On Sunday I noted that there was a 0.69 correlation between a rookie’s productivity and that player’s per-minute productivity the rest of his career. This was an exceedingly crude study. The only explanatory variable was rookie performance. Players who played more than 10 years were lumped in with players who played only two. In sum, this was not work that could be published in a real journal.
Fortunately, The Wages of Wins Journal is not a real journal. And despite the study’s obvious shortcomings, it does tell us something. NBA players, relative to baseball and football players, are really consistent.
Today I want to use the link between rookie and career performance to look back on the 2006 NBA draft. Twenty players drafted in 2006 went on to play at least 800 minutes in 2006-07. The performance of these twenty players this past season is reported in the following table.
Table One: Twenty Rookies from 2006-07
The top rookie – in terms of WP48 (Wins Produced per 48 minutes) – was Renaldo Balkman. Balkman’s minutes were limited by his head coach (Isiah Thomas), so he only finished third in Wins Produced. Per-minute, though, he was the only rookie to post a WP48 that exceeded 0.200.
This led me to wonder, how often has a player posted a WP48 in excess of 0.200 and failed to be above average in his career?
Our sample of 308 players consists of all players drafted between 1991 and 2005 who played at least 800 minutes their rookie season. Of these 308 players, only 23 managed to post a WP48 in excess of 0.200 their rookie campaign. Of these 23 players, thirteen of these players managed to post a WP48 in excess of 0.200 after their first season. Another nine players managed to exceed the average mark of 0.100. And two players – David West and Antawn Jamison – have posted a WP48 after their first campaign that is below average.
Put it all together, and we see that 91% of all players who posted a 0.200 mark their rookie season went on to be above average. Again, it’s a crude study. Still, if Thomas ever chooses to play Balkman on a regular basis, he should get an above average level of production.
After Balkman we see seven rookies in 2006-07 who were above average. From 1991 to 2005 there were 64 rookies who posted a rookie WP48 that exceeded 0.100. Of these, 15 went on to post a mark in excess of 0.200. Another 33 managed to remain above average. So 48, or 75% of the above average rookies, remained above average. Of the 16 who were below average, 13 players managed to post a positive WP48 (yet below 0.100).
Okay, what if you weren’t above average your rookie season? This past year, seven rookies managed to have a WP48 that exceeded zero, but still below the 0.100 mark. Another five rookies were in the negative range, a group that included Adam Morrison, Andrea Bargnani, and Rudy Gay.
Since 1991 there have been 140 rookies who posted a WP48 between zero and 0.100. And another 81 were in the negative range. Of the 140 who were below average but positive, 53 – or 38% — went on to be above average performers. The negative performers, though, were not quite as likely to become good players. Only nine of the 81 negative performers, or 11%, went on to become above average players.
What does this mean for Morrison, Bargnani, and Gay? Our examination of the past 16 years suggests that it’s unlikely that any of these players will be good. And this is especially true for Morrison. His WP48 his rookie season was below -0.100. Since 1991 there have been thirteen rookies who were this bad (and none played as many minutes as Morrison). Of these thirteen, only three managed a WP48 after their rookie season that was above average. And none of these were close to 0.100. In sum, for Morrison to become a “good” NBA player would be truly exceptional.
Of course fans of these players – and probably their current employers – will insist that each player is the exception to the rule. And of course, that’s exactly what many of the employers of the below average rookies in the past probably thought. The data, unfortunately, tells us that more often than not, the exceptions are indeed exceptions. The general rule is that what we see from rookies is a good indicator of what we will see in the future.
Oh, by the way, I did consider a sample of rookies from 1992 to 2000 and I still found a correlation of 0.67. What I still need to do is consider other explanatory variables. People have suggested position played and the age of the player when he was drafted. Any other ideas? Comments are always welcomed.
– DJ
Arthur
July 31, 2007
I think it would be interesting to find out if some organizations (Phoenix, Utah, Houston) are better at developing talent that others. Much as in the PhD world where having the right adviser helps in guiding the academic career (but the heaving lifting is still on the candidate) I think some places are better at nurturing as opposed to the harsh (but passionate) cities like New York + Philly. I have no idea how you could measure the average annual gain/loss of draft picks that deals with trades. There could be a survival bias on the score.
Matt
July 31, 2007
I take your point that the analysis is very
rough. If I were to control for something that
seems like it might be significant, I would
focus on age and college experience. The logic
is that a bad rookie performance is going to tell
you less about an 18 year old than a 22 year
old. My recollection, for instance, is that
Kobe was pretty shaky his first season.
This of course looks really bad for Morrison, who
had, what, 3-4 years of college. I expect it
would make, e.g., Bargnani and Tyrus Thomas look
better, since they’re younger and had less
experience prior to the NBA.
JB
July 31, 2007
I realize this blog is entirely quantitative, but a qualitative explanatory variable is the specific skill set of the rookie involved. In a case like Morrison, his one NBA skill (shooting) will never make him a star since he doesn’t seem capable of getting his shot off against NBA defences. Balkman’s skill set (rebounding and defence) is NBA-ready, which is why he shines. In Bargnani’s case, his skill set is varied, but raw. Both he and the Raptors recognized that his overall potential as a multi-skilled player is considerable, so during the season they used him in various unfamiliar roles (posting up, rebounding) for which he does not yet have NBA-ready skills. This focus on overall development is why, quantitatively, he performed poorly (just as Balkman would have performed poorly if the Knicks had decided to use him as a scorer). I think you may find Bargnani’s development going forward will follow a similar pattern to Dirk Nowitzki’s career and may likely have a similar ceiling.
