clock menu more-arrow no yes mobile

Filed under:

Monday Discussion Topic: Lies, Damn Lies, and...?

The brief respite in the schedule allows us to return to more theoretical conversation once more.  Today's discussion topic is one I've thought of several times before, but the recent article by David Thorpe at which states that Adam Morrison has the lowest PER (a John Hollinger statistic measuring the overall value of a player) of anybody averaging 30 minutes or more, while at the same time asserting that this really isn't indicative of his value, finally tripped the trigger.  How valuable ARE statistics in analyzing the game and its players?

There's no doubt that the Sabermetrics movement in baseball has caused a ripple effect in other sports.  Call it Moneyball or whatever you wish...the general idea is crunching the numbers and watching them spit out the best available player for your team by whatever criteria you set, and a lot of people are buying in.  I must admit that I, myself, do a fair amount of number crunching in preparing material for this site.  I'm familiar with Hollinger, with, and of course with your garden-variety box scores and season statistics.  At the same time I must admit some discomfort with the new wave of statistical worship, as if the spreadsheet were the boat and we're just skiing in its wake, content to go where it takes us.  Even as I read the stats, I have a healthy suspicion of them.  Perhaps it's the memory of J.R. Rider's completely useless 20 points per game.  Or perhaps it's from reading one too many message boards where people say, "The stats say this, END OF ARGUMENT!"  Is it really?  Isn't at least a small rebuttal in order?

To begin with, I'd argue that basketball, while as statistically prolific as any other sport, is less susceptible than most to absolute quantification by those statistics.  The late-90's Trailblazers were living proof that you can put five 20-point scorers together and they don't necessarily equal 100 points.  Every national team for the last decade has also proven the point amply.  The lowest bench polisher on the American squad this year would have a higher PER than the best player from Greece were the latter transported into the NBA.  As a whole the U.S. squad would have eclipsed them like Refrigerator Perry sitting on Tattoo.  What happened on the court?  It can't be anticipated, or in some ways even quantified, by the statistics.  And this was not a fluke, a probability that every stat leaves room for.  This has happened tournament after tournament.  It's not that the numbers aren't valuable...they are.  But they're hardly the unquestionable driving force and in some cases they're outright irrelevant.

I am doubly suspicious of the "cocktail" statistics out there, the ones designed to measure general ability and to rank dissimilar players.  (At the same time I must note that I respect and admire the people who come up with these formulae.  I certainly couldn't.)  For the most part, the less specific a statistic is the less useful I find it.  For one thing, I probably don't need a formula to tell me that Tim Duncan is among the best players in the NBA.  If I don't know that from the tip-off, I'm in trouble.  For another, just because your system tells me Dwayne Wade is slightly better than Lebron James who in turn is better than Kobe Bryant doesn't necessarily make it so.  Which player is better?  I'd argue it depends on your needs.  And the only way you can ascertain if those needs are being filled is by specific stats, not overarching ones.  Having Zach Randolph as the third best power forward in the league doesn't mean doodly if you needed a defensive stopper at that position.  The same goes for game planning.  If you tell me Randolph makes 62% of his shots when he goes to his left while facing up but only 41% while going right I can do something with that.  What do I do with the broader score other than to say, "Hey!  This guy is good!"?  (Again, if I don't know that already I should get another job.)  Besides, there are plenty of times, as in the case of Morrison, when the broader stats are just plain wrong, or at least misleading.   I suppose you could argue that one can find diamonds in the rough this way.  I'd agree if by that you meant "players that we might not have noticed that we should take a second look at" but it's the second look, not the stats, that make the determination, just as it's the play, not the numbers, that forms the basis of reality.

In short I'd say that stats make a very good servant, but a very poor master.  Or put another way, they're useful as a basis to start a discussion, but probably not exclusively valid as a way to end it.  They provide a different angle on the whole, but do not encompass the whole itself.  It's unwise to ignore the numbers, but neither is it wise to post a PER number and say, "End of story."

How do you view NBA statistics?  Are they more or less binding and illuminating than those in, say, MLB or the NFL?  Which stats do you find the most telling?  The least?  That's the topic of the day.  Weigh in at any length you desire.

--Dave (