Wednesday, May 7, 2014

The Hibbert Space Problem in Atlanta

The popular term now is "basketball analytics" -- it's all about numbers and statistics and advanced metrics. A word that's used far too often, however, is scientific. Scientific basketball is the ideal. It's about testable theories that are verifiable and reproducible. If you think that age matters in the playoffs, then prove it with real numbers with open methodology that can be tested by everyone else.

A few days ago, I wondered why Indiana didn't play Hibbert whenever Brand was on the floor. If Hibbert being stuck on the perimeter is the problem, then stick him on the one guy without three-point range. I was surprised by how infrequent this simple action was, but luckily Indiana played two games versus Atlanta in the regular season where Antic was unavailable and non three-point shooters Ayon and Brand played center (these were the only two regular season games Indiana won, and it seemed like a harbinger of things to come.)

The Hawks are a test lab of sorts for five-out basketball where everyone can shoot and the results were pretty terrifying for the Pacers despite the lack of star power on the other side of the court. For another site, I started a series on the power of spacing -- that outside shooting has a significant effect on defense even excluding the added benefit of efficiency. This is more pronounced with big men. Why is that? You're drawing out interior defenders far away from the basket, which they're supposed to protect. Hibbert, for instance, derives most of his value from guarding the basket. Think of this: besides his rim protection, what does Hibbert do? He is not above average in any other area with the exception of FT% for a center (for his career he's near league average in offensive rebounding compared to centers.) If you take away his rim protection and put his 7' 2" frame on the perimeter where his defense can be shaky, he is much less valuable.

Testing this, I collected all the play-by-play data for the Hawks and Pacers this season, and calculated the offensive efficiency of both teams given three situations: when every big man is a floor-spacer like Pero Antic, Paul Millsap, and Mike Scott; when there's at least one big man without three-point range like Elton Brand or Gustavo Ayon; and overall.

The defense is worse with Hibbert, which isn't surprising to anyone who saw the series, but that's true of every situation. This is a little alarming when you consider Hibbert is mostly aligned with his other starters and not the bench players, who are far worse defenders -- Scola, Turner, et al. Mahinmi was a better defender for the Pacers against Atlanta, but when looking at how the Pacers fare with the two centers in similar lineups this was true for the entire season.

Table 1: Atlanta’s offensive efficiency (points per 100 possessions)

With Hibbert
Without Hibbert
Difference
Overall
103.3
100.4
+2.9
3PT center
104.8
102.5
+2.3
No 3PT center
100.0
97.5
+2.5
*Note: it's not an error that the overall difference is greater than the other two categories because the with and without Hibbert columns have a different number of possessions for each category. 

Yet Indiana still plays better with Hibbert when a non-3PT shooting threat is on the court. Why is this? Indiana's offense unexpectedly takes off, and it's better with Hibbert. However, when there are three-point shooters in the game, their offense is much worse when Hibbert is on the court. I'm not sure why this is, since Brand is still a good defender, but so is Antic. This is also a case of noise with raw +/- data washing out any useful information. Or maybe, Hibbert struggling against Antic or Millsap translates to frustration on the other end of the court.

Table 2: Indiana’s offensive efficiency (points per 100 possessions)


With Hibbert
Without Hibbert
Difference
Overall
97.0
103.6
-6.6
3PT center
88.7
99.4
-10.7
No 3PT center
113.9
109.0
+4.9

It's still curious the Pacers didn't maximize the time coinciding with a non-3PT shooting threat for Atlanta and Hibbert -- that happened just 38.5% of the time, and excluding the two games Antic missed it's only 23.5%. They did play a bit better with him on the court guarding Brand or Ayon. The numbers state Indiana's defense with Hibbert isn't affected by the outside shooting, but it's hard to reconcile that with what happened on the court. As basketball becomes more scientific, we don't dump these results and hide the confusion; we embrace the surprises and try to find answers. Perhaps we need to adjust for something else, like Scola-Turner playing with Hibbert against Brand more often. Perhaps Hibbert isn't as bad defending the line as we thought, and the problem is that he's just flat-out worse overall. The struggle to answer this question, the journey for knowledge, is the path for enlightenment.

Whatever's happening to Hibbert, it's all-encompassing. And it's destroying Indiana's season.

2 comments:

  1. This comment has been removed by a blog administrator.

    ReplyDelete
  2. Great stuff, enjoyed reading as always.

    ReplyDelete