Fun With Numbers and Venn Diagrams

Published: December 9, 2009

The Hornets play the Timberwolves tonight and hopefully Timberwolves’ Coach Kurt Rambis still won’t have grasped that the key elements of his best starting line up would be Love-Sessions with Big Al. 

Sorry.  I couldn’t resist that.  Particularly since it’s true.

Anyways, here’s the preview I did two games ago.  The only change is Paul is playing, and I’d give Kevin Love the advantage over David West if Love starts.  If Love doesn’t start, the Wolves’ bench becomes better than the Hornets’, hands down.

Now, since I already have a recent post breaking down the matchups, I’m instead going to present something I’ve been playing with for about a week.  Brace yourself: 

Point Guard Venn Diagrams

I’m a Business Analyst at a financial services company, and one thing you learn pretty quickly when you dwell in a cube and analyze things is that there are a thousand different theories on how people should try to communicate what they just analyzed – and every business goes through phases where they experiment with a whole slew of them.  A while back, we had a manager who loved Venn Diagrams,  so I was required to learn how to make them.   I never used them, but boy, could I make them.  I realized the other day, however, that I could actually use them for this blog.  They are colorful, simple, and allow me to crunch numbers and come up with new stats en route to evaluating point guards in the NBA.  I mean, what could be more fun than that?

The Basic Point Guard Venn Diagram

There are different types of guards in the league who play the point: shoot first, pass first, combo guard, game managers, defensive specialists, three-point specialists . . . the list goes on.  I wanted to create a series of Venn Diagrams that would categorize point guards of the NBA and illustrate their strengths and weaknesses.  For my first Venn diagram, I categorized point guards according to what most basketball observers feel are the most important attributes of a true point guard: passing, ball handling, and good shot selection.  To me, those stats can be reflected by the metrics of Assist Rate(% of posessions the player ends with an assist), Turnover rate (% of posessions the player ends with a turnover) and True Shooting %.(Scoring efficiency taking into account three-pointers and free throws)

So I made a list of all the players who have played at least 10 games this season, averaged 25 minutes per game or more, and spend at least a significant amount of their playing time at the point.  I then took the average NBA point guard’s assist rate,(27.1%) Turnover Rate,(11.7%) and True Shooting %.(53.7%)  If a player fell into those categories, they are then encompassed by the appropriate “category circle” in the following Venn Diagram.  Here it is:

How do you read it?  The diagram shows that Tony Parker is in the Efficient Scorer Category, but does belong to the high Assist rate or  Low Turnover rate categories.  Deron Williams is included in both the High Assist Rate and Efficient Scorer categories, but not the Low Turnover Rate category.  CP3 is in all three categories, and therefore is above average at all three things.  Makes sense?  Good.

There are a few interesting items to be found there.  It’s surprising to see Russell Westbrook and Derrick Rose aren’t above average in any of those three skills.  It’s just as surprising to see names like Jason Williams and Mike Bibby residing in the middle section, providing above-average stats in all three categories.  And If you had asked me where rookie point Brandon Jennings would fall in a chart like this, I doubt I would have said his only above average category would be turnover rate.

The Scoring Point Guard Venn Diagram

That last point about Jennings, however, begs the question.  Are these the best three categories to use to evaluate point guards?  Gilbert Arenas and Brandon Jennings are much more to their teams than simply low-turnover game managers.  Does Lou Williams really belong where he is listed?  Some of the guards excluded from the central section of the diagram are excluded because they pass a lot – and therefore are liable to have more turnovers, but that’s not necessarily a bad thing.  So I changed my three categories.  Assist rate and Turnover rate were combined into an Assist to Turnover Ratio, and Usage(percentage of team posessions used by the player as a shot attempt) was added to the next Venn Diagram:

The result?  Jason Williams, Jose Calderon, Mike Bibby and Rajon Rondo fall out of the center section of the diagram, as they shoot less often than the average PG usage rate of 21%.  Jennings and Arenas now shift from the turnover rate category to the high usage category, reflecting the fact they get few assists to go along with those few turnovers – and also jack up below average efficiency shots at a tremendous rate.  Deron Williams and Steve Nash, high turnover-high assist guys, and Lou Williams and Chauncy Billups, low assist and even lower turnover guys, join Chris Paul in the middle of the diagram.

The Master Point Guard Venn Diagram

Again, however, I wasn’t entirely happy with the diagram.  First, I think shooting and usage need to be linked more firmly together.  If a player shoots a ton, but is below average (like Rodney Stuckey this year – 26.4% usage, 47.7% TS%) then they  are doing two things wrong.  So I created a new measure that took usage rate as a starting point and modified it based on whether a player was shooting above or below the average for a point guard.  The measure ends up rating players like Jose Calderon, who don’t shoot often(usage 16%) but are very efficient(62.3% TS%) as good pure scorers, while it rates players like Derrick Rose, who have taken a lot of shots(usage 24%) but have been terrible at converting them(TS% of 49%) as below average.  It’s fairly simple.

Once I had finished combining Usage and Efficiency, it allowed me to add a new component to the diagram: defense. Unfortunately, there is a dearth of stats on the sort of defensive impacts players have on a game.  Opponent PER has a tremendous amount to do with the team defense as a whole.  Defensive Rating does as well.  In the end, I kept it simple – perhaps too simple – and used a “defensive plays” stat that combines steals, charges drawn, and blocks into a single rating.  The resulting Venn:

As you can see, it becomes much harder to stay in that middle area, as Chauncey Billups, Deron Williams and Steve Nash all slip out due to defensive deficiencies.  Weirdly, Lou Williams is the lone player who manages to hang with Chris Paul, which makes me really want to know why I haven’t seen any articles about how surprisingly good he’s been this season.  Where is the Lou Williams love, people?  And wow, Russell Westbrook and Derrick Rose aren’t living up to their billing – they aren’t above average in almost anything.

Oh, and it also proves a point – only one name stayed in the middle, master point guard tier, no matter what categories I put together:  Chris Paul.  But we all knew that, didn’t we?

Anything on those diagrams surprise you?