Chris
July 31, 2007
Great analysis. I eat this stuff up. Also, being from Portland, I love to see any ranking where Adam Morrison is at the bottom. We had a huge “draft the ‘stache” campaign from the fans over the summer preceding that draft. So glad that didn’t happen…
Scott
July 31, 2007
I like the study, I think adjusting for age would make a huge difference in the outcomes.
It would also be interesting to see how this correlation affects international players as opposed to those who went to college.
noch
July 31, 2007
Did Morrison project to be as bad as he was when using his college numbers?
mike
July 31, 2007
Interesting. Balkman’s stellar numbers rais a couple of questions…
1) Does minutes played change the picture any? The more minutes a rookie plays, the frequency of mistakes I think increases. Basically the more chances opponents have to see a new player, the better they are going to adjust to them. Roy for example asked to do much more over an entire game then Balkman. A player like Roy was also more likely to hit the “rookie wall”.
2) Does shot attempts affect the numbers any? With WP48, its as important to not to take and miss a lot of shots. Scoring is an area that instinctual I would say is going to improve more then other factors. If Morrison for insance took fewer shots, and improved his shooting percentage, youd expect to see a major improvement.
C
July 31, 2007
As a Diehard Bobcat fan that watched Morrison
all year I should say that his struggle had
nothing to do with “not being able to get his
shot off,” Morrison’s trouble is that he was
completely and totally psyched out, not to
say that this is any more curable than any
other issue. For all I know, he could continue
to be “psyched out” indefinitely.
Owen
July 31, 2007
Noch – Adam Morrison was ranked tenth in Pawsmin in his class. See Looking at the NBA Draft part one.
It would be an interesting calculation to see how many wins the Bobcats would have had with an average player in there rather than Morrison.
If they don’t play him next year Charlotte could be surprisingly good.
Mike
August 1, 2007
Will you please stop writing that NBA players are more consistent than football or baseball players? For a guy whose blog is about quantitative analysis, it displays a striking lack of understanding of statistics. Baseball and football players only look inconsistent to you because you’re looking at stats that are largely related to luck, whereas in basketball you’re looking at stats that are related largely to skill and team functioning.
Andrew
August 1, 2007
I saw your comment at Truehoop about the point shaving thesis. Don’t you think it was gratuitous and unnecessary to note Roger Noll probably assisted with the paper? Who doesn’t receive assistance from other individuals at their university? And frankly after reading the paper it strikes me as a case of you not being familiar with the kind of excellent, interesting and rigorous work done by undergraduates at top universities.
Tball
August 1, 2007
Age probably gets to this to some extent, but pre-NBA experience would be interesting to note. European v. US v. Other foreign play. High school v. 1-2 years of college v. 3-5 years of college. Also, if college, success of the college (NCAA champs, regular sweet sixteeners, never danced). You would assume the guys who have done the most, the longest at the college level (Duncan, Battier) would have learned most of what they were able to learn before their rookie year, whereas a player like Yao or Bargnani needs time to learn and develop.
Looking at the records of the teams that play the rookies could also be telling. Winning teams are not going to let marginally performing rookies get 800 minutes, so maybe 800 minutes on Dallas is more useful than 800 minutes on Memphis (the latter being counter productive to future development). Baseball often talks about ‘rushing players’. Could rushing a basketball player cause them to learn too many bad habits and stunt growth?
Mr. Parker
August 1, 2007
Basketball players are what they are. You don’t
suddenly see increased rebounds, steals, assists
out of players who never exhibited the ability
to compile these stats. What you do see are
players who can drastically improve their
“game” by increases or decreases their shots
taken or get better at shooting.
Adam Morrison can only get better if he
starts taking better shots, and he will
only be useful if he starts taking better
so many better shots(at say a 60% success rate)
that it covers for his inability to rebound,
steal, or pass.
Paul
August 1, 2007
I don’t know if somebody suggested the same, but maybe you could track the trades, because maybe Morrisson had a chance of developing after being traded to, let’s say, any team but back to Charlotte.
Jacob Rosen
August 2, 2007
In response to all of the different reports about Morrison’s skill set, I adventured to look at my numbers regarding Rating per 48 minutes, a system similar to Wins Produced, but set around the rating system used by ESPN for the best games played of the year. In this the average for a player around Morrison’s height (6-7 to 6-9) last season was 29.829. Morrison was 24.033. The main reason it was this low, was his extremely poor rebounding (4.7 compared to the average of 9.0 per 48 minutes), and not being able to counter that with that many additional assists (3.4 to 2.6.) Sure his two-point shooting was horrid (39.1% to 47.8%,) but I think the more staggering difference came on the defensive end.
Morrison was one of the worst defensive players in the league last season. He scored a fair amount per 48 minutes, but as seen at the numbers at 82Games.com, his team did not perform well when he was on the court. I accredit this to his lack of blocks (0.12 to 0.93) and steals (0.58 to 1.36,) because we lack any more definite defensive statistics. One last thing that Morrison did poorly in his rookie year, was take way too many shots. He shot on average, about 19.44 field goals per 48 minutes, while the average around his height was 14.60.
George K
August 5, 2007
This post is both subtle and profound.
Oren
August 16, 2007
1) It would be interesting to see whether these numbers are higher or lower based on whether a rookie is the main option. For the first half of the season, Gay was one of the best players on his team with Gasol out. I wonder if this had an effect on his numbers.