Announcement

Collapse
No announcement yet.

Draft Production Expectations based on Draft Slot - Historical Review

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Draft Production Expectations based on Draft Slot - Historical Review

    With the draft being the subject of intense focus, there have been plenty of discussions regarding the value of an "unproven rookie" vs a current player. This got me thinking, what is the reasonable expectation of production for the various slots in the draft.

    So to basketballrefence.com I went and used the following parameters:

    10 different draft slots since 1980 (34 drafts) was accounted for. The draft slots I chose were #1, #2, #3, #4, #5, #9 (for DD), #10, #14 (end of lottery), #20 and #25* for a total of 10 slots. Note that the 25th pick was technically in the 2nd round prior to 1989, but is still the 25th overall selection.

    Each player's statistics in PPG and FG% were used, and then the greater of RPG or APG to discern between bigs and guards while adding a secondary production level. Defensive stats were not used as there are no advanced stats for the full range, and relying solely on BPG or SPG doesn't necessarily show D. Feel free to save your criticism on this point unless you are willing to do the research yourself.

    Each draft slot was then given averages for PPG, FG% and (when applicable) RPG and APG. Each average was then ranked against the other draft slots.

    I debated dropping the highest and lowest scores for each category to be true to statistical practices, but in the end, I don't have the energy for that. I may add other draft slots in the future (I saved my Excel spreadsheet for further population).

    Here are my findings:

    The first overall selection averages 17 PPG, 8 RPG and 6.8 APG while shooting 49%. Anthony Bennett lies beyond what is normal, so I will state that his stats are the lowest for each category. For the below comments, Bennett will be excluded.

    The lowest scoring #1 is Kwame Brown at 6.6 PPG. The highest scorers are LeBron 27.5 PPG and Allen Iverson 26.7 PPG. Only 3 top picks failed to average double digits in their career. Bennett, Brown and Pervis Ellison.

    The lowest rebounding #1 (where applicable) is Andrea Bargnani at 4.9 RPG. The highest is Dwight Howard at 12.9 RPG.

    The lowest assists for #1 picks (where applicable) is Kyrie Irving 5.9 APG while the highest is John Wall at 8.2 APG. It should be noted that there are only 4 PG's taken #1 since 1980.

    The worst FG% is John Wall at 42% and the highest is Shaq and Howard at 58%.

    Of the #1 picks, only 6 have averaged a double-double over their careers. They are Tim Duncan (20 & 11.2), Blake Griffin (21.1 & 10.3), Dwight Howard (18.3 & 12.9), Shaq (23.7 & 10.9), Hakeem Olajuwon (21.8 & 11.1) and David Robinson (21.1 & 10.6). That means there is a 17.6% chance of drafting a double-double level producer with the #1 pick.

    The #1 pick ranked the highest of the slots for PPG, RPG and FG%, while finishing 2nd for APG.


    The second overall selection averages 12.7 PPG, 6.2 RPG, 6.8 APG and 46.8%. Len Bias is the outlier here, as he died after being drafted by the Celtics in 1986 and never played a game.

    The lowest scoring #2 is Hasheem Thabeet at 2.2 PPG and the highest is Kevin Durant at 27.7. Nine players failed to average 10 PPG in their career: Darko, Shawn Bradley, Camby, Chandler, Danny Ferry, Kidd-Gilchrist, Jay Williams and Stromile Swift.

    The lowest rebounding #2 is also Thabeet at 2.7 RPG, edging out Danny Ferry's 2.8 RPG. The highest is Okafor at 9.9.

    The lowest assists for #2 is Jay Williams at 4.7 APG and the highest is Isiah Thomas at 9.3 APG. The #2 slot saw 7 PGs drafted.

    The worst FG% is Jay Williams at 39.9% (Jason Kidd was 40%) and the best was Tyson Chandler at 58.5%.

    Of the #2 picks, only none have averaged a double-double over their careers. The closest would be Okafor (12.3 & 9.9), Camby (9.5 & 9.8), Aldridge (1.8.8 & 8.1) and Isiah Thomas (19.2 & 9.3).

    The #2 pick ranked 4th in PPG, 2nd in RPG, 1st in APG and 3rd in FG%.

    The third overall selection averages 14.3 PPG, 6.1 RPG, 4.2 APG and 46%. The outlier here is 2013 draft pick Otto Porter but his production isn't necessarily that far off from Chris Washburn of the 1986 draft. Porter is dropped for the below.

    The lowest scoring #3 is Washburn at 3.1 PPG. The highest is Michael Jordan at 30.1 PPG.

