Putting Point Guards In A Box

Oct 13, 2014; Salt Lake City, UT, USA; Los Angeles Clippers guard Chris Paul (3) dribbles the ball while defended by Utah Jazz guard Trey Burke (3) during the first quarter at EnergySolutions Arena. Mandatory Credit: Russ Isabella-USA TODAY Sports

[ED NOTE: What you are about to read is about point guards, and it is awesome. For more on stylistic point guard analysis, read the annals of Pure Point methodology by the Ghostface Birdziller. – Ed. Emeritus)

Shoot first…Game manager…Coach on the floor type…More of a combo than a lead guard…

The above are all appellations commonly applied to various point guards in the NBA. But without any further definition, these labels are more often used to arbitrarily promote or disparage a given player. The desire to find ways to sort the wheat from the chaff at the NBA’s deepest position is understandable, but without more definitive categories of players it’s near impossible to sort.

In the rush of NBA fans and analysts to answer questions of universal rankings with context-free and one-size-fits all determinations of player ability, the question of “how” often gets ignored in favor of “what” or “how much?” This rush to judgement over understanding can be misguided.

Like any collaborative environment, an NBA team or lineup depends on assembling the proper mix of talent. In many ways this notion of “fit” can be almost as important as the total amount of talent on hand. A point guard who is an elite jump shooter, but less good at getting into the paint and creating for others might fit perfectly alongside a ball-dominant wing such as LeBron James or James Harden, while that same player would struggle (as would the team) if the other perimeter players were similarly dependent on teammates creating openings. Of course sometimes players have malleable talents; Steph Curry could function splendidly in either role, while George Hill and Mario Chalmers will have opportunities to expand their influence this season due to roster changes and injuries.

Still, once established in the NBA, players tend to “do what they do.” Taking the effort to describe and categorize what players are trying to do as much as it is to value their overall contributions. In that way, teams can seek to acquire players who fill apparent needs rather than duplicate areas of strength. Fans can better understand why some players struggle when in one role but shine in another.

Sep 29, 2014; Waltham, MA, USA; Boston Celtics guard Rajon Rondo (9) during media day at the Celtics practice facility. Mandatory Credit: David Butler II-USA TODAY Sports

The Point Guard Personality Test

The best friend of many a human resources director (and thus enemy of right thinking people everywhere) is the Myers Briggs Type Indicator. Joking aside, Myers Briggs can provide some cogent insight into how an individual might fit in with an organization and even what career or job that person should seek.

The MBTI, as it is known, evaluates a personality across four “dichotomies” to determine how a person perceives and interacts with the world. While not perfect and almost intentionally reductive, MBTI and similar tests measure style more than aptitude. In other words, it’s designed to determine if someone is a comedian, not if they are actually funny.

This focus on style makes MBTI a decent example of a framework around which to build a model of point guard play. The goal is to create a personality test for styles of point guard play.

This personality test is not intended as any sort of ranking. Rather the categories are purely descriptive of style of play, on offense only. Both for simplicity (to keep the number of variables small) and convenience (trying to parse available defensive metrics would be something of a nightmare at present) the other side of the ball is not a part of this discussion.

This isn’t to say defense is unimportant. The old cliche that it’s half the game is mostly true. But “point guard play” as is commonly discussed is explicitly about running an offense. No one has ever had the charge of “not a true point guard” leveled at them because of suspect defense, just ask Steve Nash.

No one style of point guard play is necessarily “best.” So far as can be made out from one season’s worth of data, it’s possible to be an effective player in virtually any combination of theses traits. A player’s level of achievement is a combination of style and ability — there are some point guards who are very effective when they penetrate towards the basket, but don’t do it especially often. Others are mediocre scorers who still shoot too much. This analysis does not attempt to measure or capture the fit between skill and style. It’s a measure only of what a player is trying to do, not how well it is done.

What constitutes a “point guard” is extremely malleable, especially in the modern era of positional versatility. For the purpose of this study, a point guard was defined as someone who would be publicly perceived as such. In this way, both Eric Bledsoe and Goran Dragic can be counted, while LeBron James (who fills the role of offensive facilitator more than many point guards) would not. I’ve included all players who played at least ten games for a given team in 2013–14, and since the categorization depends greatly only data from the SportVU player tracking system, it only measures style of performance for that season.

