/cdn.vox-cdn.com/uploads/chorus_image/image/31573637/20140320_sng_rb1_152.0.jpg)
The recent play of Gorgui Dieng and, to a lesser extent, Shabazz Muhammad have been flickers of hope in an otherwise frustrating season. Both players were deemed unplayable early in the year and posted stats in garbage time that only served to reinforce that assessment. However, over the past couple months, each player has shown definite signs of improvement. This has led to much optimism around these parts that their improved play portends similar, or even better, play in future seasons. The unspoken assumption backing up this hope is that production compiled after a rookie has become acclimated to the rigors of the NBA game is more predictive than rookie production compiled in November and December. After all, there is a normal learning curve in other professions, so it only makes sense that NBA players would improve both during and after their rookie seasons. It is exactly because this assumption seems so obvious, and widely accepted, that I decided to question it.
I have neither the time nor the skill to complete a massive, rigorous study on this question, but I was able to look at players from three consecutive recent drafts in order to gather some preliminary data concerning how common it is for players to dramatically improve during their rookie seasons, and how often that improvement is either a "real" display of the player's skill, or an intermediary point for yet another jump in production. For my sample, I used players from the 2008-2010 NBA drafts, throwing out players with microscopic sample sizes .My criteria were not strictly numerical, but focused on the question of whether the player could, at the time, give his fanbase hope with his improved play and whether or not the player actually had measurable production after his rookie season. For the purposes of this article, I decided not to look at the improvement rate of players who did not improve, or even declined, during their rookie seasons, or the overall improvement rate for players who got much better in subsequent seasons, but those topics could be the subjects of future articles.
Basketball-Reference helpfully provides pre and post All-Star break splits, so I decided to look for players whose net rating (offensive rating minus defensive rating) was at least ten points better after the All-Star break than before the break to identify players who showed a massive leap in productivity late in the year. I am aware of the flaws in using this measure as an overall measure of player quality, as it tends to underrate and overrate certain types of players, but these flaws are lessened when comparing players to themselves. It is also statistically likely that some players will shoot and play better early in the year and some players will shoot and play better later in the year due simply to random effects*. There were twenty players in the 2008-10 drafts who fit these criteria. There are seven players from the 2013 draft that have played at least 250 minutes who also fit said criteria.
*If any statistician would like to measure how many players would be expected to show such a leap in productivity due to random factors, I welcome his or her input...
Player | Net Rating 1st Half | Net Rating 2nd Half |
Cody Zeller | -12 (884 minutes) | +12 (480 minutes) |
Kelly Olynyk | -10 (816 minutes) | +1 (480 minutes) |
Gorgui Dieng | -15 (205 minutes) | +9 (515 minutes) |
Anthony Bennett | -34 (513 minutes) | -10 (134 minutes) |
Archie Goodwin | -29 (387 minutes) | -2 (115 minutes) |
Otto Porter | -36 (214 minutes) | +4 (82 minutes) |
Shabazz Muhammad | -23 (81 minutes) | -12 (209 minutes) |
Of these seven rookies, only Zeller, Olynyk, and Dieng have played significant minutes following the All-Star break, which, to my mind, makes it less likely that their improvement is the result of small sample size variance or flukish shooting. On the other hand, the improvement of their rookies from abject putridity to mediocrity should provide a small glimmer of hope to Wizards, Cavs, and Suns fans.
Or should it? What happened to other players who experienced similar improvements in their rookie seasons? I went through the sample of twenty players to catalog whether or not the player's late season improvement was "real", i.e., whether that improvement carried over to the next couple of seasons. Exactly half of the twenty player sample showed "real" improvement, while the other ten players regressed from the second half of their rookie years. Let's start with those who regressed.
Player (Career Net Rating) | Net Rating 1st Half | Net Rating 2nd Half |
Greg Monroe (+3) | +3 (1393 minutes) | +19 (829 minutes) |
Terrence Williams (-17) | -28 (977 minutes) | -9 (787 minutes) |
Michael Beasley (-9) | -10 (1241 minutes) | 0 (768 minutes) |
Toney Douglas (-3)* | -10 (374 minutes) | +1 (713 minutes) |
Anthony Randolph (-6) | -17 (462 minutes) | 0 (667 minutes) |
Nicolas Batum (+7) | -3 (909 minutes) | +20 (544 minutes) |
Patrick Patterson (+2) | -2 (340 minutes) | +17 (528 minutes) |
Trevor Booker (+7) | -5 (632 minutes) | +26 (432 minutes) |
Rodrigue Beaubois (-4) | -10 (363 minutes) | +8 (337 minutes) |
Earl Clark (-10) | -37 (303 minutes) | -2 (81 minutes) |
*Toney Douglas' O Rating went from 119 to 106 from the second half of his rookie year to his whole career. I thought that was more indicative of his play than the corresponding drop in D Rating.