    The lowest rebounding #3 is Adam Morrison at 2.1 RPG. The highest is Buck Williams at 10 RPG.

    The lowest assists for #3 is Ben Gordon at 2.6 APG and the highest is Deron Williams at 8.8 APG. There were 12 players who qualified for this category.

    The worst FG% for #3 is Adam Morrison at 37.3% and the highest is Kevin McHale at 55.4%.

    Of the #3 picks, only 1 averaged a double-double over their career, Buck Williams. There is a 0.029% chance of drafting Michael Jordan with the third pick overall, although I'm fairly certain that figure will continue to trend down.

    The #3 pick ranked 2nd in PPG, 3rd in RPG, 6th in APG and 4th in FG%.


    The fourth overall selection averages 13 PPG, 5.7 RPG, 5.2 APG and 45.3%. Cody Zeller (5 PPG, and 4 RPG) is the outlier and is omitted for below.

    The lowest scoring #4 is Bill Garnett at 5.5 PPG. The #4 pick of the 1982 draft lasted onlt 4 seasons, split between Dallas and Indiana. The highest scorer is a tie at 19.3 PPG between Chris Bosh and Stephon Marbury.

    The lowest rebounding #4 is Dennis Scott (who had even fewer assists) at 2.8 RPG. The highest is Dikembe Mutombo at 10 RPG.

    The lowest assists for #4 is Byron Scott at 2.5 APG and the highest is Chris Paul at 9.9 APG. There were 10 players that qualified for this category.

    The worst FG% is Wesley Johnson at 40.7% and the highest was Eddy Curry at 54.5%.

    Of the #4 picks, none have averaged a double-double over their career, but Chris Paul (18.6 & 9.9) and Mutombo (9.8 & 10) are very close.

    The #4 pick ranked 3rd in PPG, 4th in RPG, 4th in APG and 6th in FG%.


    The fifth overall selection averages 11.9 PPG, 5.6 RPG, 5.3 APG and 45.4%. Alex Len (2.2 & 2.7) from 2013 is the outlier and omitted from below.

    The lowest scoring #5 is Nikoloz Tskitishvili at 2.9 PPG. The highest scorer is Dwayne Wade at 24.4 (just edging his TV commerical buddy Charles Barkley at 22.1).

    The lowest rebounding #5 is a terrible tie with 2.2 RPG between Jonathan Bender and James Ray. The highest rebounder is another tie, as both Charles Barkley and Kevin Love average 12 RPG.

    The lowest assists for #5 is Steve Smith at 3.1 APG. The highest is Ricky Rubio ay 7.9 APG. Only 6 players qualified for this category.

    The worst FG% is Rubio at 35.9% and the best is Barkley at 54.1% (just edging Jonas Valanciunas at 52.7% - the only two players in this slot to avg over 50%).

    Of the #5 picks, 4 have averaged a double-double over their careers; Barkley (22.1 & 12), DeMarcus Cousins (17.4 & 10), Kevin Garnett (18.7 & 10), and Kevin Love (18.6 & 12).

    The #5 pick ranked 5th in PPG, RPG and FG%, while ranked 3rd in APG.

    The ninth overall selection averages 11 PPG, 5.4 RPG, 4.9 APG and 47.2%. No justified outlier exists in this group, but Patrick O'Bryant is the lowest in PPG and RPG at 2.1 and 1.4 each. He'll be omitted from the below.

    The lowest scoring #9 is Joel Pryzbilla at 3.9 PPG. The highest is Dirk Nowitzki at 22.6 PPG. Demar ranks 4th highest behind Dirk, Amare, and T-Mac.

    The lowest rebounding #9 is Rodney White at 2.2 RPG. The highest is Andre Drummond at 10.1 RPG.

    Only 3 players qualified for APG: DJ Augustin at 4.0 , Trey Burke at 5.5 and Kemba Walker at 5.1

    The worst FG% is Ed O'Bannon at 36.7% and the best is Drummond at 61%.

    Of the #9 picks, only 1 has averaged a double-double over their career; Andre Drummond at 10.3 and 10.1.

    The #9 pick ranked 6th in PPG, 6th in RPG, 5th in APG and 2nd in FG%.

    Since this is a Raptors forum, I must add, DeMar is below average for FG% and RPG, while 4th best in PPG.


    The tenth overall selection averages 10.5 PPG, 4.8 RPG, 3.8 APG and 45.3 FG%. No statistical outlier exists in this group.

    The lowest scoring #10 is Mouhamed Sene at 2.2 PPG. The highest is Paul Pierce at 21.5 PPG.