May 14, 2014; San Antonio, TX, USA; Portland Trail Blazers guard Damian Lillard (0) shoots the ball over San Antonio Spurs forward Tim Duncan (21) in game five of the second round of the 2014 NBA Playoffs at AT&T Center. Mandatory Credit: Soobum Im-USA TODAY Sports

Trait #1 — Shoot first vs. Pass first.

In “true point guard” discussions, the trait most often discussed is the degree to which a player looks to shoot or distribute more. Does a guy look to get his own first, but when the defense adjusts finds a teammate, a la Damian Lillard? Or is he always looking to pass and shoots more as a last resort or attempt to keep the defense honest, like Rajon Rondo?

There are many different ways to place a point along the sliding scale between “scoring” and “passing.” Perhaps the simplest is simply comparing the ratio of shots to passes. This is unsatisfying for a number of reasons. Some passes simply move the ball, while others might be described as “productive,” allowing for teammates’ scoring opportunities.

Comparing players’ scoring attempts against these “productive” passes is a better way of capturing whether a player is looking to score or pass when he decides to “do something” with the ball. Starting from last year, the data collected from the SportVU system in all 29 NBA arenas allows for an investigation of this ratio.

Across the NBA, point guards shoot just slightly more often than they set their teammates up, averaging 1.08 scoring attempts[1. For the purposes of this article “shots” and “scoring attempts” will be used interchangeably to account for free throws using the generally convention used for shot attempts; FGA + .44*FTA. This is an estimate, but close enough for purposes of categorization.] per assist opportunities generated. In this analysis players who shot more often than that are described as “shoot first” with others being “pass first.” At one end of the spectrum were those who didn’t just shoot first, but shot most. And at the other extreme, the players most inclined to pass.

[ED Note: The original version of this article included a data calculation error in this section and incorrect ratios in the chart above. They have since been updated and the preceding paragraph has been rewritten to reflect the difference. Several graphs required minor changes as well. The ensuing analysis was unaffected by the error and is thus unchanged. ]

Oct 6, 2014; San Diego, CA, USA; Los Angeles Lakers guard Jeremy Lin (17) is defended by Denver Nuggets center Timofey Mozgov (25) during the second half at Valley View Casino Center. Mandatory Credit: Jake Roth-USA TODAY Sports

Trait #2 — “Driving” versus “Probing”

At either extreme of this spectrum there is the explosive player who just wants to put his head down and get to the rim, think Jeremy Lin or Derrick Rose. On the other end is the crafty veteran who uses angles, subtle hip bumps and changes-of-pace to operate from the mid-range area, a-la Andre Miller or Chris Paul. Both approaches have merit. Drives to the basket have greater risks (turnovers) and rewards (layups, getting fouled, high quality assist chances to teammates shooting layups or threes), while “probing” as I’m going to call it is a lower risk, lower reward style.

In some ways, it’s best to show the differences visually. First, Paul’s shot chart from last year (ignore the three pointers and relative accuracy from each location for the moment) :

And compare with Lin:

While the shot charts don’t tell the whole story, it’s easy to envision Paul’s careening off defenders before lurching into his vaguely off-balance but still accurate 16-footer from the right elbow, while Lin’s natural inclination to get to the basket is only accentuated by the extreme’s of Houston’s “MoreyBall” philosophy of completely eschewing the midrange game.

But how to express this numerically? One method would be to examine the propensity of each player to drive the ball. Drives, as captured by SportVU, are instances where the ball-handler received the ball more than 20 feet away from the rim and dribbled to within 10 feet of the basket, with fast break situations not counting.. It’s an imperfect proxy for this sort of stylistic description, but still useful. Accounting for the times players touched the ball in post up situations or caught-and-shot, here were the least and most likely players to drive to the basket[2. league average for all point guards was right around 10%]:

From just these two traits, we can start to assign players to categories:

Of course, these are sliding scales of shot dominant vs. passer and driver vs. probe, and putting together one chart gives an idea of how extreme some of these tendencies can be[3. Jordan Crawford was literally off this during his time in Golden State, averaing over 4 shots per assist chance. Fully to scale he would be located well off the right edge of the chart above.]:

[ED Note: The graph above has also been changed to correct the already mentioned data calculation errors found in the original.]

