Freelance Friday: Expected Value in the NBA Draft


Freelance Friday is a regular series on Nylon Calculus where we accept submissions from around the internet to bring the broadest possible range of content and to provide opportunities for new voices in basketball analytics to be heard.

Today’s piece, from Saurabh Rane, discusses the expected value from each pick slot in the NBA Draft. Saurabh is a Bay Area native and longtime Warriors fan with an unhealthy obsession with numbers and sports. Follow him on @SaurabhOnTap, where he pretends to be a NBA coach or on his blog, Saurabh.R.

Have a question, comment or submission for Freelance Friday? Contact us via email at TheNylonCalculus at Gmail dot com or on twitter @NylonCalculus 

As we get closer to the 2016 NBA Draft, I wanted to quantify the expected value of a given draft slot. As a longtime Warriors fan, I’d find myself overrating, or rather over-projecting, the franchise’s numerous mid-late lottery picks (10 picks outside the top 5 since 2002). It was easy to become enamored by Brandan Wright’s theoretical ceiling of Chris Bosh or to envision Anthony Randolph leading the Warrior’s offense like Lamar Odom.

Obviously, neither of those scenarios came to fruition. Consequently, I wanted to quantify exactly what type of player teams can expect in a given draft slot to avoid the trap of optimism bias to which we are all susceptible.


The most difficult part of determining the expected value of a draft slot is quantifying a player’s career value in one number. For the sake of this analysis, player value will be quantified by a players single season peak Value Over Replacement (VORP). [1. A derivative of Box Adjusted Plus Minus (BPM), that accounts for time played.]

While far from a perfect measure, I felt VORP did a better job of ascertaining player for value for this purpose than Win-Shares (WS) or Player Efficiency Rating (PER). WS does a good job of identifying the top players, but fails in accurately capturing the value of average players, particular average players on good teams.[2. Harrison Barnes, a generally accepted league average to above-average player, has an equivalent WS to Dwayne Wade (4.9) because 73 is a lot more wins to share than 48.] Since PER normalizes stats to minutes played tends to overrate players that play lower minutes, but take lots of shots.[3. Michael Beasley finished 14th  with a PER of 22.5 last year; ahead of All-Star PFs such as Draymond Green, LaMarcus Aldirdge and Paul Millsap.]

I’m also only taking into account a player’s peak season per VORP. Looking at career averages would underrate players that played well into their prime, and overrate players that retired soon after their prime.[3. For example Grant Hill, who played well after his prime, has a career average VORP of 2.36 whereas Lamar Odom has a career average of 2.79 since he retired shortly after his prime, which dramatically undersells the impact of pre-injuries Grant Hill in Detroit. The debate of longevity versus peak is long-running, but for purposes of draft value, I think peak is more appropriate. Longevity is largely a function of talent and injuries. More talented players stay in the league longer, and talent is something that can be projected. Injuries, however, cannot, and which is why this analysis only looks at peak season VORP.]

The minimum requirements for a season to get recorded were 10 MPG and 41 games played. If a player did not meet those requirements in any season, they were considered a “bust” and given a peak VORP of 0. Additionally, only drafts from 1985 through 2012 were included, using the start of the Lottery on one end and the last year from which players can even plausibly be expected to have peaked at the other.

105 unique players have made an All-NBA 1st team and 407 unique players have made All-Star teams from those draft years. That means roughly 1.5 All-NBA 1st team players get drafted each year and 6.26 All-Stars. Using that information, the top 2.5% of all players since 1985 are All-NBA 1st teamers and top 10% are All-Stars. The average VORP of players with 10 MPG/41 GP was 0.9 so that was the cut-off for a Rotation player. A bust was anyone with a VORP of 0 or lower.


The below chart looks at the entire 1st round split up into 7 unique slots (1st overall, early lottery, early-mid lottery, mid-late lottery, late lottery, mid-late first, and late first).

