Nylon Calculus: Charting the first week of NBA Free Agency
After a predictable postseason, the NBA offseason has provided some much-needed entertainment for basketball fans. There has been a slew of reported wheelings and dealings during the first week of free agency with many teams looking to shake things up in pursuit of the Warriors juggernaut.
All-Star free agents like Gordon Hayward and Paul Millsap have already announced their intention to relocate and some of the NBA’s brightest stars — Chris Paul, Jimmy Butler, and Paul George — were traded before free agency even had a chance to kick off. With all the activity, it’s been hard to keep up, so I’ve made some charts to illustrate how the landscape of the NBA has shifted to this point in the summer.
Grouping players by role
My ultimate goal for this project was to create two small charts for each team — one representing how the roster looked at the end of the 2016-17 season and one representing how it looks now, after the first week of free agency. I wanted an informative way to group players that would indicate each individual’s role on the team as well as providing some indication of his skill level. For my charts, I took a cue from Hannes Becker and abandoned traditional positional designations, opting instead for more empirical groupings.
For example, consider the variety of primary ball handlers in the league last season.
Players like LeBron James, James Harden, and Giannis Antetokounmpo are not traditional point guards, but they each possessed the ball for at least 5.0 minutes out of every 36 minutes they played during the 2016-17 regular season. We’ll use this criterion (5.0 minutes of ball possession per 36) to define our primary ball handler group.
To characterize the impact that players in this group have on their team, we can use the Regularized Adjusted Plus-Minus statistic generated by Jeremias Engelmann for the combined 2016-17 regular and postseasons. You can read more about the advantages of RAPM in this plus-minus primer, but here’s the gist of it: RAPM estimates the number of points per game by which a player’s team outscores his opponents’ team while he’s on the court, while at the same time making adjustments for the other players who share the court with him (his teammates and opponents). RAPM is a stat created directly from play-by-play data, in contrast to other plus-minus stats, like RPM or BPM, which are based on box-score data. Opportunely, RAPM can also be split into an offensive and defensive component.
In the plot above, solid blue bars represent a player’s positive contribution on offense and solid pink bars represent a player’s positive contribution on defense, so that Stephen Curry (ORAPM = +6.6, DRAPM = +1.0, RAPM = +7.6) and LeBron James (ORAPM = +5.7, DRAPM = +1.8, RAPM = +7.5) have the widest bars, representing the biggest positive impacts. Pink-striped bars indicate a negative DRAPM, which diminishes the player’s net impact, shrinking his solid blue bar (e.g., Kyrie Irving: ORAPM = +3.6, DRAPM = -1.4, RAPM = +2.2). I want to avoid getting bogged down with ranking players or evaluating RAPM as a metric. Suffice to say, I think it’s a pretty helpful stat and I’m going to use it as a shorthand for classifying: 1. Two-way players, 2. Players who are good on offense only, 3. Players who are good on defense only, and 4. Players who aren’t helping out at all.
Our next easily-delineable group of players are the rim protectors, which we will define as players who defend at least 5.0 field goal attempts within six feet of the basket per 36 minutes of playing time. By this definition, most rim protectors are categorized as traditional centers, but there are some important exceptions.
An increasing reliance on small-ball tactics forces players like Draymond Green, Kevin Love, and Al-Farouq Aminu to challenge shots near the basket, with varying degrees of success. Once again, we can consider the quality of various rim protectors by looking at RAPM:
Green, the Defensive Player of the Year, and Rudy Gobert, the DPOY runner-up, have the two highest DRAPM (+4.0 and +3.9) and they also have two of the biggest net impacts among the rim protector group (RAPM of +5.3 and +4.3).
Setting aside the ball-handlers and rim-protectors, we can use a player’s ability to space the floor with 3-point shooting as the critical attribute for further classification. We will define anybody who shot 5.0 or more 3-pointers per 36 minutes as a spacer.
Kawhi Leonard (ORAPM = +5.0) and Kevin Durant (ORAPM = +3.7) are reassuringly at the top of this wing group, offensively. On the other end of the spectrum, players like Victor Oladipo (DRAPM = +2.4) and Robert Covington (DRAPM = +2.8) play great defense, but struggle to pitch in on the offensive end.
The leftovers are a grab-bag of mostly bricky wings and offense-first bigs.
This is the group of players who is getting squeezed by the pressures of modern basketball and you can see the ceiling for this type of player is a bit lower — with Blake Griffin leading the way at +3.4 RAPM.
Charting last season’s rosters
Now that we have our player groups settled (ball-handlers, rim-protectors, 3-point spacers, and everybody else) and our preferred metrics for comparing skill levels (ORAPM, DRAPM), we can look at what gaps NBA teams should be looking to fill during the offseason.
The Warriors (surprise, surprise) had the most complete roster; it comprises one ball handler (Curry), two 3-point spacers (Durant, Klay Thompson), two rim protectors (Green, Zaza Pachulia), and one other guy (Andre Iguodala) who have a positive ORAPM and a positive DRAPM (individuals in the Top-15 for minutes-played on each team were considered).
This table has some useful info and you could go through it and pick out your favorite team and see what pieces they need to add in order to match up with the Warriors unfair collection of two-way stars. But, if you’re not into reading actuary tables, I’ve also created some colorful charts that relay some of the same information.
