Draft value is real. Provable draft skill is rare.
Every June, thirty front offices spend first-round picks like they know something. Do any of them? This analysis prices every pick slot from the careers that actually followed, then asks which franchises and which colleges beat that price. The sample is every draft class from 1989 through 2015: each one with ten full NBA seasons to accumulate a career, scored in career Win Shares against a slot expectation fit isotonically over picks 1-60. No buckets, no functional form, and every table is produced twice, in Python and R, and reconciled to numeric tolerance before a word is written. The companion interactive viewer holds every lottery and full draft board since 1989.
A pick slot is a price
Before anyone gets credit for drafting well, the draft itself has to be priced. The expectation curve below is a weighted isotonic fit of career Win Shares over picks 1-60: pick 1 carries an expectation of 77 career Win Shares, pick 60 carries 2. Everything that follows is measured against this curve. A player's value is not his career; it is his career minus what his slot was already worth.
Twenty-seven of thirty error bars cross zero
Rank the franchises by career Win Shares above slot expectation and Oklahoma City leads at +489 over 69 picks, with San Antonio (+411), the Lakers (+251), Phoenix (+240), and Golden State (+204) behind. The LA Clippers sit last at −394. But the honest headline is the intervals: draft value is heavy-tailed, and 27 of 30 franchise intervals cross zero. The ordering is the estimate, not a verdict. Ten thousand fan arguments about "our front office drafts badly" rest on differences this data cannot certify.
Wake Forest, quietly, out-drafted everyone
Flip the question from who drafts to who supplies. Among colleges with 8+ draftees, Wake Forest leads at +323 Win Shares above slot with a 70% hit rate (Tim Duncan and Chris Paul will do that), ahead of UCLA (+224), Marquette (+201), Florida (+142), and Xavier (+117). Among high-volume programs with 15+ draftees, Kentucky has the best hit rate: 25 of 41 picks (61%) became 10,000-minute NBA players.
The biggest steal in the sample went first overall
"Steal" usually means a late pick who hit. Priced against slot expectation, the biggest steal of the era is LeBron James at +200 Win Shares above what pick 1 already promises. The rest of the top five: Dirk Nowitzki (+170 from pick 9), Chris Paul (+155 from pick 4), Kobe Bryant (+145 from pick 13), and Kevin Garnett (+136 from pick 5). A first-overall steal sounds like a contradiction. It is the strongest possible statement about a career: the most expensive slot in basketball, and he still quadrupled its price.
| Player | Year | Pick | Drafted by | WS above slot |
|---|---|---|---|---|
| LeBron James | 2003 | 1 | Cleveland Cavaliers | +200 |
| Dirk Nowitzki | 1998 | 9 | Milwaukee Bucks | +170 |
| Chris Paul | 2005 | 4 | New Orleans Pelicans | +155 |
| Kobe Bryant | 1996 | 13 | Charlotte Hornets | +145 |
| Kevin Garnett | 1995 | 5 | Minnesota Timberwolves | +136 |
Every lottery, every board, in the browser
The numbers above compress 27 draft classes into a few tables. The
interactive draft viewer
is the uncompressed version: every lottery since 1989 with pre-lottery odds
and movement, the full two-round board with career outcomes, colleges, and
teams, in light and dark, timezone-safe, no install. The full tables behind
this page (colleges.csv, teams.csv, steals.csv, pick_curve.csv,
each with an _r.csv twin from the independent R implementation) ship in
the repo, and the reconcile gate holds the twins equal.
Method, code, and verification
The full technical write-up (data provenance, model specification, validation gates, and stated limitations) is the README. The analysis is implemented independently in R and Python and reconciled to numeric tolerance; run_checks.sh replays every gate.