    The lowest rebounding #10 is Keith Edmonson at 1.5 RPG. The highest is Horace Grant at 8.1 RPG.

    The lowest assists for #10 is Jimmer at 1.5 APG (even Austin Rivers gets 1.9) while the highest is Pooh Richardson at 6.5 APG. This category had 12 players qualify.

    The worst FG% is Luke Jackson at 35.7% and the highest is Andrew Bynum at 55.7%.

    Of the #10 picks, none have averaged a double-double over their career. Horace Grant (11.2 & 8.1) and Andrew Bynum (11.5 & 7.7 ) are the closest.

    The #10 pick ranked 7th in PPG, RPG and FG% and 8th in APG. (#10 is the 7th highest slot evaluated).


    The 14th overall selection averages 8.7 PPG, 3.9 RPG, 3.9 APG and 44.1 FG%. There are several players with terrible numbers that could dominate the board, so I'll omit Rich King (1.9 PPG and 1.0 RPG) and Scott Haskin (2 & 2).

    The lowest scoring #14 is Yinka Dare at 2.1 PPG. The highest is Clyde Drexler at 20.4 PPG.

    The lowest rebounding #14 is Shabazz Muhammad 0.7 with special honour to Alfredrick Hughes and his 1.7 RPG. The highest is Troy Murphy at 7.8 RPG

    The lowest assists for #14 is William Avery at 1.4 and the highest is Tim Hardaway with 8.2 APG. Seven players qualified for this category.

    The worst FG% is William Avery at 33% and the highest is Walter Berry at 53.9%.

    Of the #14 picks, none have averaged a double-double over their career. Clyde Drexler leads the group for "star power" but good players like Peja Stojakovic, Tim Hardaway and Dan Majerle are in the group.

    The #14 pick ranked 8th in PPG, 8th in RPG, 7th in APG and 9th in FG%.


    The 20th overall selection averages 7.2 PPG, 3.2 RPG, 3.4 APG and 44.6%

    The lowest scoring #20 is Julius Hodge at 1.2 PPG. The highest scorer is Larry Nance at 17.1 PPG.

    The lowest rebounding #20 is Paul Grant at 1.1 RPG. The highest is Zydrunas Ilgauskas at 7.3 RPG.

    The lowest assists for #20 Julius Hodge at 0.8 APG and the highest is Jameer Nelson at 5.2 APG. 8 players qualified for this category.

    The worst FG% is BJ Tyler at 38.1% and the best is Larry Nance at 54.6%.

    Of the #20 picks, none have averaged a double-double over their career. 6 players in this group averaged double digit scoring while 8 averaged less than 5.

    The #20 pick ranked 9th in all categories except FG% was 8th.


    The 25th overall selection averages 6.1 PPG, 3.2 RPG, 2.9 APG and 43.3%. Shaun Vandiver never played a game, so he isn't included in any way.

    The lowest scoring #25 is Tim James at 1.6 PPG while the highest is Jeff Ruland at 17.4 PPG. 5 players averaged double digits in scoring (Batum, Al Harrington, Mark Price, Ruland and Gerald Wallace), while 17 averaged less than 5. That means there is a 50% chance of drafting a guy who doesn't score 5 or more PPG over his career.

    The lowest rebounding #25 is Ron Moore at 0.6 while the highest is Jeff Ruland at 10.2 RPG.

    The lowest assists for #25 is Dominique Jones at 1.8 and the highest is Mark Price at 6.7 APG. Nine players qualified.

    The worst FG% is Ron Moore at 31% and the best is Jeff Ruland at 56.4%.

    Of the #25 picks, Jeff Ruland was the only to average a career double-double. In fact, if you have the #25 pick, Jeff Ruland is clearly the guy to get.

    The #25 pick ranked last in all categories (as expected).


    ~~~

    Summary:

    This data shows largely what we already know, that it is much harder to get high quality players outside of the top 10 of the draft. While there are certainly exceptions (noted above when applicable), the reasonable expectations for each slot should follow the 34 years of draft history.

    The Raps could end up with the 20th overall selection, and based on the historical data, we should end up with player production like current seasons of Mason Plumlee, Shaun Livingston, Tayshaun Prince or Norris Cole.

    Outside of the top 10 draft slots, it is unlikely to draft a player who will average at least 10 PPG, but the difference between 5th and 10th pick, is only 1.5 PPG.

    Rebounding can be found at a fairly predictable linear scale as the draft moves on. From 4th to 10th is less than 1 RPG difference.

