web analytics

NCAA Advanced Statistical Plus/Minus Released

February 20, 2013
By

Advanced Statistical Plus/Minus (ASPM) is now available for NCAA Division 1 Men’s Basketball! ASPM was originally derived for the NBA, but the weights found for each event, as well as the methods for adjusting for opposition, should be equally valid for college basketball.

The full Tableau visualization will be permanently located at http://godismyjudgeok.com/DStats/aspm-and-vorp/ncaa-advanced-statistical-plus-minus-and-vorp/, under the “ASPM and VORP” menu above. I will reproduce it here for discussion:

EDIT: for those new to ASPM: it is an advanced box-score plus/minus estimation not based on play-by-play data. Its scale is in points per 100 possessions played (by the player). Zero is average; in the NBA, the super-elite players like Chris Paul, Kevin Durant, and LeBron James have ASPMs between +7 and +10. This is a rate stat, so playing time is not taken into account. VORP is (ASPM – replacement level) x %playing time, so it is points above a replacement-level player, per 100 TEAM possessions. Note–replacement level as a concept doesn’t really make sense in the NCAA.

First, let’s run down this visualization and what it can do:

There are 4 seasons present here, which is all the data publicly available. Filters are located at the top, for season, players, and teams.

There are two charts: the first chart shows overall average ASPM versus an estimate of the player’s position, and he second chart shows Offensive ASPM versus Defensive ASPM. Colors in all charts are based on the team color. If you select a dot in either chart, all of the seasons from that player will be highlighted on both charts. In addition, the table below the charts will be filtered to show only that player’s data. For example, select Victor Oladipo – 2013 in the first chart. He will be highlighted in both charts and only his data will show in the table below, showing his progression from an above-average freshman to an elite junior.

The team list to the right allows another method of filtering. You can select, say, A-10, and both charts and the table will be filtered to only show the A-10 teams. Or, you can multi-select (holding down CTRL and clicking) individual teams to compare a couple of teams. Since there is a lot of data to filter, though, this can be a bit slow.

Finally, you can select any player in the data table below the charts and that player will be highlighted in the charts. Again, multi-select works with this as well. You can compare Victor Oladipo’s incredible progression to Cody Zeller’s… non-progression (started elite, still exactly the same).

Enough about the nuts and bolts of the Tableau Viz… what does the ASPM say?

First of all: the 10 best player-seasons of the last 4 years (by ASPM, minimum 25 MPG):

  1. 2012 Anthony Davis (Fr) (+15 ASPM, +7 Offense, +8 Defense):  Excellent on offense, spectacular on defense.  2nd best defensive rating among qualified players, and as a freshman, also!
  2. 2013 Victor Oladipo (Jr) (+14.5/+8/+6.5):  Easily the best season by a wing in the 4 seasons.  Best wing defender in the past 4 years (well, tied with Aaron Craft in 2012).
  3. 2011 Kyrie Irving (Fr) (+13/+9.5/+3.5): Low minutes because of the injury, but a tremendous offensive season.  In 2011, only Jimmer was better on O.
  4. 2013 Trey Burke (So) (+12.5/+10/+2.5): So far this season Trey Burke is having the best offensive season of the past 4 years.
  5. 2012 Jae Crowder (Sr) (+12/+6.5/+5.5): In the same mold as Oladipo: lock-down wing defender, do the dirty work… amazingly well.
  6. 2013 Cody Zeller (So) (+12/+8/+4): Tremendous offensive post presence, relatively weak post defender.
  7. 2013 Kelly Olynyk (Jr) (+12/+9.5/+2.5): Even better than Zeller on offense, but even worse on defense.
  8. 2011 Kemba Walker (Jr) (+11.5/+8/5/+3): Excellent offensive lead guard, and not bad on defense for a PG, either.
  9. 2012 Cody Zeller (Fr) (+11.5/+7.5/+3.5): Cody was also really good as a freshman, and hasn’t changed much.
  10. 2012 Jared Sullinger (So) (+11.5/+6/+5): A beast in the post, outstanding on both ends of the court.
  11. 2010 Evan Turner (Jr) (+11/+7/+4): The best of 2010–a down season.  A uniquely talented 3 on offense.

Thoughts:

  • The results for ASPM, as applied to the NCAA, pass the smell test
  • Point guards skew towards offense, posts towards defense (as it should be)
  • There are a multitude of ways to slice this data and find interesting nuggets.
    • For example… Marcus Smart is the 5th best freshman in the past 4 years (behind Davis, Kyrie, Zeller, and Noel).
    • Or… there have not been many very good freshmen 3′s in the past 4 years.  Otto Porter?  Glenn Robinson?
    • The best defensive guards in the past 4 years are Marcus Smart and Aaron Craft, each well ahead of the rest of the pack

I wish that these advanced stats went a couple of years further back.  It’s hard to get a good handle on how elite, for example, Anthony Davis really was on D–we don’t have Greg Oden to compare him with, even.  I’d love to see Oden and Durant’s college numbers!

This is a good start for NCAA ASPM–what uses could there be for this data?

One Response to NCAA Advanced Statistical Plus/Minus Released

  1. Dan Dickey on February 21, 2013 at 5:14 pm

    Great, great stuff. Can’t say enough good things about it.

Leave a Reply

Your email address will not be published. Required fields are marked *

Current day month ye@r *

DSMok1 on Twitter

To-Do List

  1. Salary and contract value discussions and charts
  2. Multi-year APM/RAPM with aging incorporated
  3. Revise ASPM based on multi-year RAPM with aging
  4. ASPM within-year stability/cross validation
  5. Historical ASPM Tableau visualizations
  6. Create Excel VBA recursive web scraping tutorial
  7. Comparison of residual exponents for rankings
  8. Comparison of various "value metrics" ability to "explain" wins
  9. Publication of spreadsheets used
  10. Work on using Bayesian priors in Adjusted +/-
  11. Work on K-Means clustering for player categorization
  12. Learn ridge regression
  13. Temporally locally-weighted rankings
  14. WOWY as validation of replacement level
  15. Revise ASPM with latest RAPM data
  16. Conversion of ASPM to" wins"
  17. Lineup Bayesian APM
  18. Lineup RAPM
  19. Learn SQL