While spacing the floor and the time spent with the ball are certainly important in describing a point guard, the shoot-pass and drive-probe dichotomies capture the most important elements of what is generally termed “point guard play.” The other two variables add some detail within the broader categories.

Oct 10, 2014; Indianapolis, IN, USA; Indiana Pacers guard George Hill (3) dribbles the ball while being guarded by Orlando Magic guard Luke Ridnour (13) in the third quarter at Bankers Life Fieldhouse. The Orlando Magic beat the Indiana Pacers by the score of 96-93. Mandatory Credit: Trevor Ruszkowski-USA TODAY Sports

Trait #3 – On-Ball vs. Off-Ball

There is a wide variation in the amount of time point guards seem to spend with the ball in their hands. Some, like Rajon Rondo pound the ball at the top of the floor while directing traffic. Others, like Chalmers or Hill function much more as spot-up shooters as their more dynamic teammates take more of a primary creation role. There are a number of different ways to examine this dichotomy, but perhaps the simplest is just looking at how often the player has the ball in his hands while he’s on the floor.

Here are the players with the highest and lowest possessions percentages (note that these percentages account for the entire time the player is on the floor, both offensively and defensively. [4. Also note that by SportVU definitions the ball is “possessed” by either team for a total about 37 minutes of a 48 minute game with the other 11 minutes presumably including time the ball is in the air after a shot or is loose on the floor after a rebound]

May 27, 2014; Oklahoma City, OK, USA; Oklahoma City Thunder guard Russell Westbrook (0) attempts a shot against San Antonio Spurs guard Danny Green (4) during the third quarter in game four of the Western Conference Finals of the 2014 NBA Playoffs at Chesapeake Energy Arena. Mandatory Credit: Mark D. Smith-USA TODAY Sports

Trait #4 — Floor-Spacer vs. non-shooter

The first three categories mostly discuss play with the ball: how much does a player have the ball, and what does he do with it. However, an increasingly important part of point guard play is the ability to influence the game without the ball. Specifically, to space the floor. Ideally, the scale on which this aspect was captured would measure how much the defense feels it has to guard the player when he doesn’t have the ball. Unfortunately, data to accurately measure this effect doesn’t exist yet, at least not in public. One option would be to look at three point percentage, but that doesn’t tell the whole or even most of the story — a high volume 34% three-point shooter most likely spaces the floor better in terms of the defense “honoring” that shot than a guy who rarely shoots thress but does so at substantially better percentage. In other words Brandon Jennings is probably a more effective floor spacer than Tony Parker.

Catch-and-shoot threes might work, but one player alone breaks that metric. Steph Curry was roughly league average for a point guard in terms of catch-and-shoot threes attempted per minute. Any scale in which Curry is not the ne plus ultra of floor spacing point guards is probably a bad scale. In the end, it’s probably best to keep it simple and go with attempts per minute, expressed here as 3PA/36. [5. PG’s leaguewide averaged 4.3 threes attempted per 36 minutes played.]

Player Types & Visualizations

With these four scales in place, it’s possible to start categorizing the point guards played enough for any numerical analysis to be meaningful. Double-counting the 7 players who played enough with two different teams[6. With the exception of Ramon Sessions, there were small but significant differences in how each of these players operated between time spent on one team and the other.],  there were 71 point guards analyzed.

Possibly more useful for quickly sorting players into buckets is scaling each category. In this chart, 100 in category is precisely league average for point guards, with higher values being more extreme towards the high end of shoot first, driving, playing on the ball and spacing the floor effectively,

Visualizing these four traits is probably an easier method of comparison. Each player’s particular style can be represented by the shape their scores in each trait generate on a radar chart like this one:

The inner diamond corresponds with league average rates, values extending past this diamond mean a player is towards the listed side of that scale, while values which are wholly inside this diamond indicate the player’s style is more towards the opposite end of the spectrum. Again “more” is not necessarily better or worse, these charts describe style, not achievement. As an example, here is visualization of the style of play for Atlanta’s Jeff Teague:

This shows that Teague is slightly less shot-happy than the average point, is extremely aggressive in terms of driving to the basket, but not much of a floor spacer, while controlling the ball slightly more than average for a lead guard. All of which aligns quite nicely to the Jeff Teague seen on the the NBA floor.

So what are some of the more archetypal “point guard personality types?”

