies by 1000 before squaring to get more manageable numbers.) As a
result, I could have gotten much more accurate results by removing guys who played
only a handful of minutes in year two or year five, but the big issue is point
guards relative to overall positions, not the size of the numbers (which have
little power of language anyways).
Surprisingly, point guards actually show as less variable in year two. By
year five, however, the same difference as evident as with correlation -- the
average squared error is nearly 50% larger than for all players, a huge
difference.
The conclusion, in this case as throughout this column, is that conventional
wisdom is correct. Point guards are worse as rookies, they improve more, and
their improvement is less predictable than at other positions. I've argued in
the past that NBA observers are too slow in updating their evaluations of
players to reflect their performance early in their NBA careers, but that may not
be the case with point guards. We should be slower to come to judgments on
point guards, and give them more of an opportunity to prove themselves.
"The Page 23 Club"
One of the unfortunate things about this column is that, because of my
schedule, I can't commit myself to a specific day or time for publishing columns.
To help my readers, I'm starting an e-mail list. If you want, you'll receive an
e-mail whenever a new column is up with an introduction to the column and a
link. If you're interested, e-mail me at kpelton@xxxxxxxxxxxxxx and let me
know. I will, of course, make every effort to protect the privacy of your e-mail
address. >>