    Here is the chart

    PPG RPG APG FG%
    1 16.99 8.04 6.75 0.489
    2 12.68 6.23 6.82 0.468
    3 14.25 6.12 4.24 0.460
    4 12.99 5.69 5.22 0.453
    5 11.93 5.55 5.28 0.454
    9 11.02 5.38 4.87 0.472
    10 10.45 4.83 3.82 0.453
    14 8.72 3.91 3.87 0.441
    20 7.20 3.24 3.40 0.446
    25 6.14 3.19 2.93 0.433


    Summary notices

    Super elite players who contribute in multiple categories did not get additional considersation in determining averages. So even though players like Jordan (RPG), and LeBron (APG) add another dimension, they each only contributed to the category that they performed better in. For example, Jordan's RPG did not count towards slot averages as he averaged more assists than rebounds. LeBron would be the opposite. These anomalies (for lack of better word) would simply make this exercise significantly more difficult. Over a 34 year span, the affects of these super elite players should be minimalized in the calculations.
    Last edited by Axel; Tue Feb 18, 2014, 12:02 PM.
    Heir, Prince of Cambridge

    If you see KeonClark in the wasteland, please share your food and water with him.

  • #2
    Good stuff Axel

    Interesting that the 9th pick and 5th pick are almost identical

    Comment


    • #3
      whooo... nice post. must have taken some serious effort.
      "Bruno?
      Heh, if he is in the D-league still in a few years I will be surprised.
      He's terrible."

      -Superjudge, 7/23

      Hope you're wrong.

      Comment


      • #4
        stooley wrote: View Post
        whooo... nice post. must have taken some serious effort.
        Thanks, it was a labour of love, so the time flew. Also was much easier once I figured out how to remove columns in BBRef and make for easy copy and paste in Excel. From there, simple avg, and rank functions did the rest.
        Heir, Prince of Cambridge

        If you see KeonClark in the wasteland, please share your food and water with him.

        Comment


        • #5
          Axel wrote: View Post
          Thanks, it was a labour of love, so the time flew. Also was much easier once I figured out how to remove columns in BBRef and make for easy copy and paste in Excel. From there, simple avg, and rank functions did the rest.
          As an accountant, I can appreciate some good spreadsheet work. Dope post.

          Comment


          • #6
            Interesting stuff
            @sweatpantsjer

            Comment


            • #7
              =Much respect on the leg work and for sharing what was a really good read.
              Last edited by Fully; Wed May 7, 2014, 10:28 PM.

              Comment


              • #8
                Great read. Interesting that trading down from #5 to #9 doesn't affect all that much. We could see a smart team use that data to trade down and get a quality rotation player in the deal
                @Boymusic66

                Comment


                • #9
                  Here's hoping MU pulls an outlier.

                  Comment


                  • #10
                    TSF wrote: View Post
                    Great read. Interesting that trading down from #5 to #9 doesn't affect all that much. We could see a smart team use that data to trade down and get a quality rotation player in the deal
                    I think this depends on the year though. Sometimes a great player will be at 5. Sometimes a bunch of average players will be at 5-9.

                    For example, during the anthony davis draft class, there were basically 7-8 guys that noone could really say were superior prospects amongst themselves.

                    This year, the drop off from 5-9 seems much larger. Randle - lavine/ennis.

                    These stats are great for handling expectations and predicting the quality of picks down the line (like how much will NYK pick be worth, and where do we expect it to fall). When it comes to the actual draft, I'm not a big believer in averaging values, or counting all stars taken at certain picks.

                    It's about talent evaluation, who's good and who's not. Do you make the pick for the right reasons, or for the wrong reasons?

                    I guess to put that idea into a mathematical concept: I think the variance on individual picks at the same spot from year to year is massive, probably just as much, if not more than the variance of two random consecutive picks in the same year. So given the huge variance, these numbers cannot really provide accurate expected values. What they do provide is an over/under type of valuation.
                    Last edited by stooley; Tue Feb 18, 2014, 02:41 PM.
                    "Bruno?
                    Heh, if he is in the D-league still in a few years I will be surprised.
                    He's terrible."

                    -Superjudge, 7/23

                    Hope you're wrong.

                    Comment


                    • #11
                      stooley wrote: View Post
                      I think this depends on the year though. Sometimes a great player will be at 5. Sometimes a bunch of average players will be at 5-9.

                      For example, during the anthony davis draft class, there were basically 7-8 guys that noone could really say were superior prospects amongst themselves.

                      This year, the drop off from 5-9 seems much larger. Randle - lavine/ennis.