Below, each team has two small charts — one for offense (blue and green bars) and one for defense (red and orange bars). The bars are organized from left to right as they were in the table above — ball-handlers, 3-point spacers, other guys, and rim protectors. Each bar is a weighted average (by regular season minutes played) of the players in the group. The vertical axis represents RAPM with positive values above the white line and negative values below. The width of the columns represents the total minutes played by the individuals in each grouping. All of the charts in the figure are set to the same scale, so you can compare the size of the rectangles between the teams.
There is plenty to look at, but here are some things that stand out to me:
That tallest bar — the blue one — up in the upper lefthand corner, those are Golden State’s ball handlers (almost exclusively Curry).
Every team had at least one negative category, but Golden State’s profile comes the closest to going 8-for-8 on the positive side.
Houston had a very wide rectangle for 3-point spacers and very few non-shooting wings, as per the Moreyball ideal.
Teams like Memphis and New Orleans had the opposite approach — very few 3-point spacers and many non-shooting wings.
Cleveland’s poor wing defense is evident.
Russell Westbrook’s fantastic offense is evident, too.
The Lakers stunk.
With this baseline in mind, we can examine how teams have changed in the offseason so far.
Charting roster changes
The remainder of the charts will follow the same setup as the previous one, but I’ve changed the layout a bit so that the offensive and defensive charts are now side-by-side. I’ve sporadically highlighted where some key players fit into the groupings with callouts. I used the NBA’s Free Agency Tracker and Real GM’s Transaction History to keep updated on reported deals. I’m ignoring players who did not play last season, including recently drafted players. If a team did not extend a qualifying offer to a potential restricted free agent, if a team declined their team option, or if a team waived a player outright (at any point during the season), I removed the player’s RAPM contribution from the current team chart. If, however, a player is currently an unrestricted free agent with no deal yet reported, I left him on his old team for now.
Warriors: Two smart signings by Golden State (Casspi, Young) on the heels of five successful re-signings (Curry, Durant, Iguodala, Livingston, West) have kept the Warriors towering over their Pacific Division rivals.
Clippers: With Chris Paul headed to Houston, the Clippers are currently without a primary ball-handler (at least according to our definitions based on last year’s stats). Gallinari, Beverley, and Williams will add some nice offensive pop around the perimeter.
Kings: George Hill should help out in Sacramento.
Lakers: They still stink.
Suns: No activity to report yet.
Spurs: Pop et al. are quietly lurking.
Rockets: Houston doubled down on quality ball-handlers by pairing Paul with Harden. The addition of P.J. Tucker is a necessary improvement on the defensive end, but they shipped out some of their three-point shooting spacers in exchange for Paul.
Grizzlies: Ben McLemore and Zach Randolph are exchanging places, but it won’t move the needle much.
Pelicans and Mavs: Holding steady.
Jazz: Utah saved about $5 million per year by bringing in ball-handler Ricky Rubio and letting George Hill walk in free agency, but RAPM says Hill may have been the better option. The Jazz also lost a valuable piece when Gordon Hayward took a big chunk of their wing spacing rectangle to Boston.
Thunder: Oklahoma City traded Paul George for Victor Oladipo and in doing so traded improved perimeter offense for diminished perimeter defense.
Trail Blazers: Portland has limited financial flexibility and may not be able to get in on the free-agent fun this summer.
Nuggets: Paul Millsap will provide a big boost to Denver’s rim protecting efforts and he’ll be a nice front-court complement to Nikola Jokic.
Timberwolves: The Wolves’ charts demonstrate perhaps the most impressive team overhaul. Jimmy Butler brings an offensive infusion, Jeff Teague replicates Ricky Rubio’s production, and Taj Gibson balances out Andrew Wiggins defensive struggles. Their chart now appears to show a mostly-positive roster. One worrying caveat — Minnesota has parted ways with Zach Lavine, who was their only 3-point spacer last season, so spacing could be tight.
Cavaliers: Cleveland has done little to address their lack of effective two-way wings to date this offseason.
Bucks: Still have Giannis.
Pacers: Here’s the other side of the George-Oladipo swap, illustrating Indiana’s diminished perimeter offense and improved perimeter defense after the trade.
Bulls: On the bright side, they scaled down their stable of not-helpful ball-handlers by cutting loose Rajon Rondo and Michael Carter-Williams. Unfortunately, they also added a possibly-not-very-helpful ball handler in Kris Dunn.
Pistons: Building a wall.
Celtics: Boston claimed one of the jewels of this summer’s free agency by signing Gordon Hayward and he should bring offensive production on the perimeter. Giving up Olynyk, Zeller, and Johnson to clear space for his contract may weaken the Celtics interior defense and trading Marcus Smart could make things even worse on that end of the court for the Cs.
Raptors: Toronto lost two helpful defenders in P.J. Tucker and Patrick Patterson.
Knicks: Did not trade Porzingis.
Sixers: New additions J.J. Redick and Amir Johnson are buoying Philadelphia up to neutral position for a few roster categories. Ben Simmons and Markelle Fultz (not pictured) are providing other reasons for optimism.
Nets: D’Angelo Russell has room for improvement this year.
Wizards: Washington still has some defensive issues to address.
Hawks: Atlanta has been involved in a few exciting trades featuring big names, but they don’t have much to show for their cooperation.
Heat: No big moves from Pat Riley, yet.
Hornets: Charlotte plans to have Dwight Howard anchor their defensive scheme.
Magic: The Southeast Division is pretty boring on the free-agency front.
Next: Nylon Calculus -- A nearest neighbor approach to 2017 NBA Draft statistical comparisons
Obviously, rosters remain very much in flux as teams are still adding and dropping players. I will continue to update the charts as the free-agency dust settles and rosters become finalized.