Data Mining: The great defensive efficiency divide

Almost two months into this NBA season, it has become exceedingly clear just how important the Denver Nuggets bench has been to the team’s relative success. The starting lineup has frequently struggled, especially in first quarters, and the second unit has come to their rescue all too often.

In my recent post on Wilson Chandler, I compared his net efficiency with that of Timofey Mozgov, suggesting that Timo was, and Wilson wasn’t making a big impact. In the comments section, a Roundball Mining Company reader correctly pointed out that it might not be fair to compare a starter with a bench player, given the discrepancy in the caliber of players they’re going up against.

So to follow up on this notion, I used statistics from to put together the following chart:

I used on-court versus off-court data for this, but the points per 100 possessions column is an approximation of net efficiency. And as you can see, Chandler has the best numbers among the starting five, and the commenter was correct: Instead of comparing apples to oranges, I should have compared apples to apples.

Even so, the stark contrast between the starters and bench, not to mention the fact that, among other oddities, Ty Lawson is certainly not the worst player on the team, caught my attention. The eyeball test would tell us that Lawson is clearly the most important player on the Nuggets roster, and the offense generally falls apart when he’s not in the game. “Sure,” many Nuggets fans might say, “Ty’s defense might be a liability, but he more than makes up for it with his offense.”

However, looking more carefully at the numbers (you can check this spreadsheet to see them), a pattern emerged which suggests that it’s really all about the defense. For example, with Lawson on the court, the Nuggets have scored 107.7 points per 100 possessions, but this only drops to 106.9 with him off the court – less than one point. Nuggets opponents, on the other hand, have outscored the Nuggets with Lawson in the game, scoring 110.1 points per 100 possessions. But their scoring dramatically plummets to 95.7 with Lawson off the court – a decrease of a whopping 14.4 points!

This pattern of limited on/off-court differences in offensive efficiency contrasting major differences in defensive efficiency is present in nearly all Nuggets players’ numbers, differentiated mainly by the fact that the bench players have a positive, and the starters a negative defensive impact.

So the fact that it’s defense more than offense that divides the two units is clearly evident, but we still haven’t resolved the problem of comparing apples to oranges.

Using stats from, I have attempted (hopefully with some success) to make a more apples to apples illustration of just how bad the starting five’s – and just how good the bench’s – defense has been:

The first two columns show each player’s defensive efficiency and his rank among the 285 NBA players with 15 or more minutes per game. As you can see, all bench players not named Jordan Hamilton are in the top 20, while the entire starting unit is in the bottom 70.

But it’s true that the starters are going up against tougher competition, and that some of the NBA’s weaker benches might be easy meat for the deep Nuggets bench. So how do they fare among their own? How do Denver’s starters stack up against the league’s starters at the same positions, and how does the bench fare against other bench players?

Unfortunately, pretty much the same. The five columns on the right represent each player’s percentile in defensive efficiency among similarly situated players. So, for example, as of collecting this data there were 49 guards in the NBA who started 15 or more games this season. Among those 49, Lawson is ranked 45th in defensive efficiency, and Foye is ranked 46th, locating them both in the bottom 10th percentile among starting guards.

So even when we compare apples to apples, the picture doesn’t get any better for the Nuggets starters’ defense. All five are in the bottom 20th percentile, including three in the bottom10th.

And no credit should be taken away from Denver’s bench for facing “easy” opposition. Darrell Arthur may have a well-deserved reputation as a great defender, but Andre Miller, who has the 6th highest defensive efficiency among the 52 guards coming off the bench in the NBA? Dre’s defense has traditionally been so bad that lamenting it had nearly become a cottage industry here at RMC, and now he’s a key component of one of the league’s elite bench defenses?

Clearly, what the bench is accomplishing is a fantastic team effort. As David wrote about in his recent post on Nate Robinson, this was a point of emphasis that Tom Thibodeau “drilled into” their heads in Chicago, and perhaps Nate-Rob deserves a big chunk of the credit for bringing that mentality to Denver’s second unit which is, in fact, performing at an elite level defensively.

If defense has become a recurring theme here at RMC, it’s because it is currently the team’s greatest weakness. In my view, the data in this post is further supporting evidence for Matt’s recent proposal to change up the starting lineup. So check that out in case you missed it, and also see David’s post on switching for more analysis of the Nuggets’ defense.


You can follow me on Twitter here: @denbutsu


The following two tabs change content below.
Joel is a long time Denver Nuggets (and Broncos) fan from Colorado who's been living in Japan since the mid-90s, and blogging about the Nuggets since 2008. You can contact and follow him on Twitter: @denbutsu.
  • heykyleinsf

    Good to know this stuff Joel. Something I said about the grades after OKC loss..

    heykyleinsf • 4 days ago −
    IDK David.. you seemed pretty generous to Foye and mean to Robinson… Foye played 24 minutes and did what again? Nate at least did something even if one dimensional.. But maybe I’m just biased against Foye… who is to me.. by far.. the least impressive starter we’ve had all year. Don’t understand or like that he is a starter for us. I guess I’ll keep looking for good in him.

    Why does Foye still start?

  • Kyle Wurtz

    awesome analysis! couldn’t agree more that this complete lack of defense in the starting lineup has a lot to do with the fact that the team struggles to win first quarters…

  • MacKenzie Pantoja

    I think 2 man lineup data is something that is quite underrated. Individually, Lawson has horrible differentials. But, if you isolate the data, you’ll find Arthur+Lawson is equally effective as Arthur+Miller defensively (though still worse than Robinson+Arthur), and Mozgov+Miller or Robinson is only about 2 points more effective that Mozgov+Lawson. The difference, though, is that Lawson always shares the court with Hickson and other bad defenders. The fact that Lawson is always on the floor with Hickson, Faried, and Foye is out of his control.

    In short, it’s not as if all Nuggets starters are in the bottom 20th percentile for their position. It’s just that their numbers are dragged down by the fact they spend so much time with defenders in the bottom 10th of their positions.