Image 1 PeakVORP FirstRound19852012
Image 1 PeakVORP FirstRound19852012 /

Some things immediately stand out:

  • The first overall pick has historically had a very high floor. Almost 60% of first overall picks become All-Stars, and almost 25% become All-NBA First Teamers.
  • The mid-lottery range (picks 4-10) is unlikely (<25%) to produce an All-Star, and has a very low chance of producing an All-NBA First Teamer.
  • The late lottery (11-14) does not give teams a large advantage over picking in the mid to late first (15-30) in regards to picking All-Stars or All-NBA Players. It does give teams a larger chance of getting a rotation player (~50% vs ~40%).

The biggest takeaway here is how quickly team’s chances at an All-NBA 1st team player drop off. Even the chances of grabbing an All-Star fall significantly towards back end of the lottery.

This expected value data can prove useful in determining the value of a team’s assets and evaluating trades. Take for example, the rumored Boston/Philadelphia trade centered around the number 3 overall pick and Jahlil Okafor. Historically, per the chart below, the 3rd overall pick has produced a player who peaked at a single-season VORP of ~3. The players in the 2015-16 season that had a VORP of approximately 3. were Giannis Antetokounmpo, Karl-Anthony Towns, Nikola Jokic, Carmelo Anthony, Jae Crowder, Rudy Gobert, and Kevin Love. If Philly believes Okafor will not reach the level of those players 2015-16 seasons, it makes sense to make that trade.

Image 2 PeakVORP Lottery19852012
Image 2 PeakVORP Lottery19852012 /

The Markieff Morris trade is another interesting deal to look at. Morris, in the 13/14 and 14/15 seasons had an average VORP of 1.4, which is around the 66th percentile of players. Historically, a late lottery pick will produce an equivalent (or better) player ~36% of the time, and a bust 25% of the time.

While the above discussions illustrates the average value of a given draft pick, much of the purported value of lottery selections is the upside – the chances that the player drafted becomes a building block for the future. The chart below represents the historical percentage of a given pick becoming an All-NBA 1st Teamer, All-Star or rotational player.

Image 3 Draft Odds
Image 3 Draft Odds /

Quick observations from the above

  • The steep drop-off in All-Star chances in the early-mid lottery (picks 1-7), and the slower drop-off in the back-end of the lottery.
  • The low chance of getting an All-NBA First Teamer is in the mid-late lottery (picks 7-14).
  • The late lottery (11-14) is twice as likely (25%) to produce a bust than an All-Star.

This not only highlights the importance of the 1st overall pick but also suggests, from a ceiling standpoint, the mid-lottery is not especially valuable in terms acquiring corner stone talent. The expected value is of late-lottery picks is a definite contrast to how highly they have often been valued in trades.

At face value this may seem to validate tanking – if the 1st overall pick gives a 55% / 20% chance of getting an All-Star/1st Team All-NBA player and the 5th overall pick gives a 30% / 7.5% chance, it certainly makes sense to tank. However, the big caveat is the worst record does not guarantee the number 1 overall pick, only a top 4 pick. Moreover it does not even give significantly better odds than a slightly worse lottery position. With the worst record, teams are only guaranteed a bottom 4 pick, and the odds of picking 1/2/3/4 are 25%/21.5%/17.8%/35.7%. So in reality, the ~20% chance of getting an All-NBA player is closer to ~13% after adjusting for lottery odds. Conversely the 5th worst record jumps up from a 7.5% chance of 1st Team All-NBA Player to just below 10%:

Image 4 AdjDraftOdds
Image 4 AdjDraftOdds /

The above suggests that the returns to extreme tanking diminish fairly rapidly once a team is already in the range of the fourth worst record.

In regards to what this means for the 2016 draft, probably not much. Past results do not guarantee future performance, and on the micro scale of individual players in a given draft, individual player strengths, weaknesses and overall ability is more important than historical aggregates.

On the macro level, however this should provide insight into proper valuation of draft picks as well as to illuminate the dubious benefit of narrowly missing the playoffs as opposed to making the playoffs as an 8 seed. Most importantly, I hope I have succeeded in illustrating what teams can realistically expect from a given draft pick rather than predicting immediate stardom for every player selected.