20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Thursday, October 22, 2015

Basketball Reference Elo Ratings Reveal Extent of Fan Bias Against Bill Russell, Kareem Abdul-Jabbar and Kobe Bryant

Ranking the greatest basketball players of all-time is a challenging and, to some extent, subjective exercise; there are so many factors and variables that it is difficult to make a completely objective evaluation, though it is possible to reasonably classify players into various broad categories such as top 10 or top 20--but even that cannot be done without controversy. My Pantheon series examines the careers of 10 of the greatest retired basketball players of all-time, plus four active players who also deserve consideration for Pantheon status. I do not dispute that the placement of players within the Pantheon can be intelligently debated, nor do I deny that a good case can be made for some players who I did not include.

Physicist Arpad Elo, a chess master who won the Wisconsin State Chess championship eight times, developed a rating system that not only ranks players but can also be used to predict the probability of victory based on the rating differences between players. The Elo rating system is most widely identified with chess, though it can and has been used for a variety of different games.

BasketballReference.com publishes so-called Elo player ratings for all NBA players who reached at least one of the following statistical levels: 10,000 points, 5000 rebounds, 2500 assists or 1000 steals plus blocked shots. BR.com assigned each qualifying player an initial rating of 1500 and then created fictional matchups between various players; visitors to BR.com can vote on who they think would win a given matchup. The matchups are generated randomly, so visitors cannot deliberately vote up (or vote down) any one particular player.

BR.com describes this project as "a community-based project with the goal of rating the best players in NBA history." Presumably, the idea is that by collecting the votes of a large number of people, bias will be smoothed out and a reasonable consensus about pro basketball greatness will be reached.

The ratings fluctuate but certain trends are consistent. Michael Jordan is usually ranked first by a fairly sizable margin and he is the only player currently rated above 2400. Larry Bird, Hakeem Olajuwon, Julius Erving, Karl Malone, John Stockton and David Robinson fill the next spots in various orders and players in that group have ratings clustered between 2300-2360. One could quibble with some of those choices--I would not put Malone or Robinson in the top 10--but Stockton is the only player who never came close to winning an MVP and who does not even belong in the top 30, let alone the top seven.

The BR.com ratings look objective superficially--each player has a four digit rating, just like chess players do!--but of course the ratings are not a product of actual competition; they are a product of fan voting, which is little more than a popularity contest. It is revealing to look at the list as a whole and see which players are so hated by the fans that the fans are unwilling or incapable of separating hate from reality.

For instance, an excellent case could be made that Kareem Abdul-Jabbar is the greatest basketball player of all-time. He holds the career regular scoring record (38,387 points), he holds the record for most regular season MVPs (six), he won six championships and he was a dominant player for most of his 20 year career (winning the Finals MVP as a 23 year old and as a 37 year old). It would be very difficult to rationally argue that Kareem Abdul-Jabbar is not at least a top 10 player of all-time. However, Kareem Abdul-Jabbar is not very popular and thus in the BR.com rankings he barely cracks the top 50, trailing Ray Allen, Chris Mullin, Dikembe Mutombo and Reggie Miller, players who had a fraction of the impact that Abdul-Jabbar did.

However, by far the most ludicrous rating for an all-time great player belongs to Kobe Bryant. The last time I checked (ratings change continuously), Kobe Bryant ranked 326th out of 560 qualifying players, just behind Darrell Armstrong, David Wesley, Joe Smith and Caron Butler and just ahead of Michael Adams, Johnny Green, Kurt Thomas and Mo Williams. One can have an intelligent conversation about Kobe Bryant versus LeBron James or Kobe Bryant versus Michael Jordan; one cannot have an intelligent conversation that begins with the premise that Bryant is not just outside of the 50 Greatest Players list but that he misses the cut for the top 300! If you hate Bryant that much, then why vote? How does that vote contribute to the conversation or add value to BR.com's Elo ratings? Placing Bryant in the midst of players who were fringe All-Stars at best makes BR.com's Elo ratings look amateurish and silly.

Then there is Bill Russell; the greatest winner in the history of North American professional team sports barely makes BR.com's top 40. I don't think that fans hate Russell the way that they apparently hate Abdul-Jabbar and Bryant but I do think that fans fail to understand the nature and extent of Russell's greatness. I am baffled that Russell's accomplishments are not more widely appreciated and that so many fans denigrate his skill set, particularly as an offensive player; he scored at least 16 ppg for six straight seasons en route to posting a solid 15.1 ppg career average, he ranked in the top 10 in assists four times and he ranked in the top four in field goal percentage four times. No, Russell was not an offensive powerhouse but he was hardly a liability at that end of the court and the things that he did well offensively--run the floor, pass and rebound--would translate very well to the modern era. Furthermore, the idea that a player of Russell's size could not play post defense in today's game is belied by the success of players like Dennis Rodman and Ben Wallace.

In theory, having fans vote about thousands of player simulations in order to rank the greatest players of all-time sounds like a fun idea that should produce a fair list, but in practice BR.com's Elo ratings simply reveal widespread ignorance and bias even among the subset of fans who are interested enough in analytics to visit BR.com.

Labels: , , , , , , , , , , , ,

posted by David Friedman @ 4:30 PM

39 comments