Pace Vs Possessions Used – One Man’s Stat Debate

by

After generating pace-adjusted stats for our inter-generational NBA stars comparisons, I’ve grown suspicious of the value of the exercise and even the nature of possessions-based metrics. I understand the need to break the game down into equal sample sizes, and that a possession is our best option in basketball, where there’s no such thing as an at bat. And as we saw in the ‘60s players conversion, when there is a significantly larger available pool of possessions, it can dramatically impact a player’s output – see Chamberlain, Wilt and Robertson, Oscar who quickly morphed in Shaq and Wade in short-shorts when we made the pace adjustment on their stats.

That said, I’m unconvinced that looking at the proportion of possessions tells us anything about a player’s production compared to another who plays at a different pace. Let’s take a common comparison like Larry Bird vs. Dirk Nowitzki as an example. For his career, Dirk’s Usage rate (the percentage of his team’s plays where he ends the possession via shot or turnover) is about 27. So 27% of the time that he’s on the floor, he makes the play to end a possession. The used possession calculation shows that he actually ends about 21 possessions per game. Bird’s Usage rate is 26; so he makes the play 26% of the time for his team. Bird actually ends about 24 possessions per game.

So while Dirk uses a higher proportion of his team’s possessions by 1%, Bird makes 3 more plays per game. The going theory is that from a simple production standpoint – not factoring efficiency yet – Dirk’s contribution is more valuable because it consumes a larger slice of the possession pie, thus keeping his non-Dirk teammates from taking shots and messing everything up with their non-German lack of touch. I can understand that line of reasoning. BUT, and this is important, Bird still does more. He actually uses more possessions. Are we to assume that if Dirk played in a faster environment, his usage would scale up and he’d continue to provide 27% of his team’s playmaking? Logic dictates that there’s a breaking point to this sort of increase. We’re working towards limits, not infinites.

This has always bothered me somewhat when comparing stars. Because Rick Carlisle can take the air out of the ball and filter a healthy percentage of his team’s shots through Dirk, that doesn’t mean that Dirk could generate that same percentage of offense while playing in a faster style where ball-distribution will naturally be more diverse. Conversely, because Bird manages to use 26% of possessions on a fast team, does that mean he will use that same percentage on a slowed down team where the coach would want to funnel shots to its best shooter and passer? Is it more likely that he uses the same total number of possessions or the same percentage of possessions?

From my swivel chair, I’d say the former is more likely. In fact, I think Dirk himself is a decent example of this:

2003: 19 FGA, 7 FTA, 2 TOV, Team Pace 92.4 Usage 27%

2006: 19 FGA, 7 FTA, 2 TOV, Team Pace 87.8 Usage 30%

Nash and Don Nelson leave. Pace changes drastically. Dirk’s Usage% changes to compensate. His used possessions remain identical.

Obviously I didn’t choose those seasons at random, but you get the point. It’s simply more reasonable, in my estimation, to believe a player can maintain a certain level of production than to believe he will scale his production up to match the pace of his team.

Pace cannot be ignored in three categories: rebounds, steals, and blocks. The number of the opposing team’s available possessions does have a determinant impact on the total steals a player has the chance to nab. The number of field goals the opposing team shoots is the number of opportunities a player has to record a blocked shot. The total number of missed field goals and possession-ending missed free throws an opposing team generates is the pool of available rebounds to be snatched. Pace, while not the only factor since opponent shooting percentage and free throw rate come into place, is critical in these statistical columns.

Where pace seems to me to be less critical is on the playmaking side. Players are who they are on offense. Michael Jordan led the league in scoring and in usage most of his career, despite playing at wildly different paces. The year he averaged 37 points per game was not the fastest paced or his highest minute total, just the chuckingest. If you moved him to a slower era, would he have scored less because the Bulls had fewer possessions or just shot a larger proportion of the time? I don’t really know, but he had clearly checked his conscience at the door when he left the Boston Garden after the previous playoffs.

Magic Johnson played in the hyper paced Showtime era and then a more slowed down post-game system later on, and his production shifted somewhat, but he always got up about 12-14 shots from the field, 6-7 free throws, dished out 11 or 12 dimes, and lost three or four turnovers. That’s Magic whether he’s running the Showtime squad or the big post-and-toast group of ’90 and ’91. Players do get better or worse with age. Efficiencies dip and spike. But what a star player gives, while it varies to a certain extent, is basically the same performance from year to year, influenced of course by team need.

The other issue I have with scaling offensive production with pace is the idea that efficiency could be constant. That doesn’t necessarily make sense. Playing faster might mean more layups or it might mean more rushed jumpers. Playing slower might mean more time to run good plays or more time spent pounding the ball aimlessly. I’d prefer not to make any changes to scoring, assists, and turnovers unless I have to – as with the ‘60s when certain players used more possessions than would even be available to their modern counterparts.

So, I’m working now on some data using pace to adjust ORB, DRB, STL, and BLK stats while leaving PTS, AST, and TOV alone. I’m not sure how it will come out, but I think it may be more representative of who players really are and how they function than adjusting everything has been. It seems like a fair approach, and I’ll share the results as soon as they are complete.

Advertisements

Tags: , , , , , , , , , , ,

7 Responses to “Pace Vs Possessions Used – One Man’s Stat Debate”

  1. NBA Players Top 40 List: 1980-2013 « Double Dribble Says:

    […] Shoving Basketball Knowledge Into Your Dome. « Pace Vs Possessions Used – One Man’s Stat Debate […]

  2. Neil Paine Says:

    I touched on this same idea in the BB Prospectus book from a couple years ago (2011-12 I think)… I found that, indeed, teams that play a faster pace tend to be more egalitarian in the distribution of their possessions, which supports your theory. IIRC, though, the impact of this wasn’t very big. The correlation between the go-to guy’s Usg% and his team’s pace was something like -0.10 or weaker.

    • jpalumbo Says:

      Damn it I miss your blog! So you mean that for every change of Pace +1 there was a -0.1 correlation in possession usage?

      It’s so difficult to isolate any factors in basketball. I’m looking at the last 33 years worth of league stats, and here’s a preview: teams shoot more 3 pointers than they used to.

      • Neil Paine Says:

        I miss it too. 😦

        -0.1 was just the correlation (remember, correlations run from -1 to 1, where the further from 0 => bigger relationship between variables). I don’t remember what the regression coefficient was (aka the slope of the line). That’s what would tell you how much a change in pace would affect the usage of your go-to guy.

  3. jpalumbo Says:

    Ahhh….

    I was going to tell you to quit sassin’ me, boy, but then I realized that I deserved your math lesson, and settled down.

    Let’s do a podcast this weekend.

  4. jpalumbo Says:

    Love that idea! I won a drink in my high school cafeteria on NBA trivia. I also completely screwed up my 7th grade percentages math class because the teacher gave us a bunch of Michael Jordan stats to parse out, and I knew all Mike’s numbers by heart. They thought I was a savant for a few a seconds.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: