The Usage Poster

It occurred to me that I had not done a cool visualization in a while.

Like this one from http://www.digitalsurgeons.com/ that's for a very good cause.

 

No more!

As you may have seen, I’ve been playing with Usage Curves and that work has led to some very cool stuff:

That’s the usage curve for every game played in the current season >24 minutes thru 3/26. I also did this :

That a tool that lets me look at team’s usage curves ( The example shown is Denver and every player that played 24 minutes for them in a game this season more than 20 times).

But the really cool thing I want to do today is a poster.

The Usage Poster. Enjoy!

13 Comments

  1. Tom
    4/2/2011
    Reply

    Arturo, can you explain why we should care about usage? And points per 20 possessions plotted against usage?

    In your previous post on usage you plotted WP per 20 possessions to show how it decreased with usage but that it had higher variance at lower usage. Isn’t this essentially a given based on the fact that WP includes the variables in usage but then adds all the others that mostly contribute positively to WP (average is ~0.1)?

    I’m having a hard time seeing the importance in it based on what you’ve presented so far, but perhaps the payoff is coming in an upcoming post.

    • Tom,
      The payoff is coming.

      The poster is about visualizing offensive efficiency. Some guys (like Dirk or Ray-Ray) are always consistent . Some aren’t. It’s interesting to look for inefficient strategies for teams.

      For example, Detroit should run the offense through Greg Monroe more. Mike Beasley Bad /Darko Worse & Love /Randolph Good. Denver had way more than enough firepower to replace Melo.

      • Rex
        4/3/2011
        Reply

        This is cool data, but to me, it seems too fast to use it to conclude, straightaway, that Player X should shoot more, and Teammate Y less. Seems like you have to first understand *why* these guys are shooting at the rates they are. Maybe Y is trying too hard to be The Man, but then again, maybe X is busting his gut on D, or there are limits to his game, such that ramping up his usage game in and game out won’t improve things. By itself, this data can’t answer that, but looking at some tape could. True, no?

  2. EntityAbyss
    4/2/2011
    Reply

    From what I’m seeing, it looks like usage has a small impact but as the shot attempts go up, so does scoring efficiency. Am I seeing this right?

    • EA,
      I believe that each player is unique. You want to look for optimization strategies.

    • Some Dude
      4/2/2011
      Reply

      my guess is there’s selection bias in here. Players shoot more when they are

      A. exploiting a mismatch
      B. “feeling it.”

      Problem is we don’t have big examples of guys getting 20 games at 15% usage and 20 games at 30% usage.

      Arturo – it’s easy to say to run the offense more though Love, but it’s hard to get a PF who doesn’t really take you off the dribble or post up against other legit bigs higher usage than the 22% he’s at. It’s just not very feasible.

      Kevin Love has been assisted on over 60% of his shots. Only 40% of his baskets are of his own doing. Why not figure in unassisted points per possession vs usage if you want to figure out who should have the ball more? Maybe the ball should be in Ridnour’s hands more?

      • Some Dude
        4/2/2011
        Reply

        almost forgot, how many of the 40% unassisted are rebound putbacks? Once you take those out, you’d see that it would be foolish to actually “run the offense” through Love, although it would be a good idea to run plays to get him wide open spot up shots.

  3. Tom
    4/3/2011
    Reply

    But the metric in the poster is points per 20 possessions. I would say that is approximately the antithesis of Wins Produced. Even if it is +/- and not actual points scored (not sure which it is), you have spent some time recently convincingly tearing apart +/-. What am I missing?

    • Tom,

      This is straight up Points scored by player vs possessions spent by player. This is very complementary to Wins Produced. It’s all about efficient scoring and being more efficient than the competition.

  4. Crow
    4/4/2011
    Reply

    What are the string of numbers in the middle and right part of the chart showing? What are the 5% bands across the top of each section? I am not clear on these things. Would you talk thru a full line of the table to help explain?

    • Crow,
      Fair questions.

      Let’s take Player #1 on the List, Al Horford.
      I’ve got Team,Name , Player average Usage for the games shown (all games >24 MP), # of games >24 minutes for the player and the average points per 20 possession (Possesions are equal to FGA + .44 FTA + TOV).
      Then comes the Average points per 20 possessions divided by usage group followed by the amount of games per usage group.

      So for example we see that Horford is mostly in the 15% to 20% usage rate with 27 games in that range and a points per twenty possesion average of 22.8.

      Hope that helped.

  5. Crow
    4/4/2011
    Reply

    Thanks, that makes sense.

Leave a Reply

Your email address will not be published. Required fields are marked *