Joe Dumars and the eye test are failing the Pistons

Joe Dumars isn't using statistics or metrics

If you have kept up with the "moneyball" trend in basketball that's sweeping the NBA -- there's even an annual conference about it now -- you probably would have guessed that the Pistons have been slow to jump on this bandwagon.

However, when Tom Gores purchased the team, there were subtle hints that just maybe this would change. There were reports that Gores was bringing in his own staff who were trained in some sort of analytics to advise Joe Dumars. Lawrence Frank and his staff utilized analytics. Phil Jackson (one of the only coaches who has been demonstrated to impact the statistical production of players) was brought in to help advise Dumars on "coaching in general," and he supported the Cheeks hiring.

For those of us who were looking for it, there appeared to be some hope that things might change. Even if Dumars remained a "just watch the games" guy forever, maybe Gores could find a way to surround him with people who would give him solid analytical advice.

If that's the strategy, Dumars isn't buying in.

David Mayo of MLive.com  recently fielded a question from a reader asking to what extent the Pistons have embraced statistical analysis as a tool to inform decision making.

The long and short of it, they aren't.*

Go read the whole thing if the topic interests you; here are the important bits:

This is largely a top-down approach and the Pistons have a couple of old-school guards in the top chairs.  I've never heard Joe Dumars utter the word "analytics," much less apply one.  Maurice Cheeks, asked about advanced statistics earlier this year, responded with something about 3-point percentage.  They govern by eyeballs.

[...] Having a staffer tuned in to analytics is advisable, but whether it warrants more than a subscription to Synergy Sports, Sports VU and other statistical clearinghouses, and that one "Rain Man" cat who never forgets a statistic and can slice the numeric clutter and find that game-winning nugget, is debatable.

I want to zero in on one statement: "they govern by eyeballs," and I want to suggest that attempting to manage and coach an NBA franchise in today's NBA -- where many of the most successful franchises are turning to analytics -- by using your eyeballs and only your eyeballs will fail.

And I'm going to start with some Psychology.  Stay with me, I promise there's a point.

What we're learning about our eyes, our brains, and our memories

In undergrad, I studied Psychology and Sociology. I took a class on memory (fittingly, I can't remember the name!), and we were assigned a book called "The Seven Sins of Memory: How the Mind Forgets and Remembers," by Daniel L. Schacter. I'll spare you a full literature review (Wikipedia has a really good summary); the important application of Schacter's research for sports is that human memory is malleable, that is it can change over time, and in a variety of ways is very unreliable.

Here's an easy example. It's a memory test that takes about a minute and a half to complete.

How'd you do? I consider myself to have pretty good memory, I tried really, really hard to do well, and I correctly answered only one of the five questions.

Here's something a little more interesting. Note that you'll be watching a classroom of law students, a group that is generally very intelligent and that generally has "good" memory.**

Here are the two points I want to tease out.

First, our perceptions through our eyeballs are not always as accurate as we think they are. Second, our memory of what we perceive through our eyeballs is almost always less accurate than we think it is.

That isn't a criticism of Joe Dumars, me, you, or anyone else. It's simply an observation about we creatures called humans. Unless you're some sort of savant, you have eyes and a brain that play tricks on you, and you shouldn't trust either one completely.

Let's have our own thought experiment

Now, on to basketball.

In the spirit of science, let's have a quick thought experiment. Imagine yourself as an old-school General Manager in the NBA, like Joe Dumars, in a practice gym that's empty, except for yourself, your team of scouts, and two players you're thinking about taking with your next lottery pick. You have determined that your team needs to add perimeter shooting in order to improve, and you want to assess these players as shooters.

In order to do so, you've set up a sort of three-point completion. Each player will attempt 1,000 three-point shots. There is no time limit, and each player is alone on one side of the court. There are five points at which players will shoot two hundred shots each: the corner (on each side), the extended elbow (on each side), and the top of the key. The players will be shooting simultaneously (for the purposes of the "experiment" to simulate players playing simultaneously in a game).  For giggles, let's say the two players are Kentavious Caldwell-Pope and Trey Burke, who were both respectable three-point shooters in college.

Obviously, there are important things you'd be looking for that you can only see with your eyes. You might be interested in whether the player is a better set shooter or pull-up shooter, so you might split those types of shots 50-50. You might be interested in elevation, arc and point of release, because NBA defenders contest shots more stringently than NCAA defenders. You might also be concerned with things like mechanics and fatigue. You'd be right to point out to me that you can't pick those things out from a stat sheet.

But, using only your eyeballs and your memory, how would you know who the better shooter is at the end of the drill? You might have some subjective feelings about whose jumper is smoother, which player has better mechanics, and so on. But, would you know any more than that?

Without one of the most basic statistics in basketball, three-point field goal percentage, there is simply no way that you could. Without someone on your staff writing down makes and misses, there's no way you could make an informed judgment. One thousand jumpers is a lot of jumpers; there will be far too many makes and misses for your eyes to see or your brain to remember. And on top of that, you'd be trying to assess two players at once.

There are a couple of objections you might make here. First, this is a simple example. Second, field-goal percentage has been around for basically forever (although given Dumars' history in free agency, it's debatable he pays attention to it), and it's not an argument for any type of "advanced" statistical analysis.

Yes, it is a simple example, but that actually underscores my point. What is happening in the gym in our experiment is actually much more simple than what happens in any given NBA game. If there is ever a time where we could completely trust our eyeballs and our memory, it would be a situation like this. To your second object, I actually agree, so, let's look at two recent NBA examples.

Comparing the eyeball test with the simplest "advanced" statistics

Sticking to the theme, let's look at Trey Burke's recent performance against the Pistons. In the recap of that game, I praised Burke's ability to orchestrate a five-man offense, and I lamented that the Pistons have gone far too long without such a player. That was my subjective eyeball assessment, and even a simple box score perspective seems to support it. He had a double-double, a rarity for a point guard, with 20 points and 12 assists, and he even had two steals and a block. Plus, he shot almost 50 percent from the field -- that's good, right? Those seven turnovers were ugly, but a lot of them came early, and he settled down when it mattered. Predictably, fans everywhere saw the good things I saw, and coupled with KCP's disappointing performance, were calling for Dumars' head.

But, was his game really all that great? Did our eyeballs get it right?

Let's use one of those advanced stats -- points per possession -- to find out.

In all levels of basketball, a team needs to spend possessions in order to get points. This is an idea that is widely accepted by analytical types everywhere. Possessions in this sense are a kind of currency. Highly successful teams can be thought of as teams who invest their money wisely and avoid wasting it. In other words, they try to get the highest number of points possible out of the possessions that they use, which means taking high percentage shots and avoiding turnovers.

Let's apply that to Trey's game. Instead of just counting up his points, assists, and steals and thinking of his turnovers as bad, let's ask how much Trey had to "spend" in order to "buy" all that.

It's actually pretty ugly. Trey used nearly twenty-five possessions to get those twenty points. That means he only bought his team 0.81 points for every possession that he used. As a team, the Jazz are scoring 1.024 points for each possession they use.

Here's another one. David Mayo had this to say about the Memphis Grizzlies:

Memphis got an analytics-happy owner in Robert Pera, Hollinger was hired, and Memphis quickly made a couple of big trades, most notably involving Rudy Gay, as if you need a statistical analyst to tell you Gay requires lots of shots and lots of cash.  Just watch the games and the bottom line and you'd know that trading Gay made the Grizzlies leaner offensively and fatter in cap space.

Leaner offensively, huh?

Offensive rating applies the idea of points per possession at the team level and tells us how many points a team scores for every 100 possessions used. In 2012-13 - the year Rudy Gay played 42 games before he was traded -- the Grizzlies scored 104.9 points per 100 possessions, good for 17th in the league.  So far in 2013-14, they are scoring 105.6 points per 100 possessions.  Mayo's eyeballs have fooled him. The Grizzlies' offense has been better since Gay's departure. Turns out that replacing guys who take a large number of shots but only make them at around the league average aren't that difficult to replace.

The eyeball test is failing, and it's time for a change in Detroit

Here at DBB, we've been telling similar stories about Joe Dumars and the Pistons for a long time now. Back in February 2010, Mike Payne wrote this gem (which I happen to think is the best piece ever written here) explaining the failed rebuild around Rodney Stuckey, Ben Gordon, and Charlie Villanueva. I've been singing the same tune, agreeing with Charlie Villanueva that it's insane to do the same thing over and over but expect different results. This summer, Mike predicted how a Josh Smith Pistons' era was likely to play out; he has been eerily correct.

We know what Joe Dumars' and Maurice Cheeks' eyeballs have told them, and we've seen the results. Using statistical analysis, we had good reasons to expect them. Josh Smith averaged 0.98 points per possession (ppp) last season; he is averaging 0.90 ppp this season. Greg Monroe averaged 1.10 ppp last season; he is averaging 1.21 ppp this season.

Of course, we haven't gotten it right 100 percent of the time. Like everything in science, statistical models are still adapting and evolving. I whiffed on Andre Drummond, for example, whose college production suggested he'd be an NBA bust. Statistical analysis struggles as much as the eyeball test when it comes to very young players -- especially very young, very big, very athletic players like Dre -- but we're learning from that.

Our eyeballs, brains, and memories aren't completely unreliable, either. Joe Dumars did build an NBA champion, and there's no taking that away from him. But as our last example, contrast the San Antonio Spurs' approach to retooling with the Pistons'. On the one hand, the Spurs are a franchise that employs a variety of statistical tools (most of which are proprietary and which they guard carefully). On the other, the Pistons are a franchise that spurns analytics. Both teams were built around a core of players of similar ages, yet one team remains a perennail powerhouse, while the other finds itself fighting to stay out of the lottery for the fourth consecutive season -- or even worse, missing the Playoffs and being forced to trade its pick because of the salary-dump trade sending Ben Gordon to Charlotte.

If you were Tom Gores, which future would you choose?

For me, the choice is simple. The future of basketball is happening now, and the Pistons need to make a change before it passes them by.

* Obviously, this is Mayo talking, not Dumars, so we have to use a little bit of caution here, but it seems reasonable to assume that the beat writers who are actually talking to Dumars and other franchise executives and employees on a consistent basis are at least somewhat in the know.

** Even though it goes without saying, I'll point out that the application to eyewitness testimony is almost infinitely more important than the application for sports. I get that.