What do all of these players have in common? You're a better analyst than I if you can find a common denominator. There are young players, old players, smart players, dumb players, athletic players, skilled players, players who played more during the first half of the year, and players who played more during the second half of the year on this list. The list also contains players that have gone onto productive NBA careers despite never matching these few months of scorching play, such as Nic Batum, and players whose careers have completely cratered, such as Rodrigue Beaubois. What about the players whose development was "real"?
Player (Career Net Rating) | Net Rating 1st Half | Net Rating 2nd Half |
Taj Gibson (+5) | -5 (1256 minutes) | +5 (948 minutes) |
Jordan Crawford (-11) | -37 (160 minutes) | -18 (867 minutes) |
Brandon Rush (-7) | -22 (990 minutes) | -9 (813 minutes) |
Serge Ibaka (+14) | -2 (707 minutes) | +11 (616 minutes) |
Darrell Arthur (-5)* | -17 (917 minutes) | +2 (547 minutes) |
Goran Dragic (+1) | -27 (312 minutes) | -12 (416 minutes) |
James Johnson (-7) | -24 (369 minutes) | -8 (388 minutes) |
Chris Douglas-Roberts (-7) | -22 (209 minutes) | -5 (375 minutes) |
Jeff Teague (-1) | -17 (436 minutes) | +2 (283 minutes) |
Greivis Vasquez (-6) | -17 (649 minutes) | -6 (211 minutes) |
*Arthur's career has been torpedoed by an Achilles injury, but in the two years following his rookie season, he played at a similar level to the second half of his rookie campaign.
One of the immediately noticeable aspects of this group's performance is that, for most of the group, their second half outburst represented their full statistical potential, despite another eclectic grouping of players. The only exception to this rule is Goran Dragic, who has made several gigantic leaps in productivity throughout his NBA career, completely confounding normal developmental curve expectations along the way. Another piece of data to be gleaned from these charts is that the players in the second chart were much worse than the players in the first chart during the second half of their rookie seasons. The players who held onto their gains had an average (unweighted by minutes) net rating of -4. The players who regressed had an average net rating of +8. Finally, these players were, on average, drafted lower than their unimproved counterparts, perhaps a result of random variance or a suggestion that some players drafted in the lottery may have less room to grow.
What does this data mean for the questions asked at the beginning of the piece? It appears it should not be expected for players to show dramatic improvements during the second half of their rookie seasons. Only 20 of the 100+ rookies showed this substantial improvement. It is also clear that showing this dramatic improvement is not sufficient evidence that a player is a "quick learner" or will improve in the future. Anthony Randolph and Michael Beasley provide evidence to the contrary. In fact, this sort of improvement may suggest that they have already made their possible adjustments or have merely enjoyed a couple months of unsustainable shotmaking.
Gorgui and Shabazz have played the majority of their minutes after the All-Star break, which means that their overall statistics are closer to their second half averages, providing a reason for optimism, but the same could be said of Toney Douglas, Anthony Randolph, and Patrick Patterson (and nearly Rodrigue Beaubois), all of whom have spent the last few years playing close to, or below, their overall rookie production. It's also worth noting that players drafted later, like Dieng, have a better track record (in these few years) of holding onto rookie improvements.
I don't want to throw cold water on dreams of Dieng becoming the next Mutombo or Shabazz becoming a productive NBA player, but each player faces an uphill battle, especially considering how much each has improved since entering the league. While there are plenty of examples of horrible rookies becoming useful players, there aren't many examples, in these three years, of horrible players getting better in their first year and taking another leap in subsequent years. Based on this evidence, which is far from complete, the Wolves should treat the consolidation of any improvement shown by their rookies as something closer to a 50/50 proposition than a given when planning any moves this summer, not limited to but including, the jettisoning of any Montenegrin folk heroes.