Last week, John Hollinger noted that Brandon Rush will likely end this season as one of the worst NBA players to ever lead his team in minutes. Aside from Bruce Bowen and Jason Collins, who he rightfully points out as superior defenders, Brandon’s 9.97 PER (player efficiency rating, an advanced stat created by Hollinger) will make him the only player to lead his team in minutes with a single-digit PER. (For reference, a 15.0 PER is exactly NBA average.)
This, really, is all Hollinger was saying. He did drop a line that “[Rush will] claim the dubious distinction of being the worst player ever to lead his team in minutes,” but the word “worst” in that sentence is likely just blunt Hollinger shorthand for “worst under this metric of defining a player’s worth.” If you talked with the man, I’m sure he would readily admit that he hasn’t watched a ton of Pacers games this season. Very few people who have no connection to this team have. They’re just not very watchable.
Meanwhile, as Indy Cornrows points out, another stat-head has seemingly lept up like a knight in shining armor to defend Brandon’s honor. Wayne Winston, who spent years employed by the Mavericks as Mark Cuban’s right-hand man for advanced stat knowledge, states that under his preferred system, adjusted +/-, Rush has more worth.
Specifically, he has more worth than Troy Murphy.
Rush has an Adjusted +/- of around 0 and Murphy has a -7 Adjusted +/- which means we estimate that Troy Murphy in for 48 minutes is 7 points worse than an average NBA player. So we believe Murphy must be doing some bad things that do not show up in the box score and in all likelihood Rush does many good things that do not show up in the box score. Jim O’Brien must also believe that Rush has “hidden virtues.”
He goes on to note that the Pacers play better when Danny is on the floor with Brandon than they do when he is on the floor with … you guessed it … Murphy.
For exmaple, [sic] we find that when Granger is in the Pacers play 2 points per game worse with Murphy in than Rush in. When Hibbert is in, the Pacers are 12 points per game better with Murphy out than with Rush out. A bad PER does not necessarily mean you hurt the team and a good PER does not imply that you surely help the team!
Interesting stuff. And stuff that definitely adds some more nuance to Brandon’s statistical contributions to the team.
What Winston doesn’t take the time to point out, however, is a player’s adjusted +/- over a single season can often be misleading. The Mavs themselves learned this the hard way, as I heard Mark Cuban himself say twice at this year’s MIT Sloan Sports Analytics Conference, aka Dorkapolloza. (Here is one of the panels I covered there.)
Cuban overvalued a young Evan Eschmeyer at one point based on his high adjusted +/- numbers, signing him to a deal that later turned out to be way more than Eschmeyer was worth. After that — and just with a lot more experience, evidence and understanding acquired over the years — Cuban said that he now values the insights shown by adjusted +/- over multiple seasons in his decision making a lot more so than anything it can turn up in single season. Essentially, for Cuban, seeing a good adjusted +/- number for one year is like having a hypothesis. Seeing it recur over several seasons starts to give you a theory.
As a billionaire, it comes as little shock that Cuban prefers to make decisions based on theories rather than hypotheses.
And in addition to that, Cuban also has about a million other metrics — many of which are gathered through proprietary means and don’t exist in the public domain — he uses to cross-reference findings to see if something like a negative PER is an outlier. Does it match with adjusted +/-? Does it sync up with other defensive metrics, like the opposing team’s FG% while a player is on the court? Does it look similar to what you see on that players customized Synergy Sports video channel? Does it align with any of the other likely-amazing metrics to evaluate a player that I have never even heard of?
Getting back to Winston … let’s just remember that he really isn’t saying that he thinks Brandon is great. His main point is merely that Troy Murphy is worse. And more than even caring one way or the other about these two players, he seems to mainly be advocating that this is one time when adjusted +/- tells a better story than PER.
In other words, this feels more like a statistician pissing contest than it does a nugget of enlightenment for Pacers fans.
In his short response to Hollinger, Winston may have used better gravitas to state his case than did John, who I imagine many Pacers fans think overstated his case when he brought the word “worst” into the equation. But, as with most all of us who talk flippantly on the internet about the NBA, Winston isn’t exactly above being a little overly blunt himself.
Here are some of his previous thoughts on Murphy:
In an interview with True Hoop I said Troy Murphy had done little to help his team. I apologize. I should have said that with Golden State he had done little to help his team (average rating of -8 points in his last 4 years with Golden State). But in the last two years with Indiana he has improved to the level of an average NBA player(which is very good indeed). Kudos to Troy and the Pacers coaching staff for this improvement. It is rare to see such an amazing improvement during the middle of a player’s career,
There is a good deal of sarcasm in there if you didn’t notice.
Basically, Winston’s preferred adjusted +/- system of analysis finds that Troy Murphy is very bad. John Hollinger’s preferred PER system of analysis finds that Troy is pretty good. Winston’s adjusted +/- finds Brandon Rush to be decidedly average. Hollinger’s PER finds Brandon to be very bad.
Ultimately, what we really have here is John Holinger saying, “Statistically, Brandon Rush sucks.” And Wayne Winston’s response is, “Yeah, well, statistically, Troy Murphy sucks more.”
No offense to either man, both of whom I think are invaluable basketball minds to the NBA community, but we’re not exactly looking at a Plato’s Republic-level debate here. We just have two guys pointing out what their preferred systems of statistical analysis says about two guys on the Pacers.
I love advanced stats and think both PER and adjusted +/- belong in the discussion of player’s worth (and, please, let’s use some of these numbers when talking about Brandon and Troy in the future). But if all we were going to learn here is that Brandon Rush hasn’t played that great over the past 75 games and that the Pacers get smoked when Troy Murphy is on the floor …. well, I could have told you that a lot more simply just by saying:
“Yes. That has definitely happened.”
{ 0 comments }