                      These stats are great for handling expectations and predicting the quality of picks down the line (like how much will NYK pick be worth, and where do we expect it to fall). When it comes to the actual draft, I'm not a big believer in averaging values, or counting all stars taken at certain picks.

                      It's about talent evaluation, who's good and who's not. Do you make the pick for the right reasons, or for the wrong reasons?

                      I guess to put that idea into a mathematical concept: I think the variance on individual picks at the same spot from year to year is massive, probably just as much, if not more than the variance of two random consecutive picks in the same year. So given the huge variance, these numbers cannot really provide accurate expected values. What they do provide is an over/under type of valuation.
                      Some interesting points. Perhaps my next task (well after completing picks 6-24) would be to try and track the variance between years and draft groupings (top 5, 6-10, 10-14).

                      I am fascinated by the NFL model of valuing picks for trades. Obviously the NFL model would be very complex considering the greatly differing values of production when comparing left guard vs TE vs MLB and the larger draft pool (7 rounds of 32 as opposed to 2 rounds of 30). It should be theoretically easier to calculate a similar model for the NBA since players are typically defined as productive using only a few categories. As advanced stats are better tracked, we would be able to achieve better statistical models in the future.
                      Heir, Prince of Cambridge

                      If you see KeonClark in the wasteland, please share your food and water with him.

                      Comment


                      • #12
                        Axel wrote: View Post
                        Some interesting points. Perhaps my next task (well after completing picks 6-24) would be to try and track the variance between years and draft groupings (top 5, 6-10, 10-14).

                        I am fascinated by the NFL model of valuing picks for trades. Obviously the NFL model would be very complex considering the greatly differing values of production when comparing left guard vs TE vs MLB and the larger draft pool (7 rounds of 32 as opposed to 2 rounds of 30). It should be theoretically easier to calculate a similar model for the NBA since players are typically defined as productive using only a few categories. As advanced stats are better tracked, we would be able to achieve better statistical models in the future.
                        Yeah. I think there's a lot of cool stuff you can do with this kind of info. I think, for example, calculating the average of the best player taken between picks 20-30 each year, could provide some interesting info too.

                        Like, if we nail our pick at #20 or #21 and get the BPA, what is a reasonable expectation?
                        "Bruno?
                        Heh, if he is in the D-league still in a few years I will be surprised.
                        He's terrible."

                        -Superjudge, 7/23

                        Hope you're wrong.

                        Comment


                        • #13
                          stooley wrote: View Post
                          Yeah. I think there's a lot of cool stuff you can do with this kind of info. I think, for example, calculating the average of the best player taken between picks 20-30 each year, could provide some interesting info too.

                          Like, if we nail our pick at #20 or #21 and get the BPA, what is a reasonable expectation?
                          Reasonable expectation is exactly why I started down this road, regardless of the draft position. I will admit I was curious to see how our own draft picks (DD especially) stacked up to "reasonable expectation". I figure now that I have the framework and a better idea of how to best manipulate the Bbref data tables, it should only be a matter of time before I can populate the entire draft (both rounds) if I chip away at it here and there. Once the data exists in a nice easy spreadsheet, then there are further options available for data analysis. Another thing id like to look for is tracking the rare 3 stat star; guys like Jordan and Pippen who score, rebound and pass well and to incorporate defensive stats somehow (highest stat of steals or blocks seems the most likely way).
                          Heir, Prince of Cambridge

                          If you see KeonClark in the wasteland, please share your food and water with him.

                          Comment


                          • #14
                            Axel wrote: View Post
                            Reasonable expectation is exactly why I started down this road, regardless of the draft position. I will admit I was curious to see how our own draft picks (DD especially) stacked up to "reasonable expectation". I figure now that I have the framework and a better idea of how to best manipulate the Bbref data tables, it should only be a matter of time before I can populate the entire draft (both rounds) if I chip away at it here and there. Once the data exists in a nice easy spreadsheet, then there are further options available for data analysis. Another thing id like to look for is tracking the rare 3 stat star; guys like Jordan and Pippen who score, rebound and pass well and to incorporate defensive stats somehow (highest stat of steals or blocks seems the most likely way).
                            Right, once you get the spreadsheet, there's so many things you can look at.
                            "Bruno?
                            Heh, if he is in the D-league still in a few years I will be surprised.
                            He's terrible."

                            -Superjudge, 7/23

                            Hope you're wrong.

                            Comment


                            • #15
                              Great job Axel. Lengthy read, but definitely worth it. Cool that 9th picks are 2nd in FG%! Must be Demar keeping up the standard ...

                              Comment

                              Working...
                              X