The Game Manager –More of a football term, a “game manager” of a quarterback is a player who doesn’t take many risks, runs the offense competently and isn’t going to lose you the game. He likely won’t win it for you either, a well-timd spot up three pointer aside. Jose Calderon is the model game manager, due to his ability to shoot an elite percentage while rarely turning the ball over.

The Undersized Two Guard — Really this is two different types but very closely related types of player. Generally stationed off the ball, much more inclined to shoot than pass and with the ability to stretch the floor. The biggest difference is between the pure “bomber” version of this player type, a la Patty Mills:

and the player more willing to waterbug his way into the lane, such as Aaron Brooks (in both Houston and Denver last season)

The Scoring Point — The main difference between what I’m calling a “scoring point” and an undersized shooting guard is ball-dominance. A player like an Isaiah Thomas may look to shoot a fair amount, but he also initiates the offense with more regularity (3rd largest possession percentage among the players tracked in the case of Thomas) than do the ersatz shooting guards:

The Slasher — Another variation in the scoring point/undersized two category is the slasher — a player who looks to get his own shot, but more accurately, is looking to get to the basket to score whenever possible, often eschewing the three pointer. Reggie Jackson and Jeremy Lin are two practitioners of this style.

The Drive and Kick — Some guys you just know are driving to create passing angles and easy looks for teammates, but the best ones are gifted enough as penetrators and passers they can make it work. Ricky Rubio and Rajon Rondo are the exemplars here.

The last two styles are fairly similar, and Teague falls somewhere in the between the two.

The Jack of All Trades — Some players are more difficult to typecast due to their versatility. In addition to being one of the very best defenders at the point in the NBA, Mike Conley has the versatility to mold his game to what the team needs. Other than being somewhat ball-dominant, Conley is right around average in terms of his propensity to shoot or drive and his affinity for shooting threes. He does all of these things well enough that he can choose to make the most appropriate play for the situation rather than stick to his prefered methods, rain or shine.

Of course, these are just a few examples of the typologies. Using sliding scales allows for a bit more refinement than simply describing a player as shoot first — there are gradations. Both Kyrie Irving and Steph Curry are “shoot first” players, but Irving to a much larger degree. The degree to which LeBron James being back in Cleveland alters that tendency is a mystery to be solved over the course of the coming season.

This also illustrates a caveat of going too far down the rabbit hole of this analysis: there is only one season’s worth of data on many of these metrics, and it is impossible to determine how a point guard’s “type” changes over time or across various situations until there are more seasons of data. The degree of change most players who were traded last season experienced suggests there is still a fair amount of context at play here. Andre Miller handled the ball a lot in Denver, very little (relatively speaking) in Washington. Greivis Vasquez looked for his shot, especially from deep, more in Toronto than he had in Sacramento

Applications

What can these charts actually show? Certainly one use would be to inform how a change at the point might affect a team the following season.

For example, one of the 2014 offseason’s more curious moves was Sacramento’s choice to replace Isaiah Thomas with Darren Collison at the point. The justification for the move was that Collison is more of a “pass-first” player than Thomas. Of course, this isn’t really true. Skill differences aside, how might replacing Thomas with Collison affect the shape of the Kings offense this coming season?

While the Clippers were certainly a more talented offensive team than the Kings last year, there are many reasons to suspect Collison won’t be able to fill Thomas’s role. The lack of ball-dominance by Collison isn’t simply a result of playing with Chris Paul, as Paul was on the floor just under a third of Collison’s minutes in 2013–14.

Collison drove and shot far less than Thomas even controlling for playing time. It’s hard to imagine DeMarcus Cousins taking on even more of the load than he carried last year, when he lead NBA bigs in usage, so the remainder is likely going to be picked up by some combination of Nick Stauskas, Ben McLemore and Rudy Gay.

McLemore was underwhelming as a rookie and had a moderately disappoiting summer league showing. Stauskas has potential as a ball-handler, but is also going to be a rookie. Which probably means Gay will get the lion’s share of the additional duty. Unfortunately, Gay demonstrated fairly conclusively in his time in Toronto that he can handle only so much offensive responsibility before completely torpedoing an offense.

So replacing Thomas with Collison is already looking like a loser just in terms of the effects on the other players. Add in the fact that Thomas is flat better at anything you want a point guard to do offensively, and it’s going to be a(nother) long season in Sacto.