Trying to figure which college kid or 19-year old international player is going to be able to transition to the NBA is a pretty dicey proposition. Many things can go unexpectedly wrong for even supposedly can’t miss prospects and occasionally a “Mr. Irrelevant” picked at number 60, like Isaiah Thomas or an undrafted player like Jeremy Lin can find his way in the league ending with a decent payday (though not too often).
However, there is a definite form to the value available in the draft, the value of draft picks on average falls as the draft goes on in something of a logarithmic pattern. Meaning that the average value of draft picks falls fairly quickly in the first few selections, then flattens out to fall more gradually. So, for example, the typical difference between having the first pick or the tenth pick is much bigger than difference between the twentieth or the thirtieth.
This spring I created a draft model based on prospects’ playing performance prior to the draft to project how well they would perform in the NBA. The model utilized data from 2002 to 20011 measuring how well the players performed by their third or fourth year in the league based on a stabilized version of Alt Win Score (ASW). If the player didn’t last that long in the NBA they were assigned a replacement level value.
The model was able to beat the performance of the average NBA general manager over that time, as well as out of sample in both of the last two years. Of course, not all GM’s are necessarily average. Since 2002, when the draft training data starts, only Miami’s Pat Riley, San Antonio’s RC Buford and Mitch Kupchak have managed to hold onto their jobs through to the present day. They are then only followed by Boston’s Danny Ainge and Washington’s Ernie Grunfeld, for some reason.
As George Carlin might say, laundry doesn’t draft players, front offices do. So I focused on two of those teams: the San Antonio Spurs and RC Buford, because of their legendary pick ups in the draft — Tony Parker, with the 28th pick in 2001, and Manu Ginoboli, with the 58th pick overall in 1999 — and the Boston Celtics, because: homer. The data I’m working with comes after the Parker and Ginoboli drafts, so I can look at whether their selections were anomalies or part of a larger pattern.
Using the DraftExpress historical draft data, I tracked both the Spurs and Celtics selections and their performance in the NBA compared to both my model’s projected rating and the expected value at that draft spot by the end of their rookie scale contracts, where they should have the most value for the team.
In that time there were 13 Spurs draft selections who played in the NBA, and a heroic 22 Celtics picks. Quantified in a couple different ways the answer indicated the same thing, even after the draft steals with Parker and Ginoboli the Spurs have continued to out-draft the average for the rest of the league. The Celtics, on the other hand, under Ainge have looked much closer to average. For every Rajon Rondo there has been a Marcus Banks, and for every Jared Sullinger there has been a Fab Melo (in same draft, back to back!).
Below are both the expected production based on the Spurs draft position and the actual production their picks have shown, with the key column in my mind being the last one with the equivalent pick to the actual production.
In the overall draft the Spurs have had a median pick of 33rd with the players in the study (they traded away some 2nd round picks or never signed others). But, the median value they have gotten out their picks has been more like a late lottery selection. The Spurs first round picks have been even better with median production of a 10th overall selection.
The Celtics, however, under Ainge have been less consistent, and much closer to their expected production based on their draft position. In part the Celtic’s score has been hurt by players that were unable to stick in the NBA, like JaJuan Johnson and Fab Melo. On the other hand, the volume of picks accumulated by the Celtics front office has given them more opportunities to find value in the draft.
While the Ainge picks have held their own, the performance is virtually identical to the NBA average over this time period.
I should point out here that even for the most emulated franchise in the association selecting draft picks is not a sure process. The difference between the Spurs picks and the rest of the league is not quite statistically significant at the 90% confidence level, however, given the omission of their two signature draft successes in Parker and Ginobili it is still an impressive record.
The Celtics have played the role of volume drafters over the study period, sort of the Allen Iverson approach to the draft, and they look to continue that holding the Nets, Clippers and now Cavaliers picks in the next four years.
Finally, the draft model can give us a little insight into the draft vs development debate. The prediction of my draft model based on pre-NBA production for the Spurs draft picks is almost exactly half way between the expected value for their draft spot and the player’s actual performance for the Spurs players. That lends support to the idea that the Spurs are both maximizing their selection and development of talent.