2012 Assist Infographics: Point Gods, Guards, and Quetzalcoatls

So, earlier today, the fantastic folks at Basketball Reference released to the public a marvelous database. It includes a highly interfaced and searchable play-by-play database for the past 10 years. This is, quite frankly, incredible. There's virtually nothing your heart desires that you can't search for -- player performance by the time left in the game, detailed stats of what happens when players make certain plays, team performance in certain situations... I don't think I'm being overly sycophantic when I say that this is among the greatest single advances in easily interfaced, searchable, and public statistical databases in the world of NBA statistics. There may never be any one or two people who mine the database for all the insight you can get from it. Ever. Long story short, it provides an easy way to answer certain questions, and the ability to learn how to raise better ones. To that end, I'll be doing a series of posts where I graphically demonstrate certain things that this database allows one to easily find. I hope that these will be useful to you. They're certainly interesting, if nothing else. Today's introductory topic as I sift through the data for interesting insights: how are assists distributed among the league's best setup men? Who are their most prolific partners? When in the game do they get them -- and what's the score when it happens?

Interesting questions. And it's easier to scratch at answers than ever. Let's dive in.

 • • •

For this post, I'm producing graphs for the top 15 players in the league in AST% as of all games played on the 14th of February, 2012. Valentine's day! Because we love assists, here at the Gothic. For the data, you can go to our spreadsheet (or just search Basketball Reference yourself, obviously). I'm going to abstain from commentary in the post proper, because these graphs took a while to make and I think they speak for themselves quite well. I will commentate in the comments, if anyone wants to start a discussion. A few definitions, in case it's not immediately obvious. The charts are ordered 1 to 15 by assist percentage (or, the percentage of shots they assist when on the floor). Assists by margin refers to the margin of the game when the assist was made -- in theory, the "ideal" point guard would have more assists in a close game than in a game broken open. The assist target pie chart refers to, as one might expect, the players who completed the most baskets that the point guard (or the only non-point on this list, LeBron James) assisted on.

An interesting aspect (and one that I highlighted by using the team's secondary color to emphasize) of point guard play is what percentage of assists came from players outside their top 5 targets -- or, how many of their assists came from (usually) players they share less time with. Ricky Rubio is the best guard at taking advantage of the "others", with more assists coming to players outside his top five than any of those five individually. On the other end of the spectrum, Kyrie Irving has only registered eight assists outside his "main" five targets -- partly a reflection of the dismal quality of those others at finishing plays, perhaps partly an indicator that Kyrie's chemistry with the outer fringes of the Cavaliers pales to that of Ramon Sessions. The why is not our object, here, only the what. Assists by quarter is relatively self explanatory, and at the bottom of each infographic, you can see a demonstration through bars of how many of that guard's assists came on three point shots as opposed to two pointers (one area where Deron rules all comers).

All that said, here are the rest of the graphs. Enjoy.

 • • •

Hope you enjoyed this trip through the seedy world of "hand-making infographics in Photoshop". For my next trick, I'll pass out in bed! Adieu, readers.

11 comments on “2012 Assist Infographics: Point Gods, Guards, and Quetzalcoatls

  1. Pingback: Rants, References and Revelations | Hickory-High

  2. Does basketball-reference have any data around where the player was on the court compared to the type of assist? I would be curious to see how a players ability to get into the paint effects the type (3pt, short 2pt, mid range, etc...) of assist. Also, it would be interesting to see how many assists come off of fast breaks vs pick and rolls (or pops).

  3. Did you somehow script photoshop or you went the manual way all the way? Those are some nifty looking inforgraphs. While looking through I thought you had for sure scripted them with gnuplot, matplotlib or similar software package. Whatever the case, I commend you for your perseverance.

  4. Pingback: The Point Forward » Posts Court Vision: Latest news in the NBA «

  5. love this.

    thoughts:

    *steve nash is good. the distance between him & #2 is greater than #2 to #8 (or #9, the rose graphic didn't render for me).

    *totals all dip in 4th quarter (approx.) ... fewer mn b/c games are already decided, fewer easy shots... what else?

    *the by-quarter totals for lebron seem way off (they show 275+ assists) ... is the trend line - i.e. most assists in first quarter, fewest in second - still representative?

  6. @ Sam: That isn't available on B-R, however, that kind of data is available on Hoopdata. As an example, look at the last table on Steve Nash's page. The data entry process was tedious enough as is, so I didn't incorporate other data sources into the graph-making. As for the last part, you'd need synergy for that. I have access, but it's such an aggravating tool to use (and tends to crash my computers) so I definitely didn't want to look into that for these guys.

    @LasEspuelas: I'm not all that good at matplotlib, though I've used it before. I figured I could probably do these by hand in the time it'd take me to learn the code to generate what i wanted properly. I went manual all the way on them, and I admit that it was definitely more trouble than I expected, but I think the output was good enough to be worth it.

    @trevor: Yeah, Nash is having an insane season when it comes to generating assists. Though I wouldn't sleep on Calderon's excellent season on that end either, or Tony Parker's. My assumption would be that were I to generate graphs of minutes per quarter (not available on the Play Index +, though it'd be a good thing for them to add) they'd explain some of the generally low totals in the 2nd and the 4th. Most coaches rotate the starters out so that they naturally play fewer minutes in those quarters, but add the blowout effect and the low fourth quarter numbers become more reasonable. And as you noted, fewer easy shots is a good explanation as well -- coaches tend to try and shut down ball movement more as the game goes on. As for LeBron, good catch. When I get home next week I'll move that down -- I put the graphs there by lining up the gridlines, but it looks like I just lined them up one too high. The trend line is still representative, just with 25 less assists at each point.

    Thanks for the feedback / discussion, everyone.

    • Aaron, the output is fantastic. It is so consistent that I seriously thought it was scripted and I wondered why you stopped at 15. Props on the hard work and the wonderful result.

  7. Pingback: Corporate Knowledge: February 20, 2012

  8. Interesting analysis and GREAT graphics. I have never seen a detailed breakdown of which players are the recipients of a given plaeyr's assists. This suggests quite a few topics for further analysis, such as comparing the shooting percentages of the recipients when assisted by this player versus all of their assisted shots or the general shooting percentage. The one metric on these charts whose value I question is numbers of assists by point margin at the time the assists were made. Those totals ought to be normalized using the amount of time the player was on the court that their team had a point margin in the respective range. The reason is that the total counts by point margin may well reflect other attributes of the team, such as defense, that affect how many opportunities are available to make assists in a given margin range. I suspect this to be the case when I consider the teams of the players with monotonically descending assist margins, versus the teams of those with assist margins that go down for 6-10 then up sharply for >10.

  9. Pingback: Top Passers in the NBA – Using Assist % « The Basketball Quant Project

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>