Georgetown NBA Analytics Case Presentation for the 2014 UNC Basketball Analytics Summit
-
Upload
nikhil-oza -
Category
Sports
-
view
140 -
download
1
description
Transcript of Georgetown NBA Analytics Case Presentation for the 2014 UNC Basketball Analytics Summit
Show Me the Money:The Factors that Affect the NBA Free Agent Market
Nik Oza, Camden Hu, Xavier Weisenreder and Nick Barton
Introduction• What factors contribute to a team’s ability to land
a franchise-altering free agent? • Which teams in the NBA could reasonably expect
to do so in the next five years? • How does the probability vary on a case-by-case
and year-to-year basis?
Franchise-Altering Players• Our Definition:• 25.6% of Salary Cap (15 million in 2013-14) for UFAs• Maxed-out RFAs
• Makes sense given max salary constraints with NBA contracts
• Definition helps to reduce hindsight bias; just because a player didn’t pan out doesn’t mean he wasn’t considered a franchise-altering player at the time
Restricted Free Agents• 2005 CBA – present: 27/27 maxed out RFAs signed
with same team (right to match)• Max offer sheets always get matched• Max RFAs are very unlikely targets, and should not
be part of other teams’ plans
Case Studies: RFAs• 2 RFAs changed teams (Kenyon Martin & Lamar
Odom) *1999 CBA• Kenyon Martin sign and trade to Nuggets• Lamar Odom signs with the Heat
Unrestricted Free Agents• 9 of 26 UFA who hit the market changed teams• Another 12 were extended before hitting the
market as a UFA
Case Studies: UFA “Overpays”• Define “overpay” as more than any other offer• Bulls sign Carlos Boozer for 5 years/$90 million• Bulls sign Ben Wallace for 4 years/$60 million• Magic sign Rashard Lewis for 6 years/$118 million
• UFA “overpays” never justify cost
Case Studies: UFA• Amar’e Stoudemire signs with the Knicks to lock
up max contract• Very probable max offers elsewhere
• Heat re-sign Dwyane Wade, sign-and-trade for LeBron James and Chris Bosh• Max offers elsewhere• LeBron: coach/management conflict
• Rockets sign Dwight Howard in free agency• Max offers elsewhere• Coach/management conflict
Statistical Analysis
Logistic Regressions• Limited Sample Size• No Significant Variable
Found
1. K-Means Clustering, 2-5 means
2. Poisson for count of UFA signed by each Cluster
• Team Win% last 10 seasons (significant)
• Total Championships• Avg Temperature in-
season• Rank of Market Size
(significant)• State Income Tax Rate
Strategy 1: Overpay!• If a team lands a free agent due to paying more
than any other team, he is probably not a franchise-altering free agent
• The Winner’s Curse• Examples:• Rashard Lewis• Carlos Boozer
Strategy 2: Other Superstars• James, Wade, and Bosh • Dwight Howard (joining with James Harden)• Might need a coach/management conflict• Requires intelligent cap foresight
Strategy 3: Don’t Count on It
• It takes an extraordinary situation in order to acquire a premier free agent without overpaying or other stars
• 67 “franchise-altering” players since 2003:• 56 signed with same team• 7 “overpays” at the time• 3 joined up with other stars• Amar’e Stoudemire locks in max deal in NYC market
The Other Route: Trading for Franchise-Altering Players• Shaquille O’Neal• Pau Gasol• Vince Carter• Ray Allen*• Kevin Garnett• Carmelo Anthony*
• Deron Williams*• Chris Paul• Joe Johnson*• Jason Kidd• James Harden• Dwight Howard (LAL)
*xRAPM-based WAR (Replacement Level = 4 pts below avg) didn’t justify contract
Projecting the Next 5 Years• Big Market• Winning teams• Cap space• Star(s) under contract (or ability to match via RFA)
Projecting Summer 2014
Players• James*, Wade*, Bosh*• Duncan*• Nowitzki*• Carmelo Anthony
Teams• Lakers• Suns?
Projecting Summer 2015
Players• Rondo• Dragic• Love• Aldridge• M. Gasol• Hibbert
Teams• Knicks*• Mavericks• Rockets• Suns• Blazers
Projecting Summer 2016
Players• Conley• Durant*• Noah?• Howard?
Teams• 76ers• Raptors• Celtics• Pelicans• Jazz?• Cavs?• Hornets?• T’Wolves?• Wizards?
Projecting Summer 2017
Players• Rose• Curry• Chris Paul• Blake Griffin• Westbrook• Ibaka• Jrue
Holiday?
Teams• 76ers• Warriors• Pacers• Raptors• Magic?• 2016 teams
that save cap space for 2017
• Possible franchise-altering player movement!!!
Conclusion
• It is very unlikely that a franchise-altering player is acquired from another team through UFA/RFA
• Slight advantage for:• Big Market Teams•Winning Teams• Teams with Stars Under Contract
• With multiple stars available in UFA, more likely to be franchise-altering player movement• 2017?
Appendix
Estimate Std. Error Z value Pr(>|z|)
Intercept 2.015e+00 7.751e+00 0.260 0.795
Age -1.742e-02 2.080e-01 -0.084 0.933
Old team Win% -1.192e+00 3.603e+00 -0.331 0.741
PER -1.442e-01 1.978e-01 -0.729 0.466
Last Contract Total Salary
1.947e-08 2.161e-08 0.901 0.368
Rings at Signing 9.211e-01 1.009e+00 0.912 0.362
No. of Old Team AS
-5.981e-02 6.560e-01 -0.091 0.927
Old Team Market Rank
1.167e-02 6.133e-02 0.190 0.849
Old Team Weather
3.587e-02 5.075e-02 0.707 0.480
Old Team State Income Tax Rate
1.956e+00 1.205e+01 0.162 0.871
• Logistic Regression for UFA change teams or no
No Variable shows up as significant
Appendix• K-means clustering on:• Win% last 10 years• Rank of Market Size• Income tax rate• Total championships• Weather
• Poisson Regression for count of UFA signed/resigned/extended by 3 clusters of teams
Estimate
Std. Error
z value Pr(>|z|)
(Intercept)
-2.66065 2.99696 -0.888 0.3747
Win% over last 10 years
7.67806 5.71029 1.345 0.1788
Market rank
-0.06993 0.03247 -2.154 0.0313*
Appendix• Poisson Regression for count of UFA
signed/resigned/extended by 4 clusters of teams
Estimate
Std. Error
z value Pr(>|z|)
(Intercept)
-1.31413 2.26635 -0.58 0.56202
Win% over last 10 years
5.72779 4.57953 1.251 0.21103
Market rank
-0.10258 0.02903 -3.5340.00041**
*
Appendix• Poisson Regression for count of UFA
signed/resigned/extended by 5 clusters of teams
EstimateStd. Error
Z value
Pr(>|z|)
Intercept -5.53178 2.97049 -1.862 0.06257 .Win% over
Last 10 years13.79608 6.11270 2.257
0.02401 *
Market rank -0.08832 0.02878 -3.0680.00215
**
References• http://www.basketball-reference.com• http://www.shamsports.com• http://www.cbafaq.com/salarycap.htm• http://www.gotbuckets.com• http://stats-for-the-nba.appspot.com