What Happens When PER and +/- Collide?PER and +/- separately and noted the strengths and weaknesses of each. Well, Roland Beech from 82games.com has combined variations of the two in an ad hoc measure that he calls the Roland Rating. Mr. Beech has also been kind enough to weigh in with some thoughts about his measurement, to help us understand the strengths and weaknesses of using it.
What is it: The Roland Rating is best understood by understanding PER and +/-. The measure is comprised of 3 parts: Production (similar to PER), Opponents’ Production (basically the PER of the person being guarded), and on-court/off-court +/- (The difference between the team’s +/- with a player on the court vs. when that player is off the court). The idea is that Production is a measure of how well a player is doing using box score stats, Opponent’s Production is a measure of how well a player is preventing their individual opponent from getting box score production, and on-court/off-court +/- is a team impact measure (i.e. how much is the team better or worse with that player on the court, as opposed to when someone else is replacing him). Roland Rating rates the net difference in Production a bit more than +/-, but combines them into one output number where higher is better.
Strengths: The biggest strength of Roland Rating is that it gives a more comprehensive view of a player’s impact than many individual stats do. It uses tangible offensive and defensive box score stats, as well as +/- stats to cover all of the major parts of the game.
Also, by combining PER and +/- into one stat, the Roland Rating lessens the weaknesses of each stat. For instance, PER has the issue of not being directly tied to either defense or team impact and +/- has the weakness of not being correlated with team usage (i.e. a role player that plays consistently with good players can have a higher +/- than a flawed impact player, even if the flawed impact player is a better overall basketball player). For instance, looking purely at on-court/off-court +/- you would find Sebastian Telfair (+8.7) rated above Pau Gasol (+7.1) while looking purely at PER would put Marreese Speights (20.75) over Chauncey Billups (19.75). Neither of these results would pass the sniff test. But Roland Rating, which uses the composite measure, ranks these four in a much more intuitive order with Gasol and Billups being significantly higher up the list than Telfair and Speights.
Weaknesses: Mr. Beech was very clear that he does not believe that there is one magical statistic that completely and accurately measures a player’s impact. In Beech’s words:
“I'll tell you up front though that I'm not a believer in 'one number overall player ratings' since I think player value is not some kind of constant, but is very much dependent on the specific teammates, coaching schemes, role the player is asked to fill, etc.”
This is a weakness for not just the Roland Rating, but for every statistic, and one that I try to hammer home with each blog on the subject: advanced stats are tools to be used to gain further understanding of the strengths and weaknesses of players. They are NOT replacements for making individual judgments based upon a composite of watching games as well as using many different types of analysis.
Also, Mr. Beech tells us that Roland Rating was “just kind of a quick wag at a player rating”, and on his website he mentions that there are more sophisticated analyses available. As mentioned in recent blogs on Shane Battier and Wins Produced, there is a lot of high order math being performed these days to really break down how a player’s performance impacts wins. While Roland Rating produces a good, fast read on player performance it is not necessarily optimized to be the best way to do so.
Usage: The Roland Rating is a comprehensive player measure, so in theory one could simply say “player X has a higher Roland Rating than player Y so he is better”…but as Beech pointed out, we should all know by now that there is no holy grail stat that is always right. You should still corroborate the Roland Rating with a variety of other stats and personal observations before making a final conclusion. Keep the pros and cons of the statistic in mind, know exactly what the stat is saying and what it isn’t, then decide in your own head how much weight to give to the results of this measure when making your own evaluation.
Quick Links for series:
Shane Battier (Thornbury's blog)