As NBA Empires Rise and Fall

Jan 12, 2015; Washington, DC, USA; (left to right) San Antonio Spurs forward Tim Duncan, Spurs guard Manu Ginobili, and Spurs guard Tony Parker present President Barack Obama with a jersey during a ceremony honoring the NBA Champion Spurs in the East Room at The White House. Mandatory Credit: Geoff Burke-USA TODAY Sports
Jan 12, 2015; Washington, DC, USA; (left to right) San Antonio Spurs forward Tim Duncan, Spurs guard Manu Ginobili, and Spurs guard Tony Parker present President Barack Obama with a jersey during a ceremony honoring the NBA Champion Spurs in the East Room at The White House. Mandatory Credit: Geoff Burke-USA TODAY Sports /
facebooktwitterreddit
Jan 12, 2015; Washington, DC, USA; (left to right) San Antonio Spurs forward Tim Duncan, Spurs guard Manu Ginobili, and Spurs guard Tony Parker present President Barack Obama with a jersey during a ceremony honoring the NBA Champion Spurs in the East Room at The White House. Mandatory Credit: Geoff Burke-USA TODAY Sports
Jan 12, 2015; Washington, DC, USA; (left to right) San Antonio Spurs forward Tim Duncan, Spurs guard Manu Ginobili, and Spurs guard Tony Parker present President Barack Obama with a jersey during a ceremony honoring the NBA Champion Spurs in the East Room at The White House. Mandatory Credit: Geoff Burke-USA TODAY Sports /

First of all, credit where credit is due. I got the idea for this article from the 538 Elo NBA history, which made me think ‘if only people would start to like heatmaps, they are such a nice way to show several things at the same time!’. Thankf for much of the data for this article from Krishna Narsu (aka ‘Sir Scrape-a-Lot’) and of course the invaluable basketball-reference.com (aka ‘Basketball nerd heaven’). So, let’s start!

NBA teams have a natural life cycle. Sometimes you are up, then your best players start to decline due to age, or go somewhere else for more money, or the league itself simply moves on. I wanted to find a way to illustrate these rhythms over time. Starting from the ’75-’76 season (one year before the ABA merger) I used SRS, short for ‘Simple Rating System’, to illustrate team strength. In a nutshell[1. SRS is explained more fully here.] SRS combines a team’s average margin of victory with a strength of schedule measure based on opponents.

The following chart shows all 30 teams and how they fared over time. I am sorry Sonics fans, but for simplicity I merged teams that moved cities or changed names. [2. I also forgot to include that the New Orleans Hornets played in Oklahoma City for two seasons.] The green squares mark the each season’s champion:

Nba history
Nba history /

A few observations:

  • I find this figure a very nice visualization of how continuously dominant the Spurs were for the last 17 years. I first thought about acknowledging the Lakers threepeat, but changed my mind upon seeing this endless line of shades of red
  • Tom Haberstroh estimated how many titles each team was expected to win during its history, giving the Suns an expected 1.9 titles. You can see on the plot that they were really good during large stretches of both the MJ and the Duncan era. Alas…
  • Tom also mentioned that the Lakers are the most overachieving team, with 10 titles since 1976. But he might have ignored the smaller league size during the 80’s. Intuitively, a 60 plus win team in a 22 team league should already win way more titles than in a 30 team league. Plus, of the four decidedly strongest teams during that era, the other three (Boston, Philly, Milwaukee) were in the Eastern Conference.
  • Basically all Champions were either dark red during the season they won the title or in the season directly afterwards. Which makes the Rockets light pink repeat even more impressive

There are many more things to see in this figure, but let’s move on for now…

Changes in #Winning

One question I was asking myself was which team characteristics were the most related to winning over history. Skipping obvious things like shooting percentages, I looked at the correlation between SRS and 9 categories.[3.

  • Age (meaning average team age weighted by minutes)
  • Pace
  • 3 point attempt rate (T3Pa)
  • Offensive and defensive free throw rate (OFTpFGa, DFTpFGa)
  • Rebound percentages (ORBp, DRBp), and
  • Turnover percentage (OTOp, DTOp)

] The next figure shows the results over time, using a smoothed function for the thick lines, as the real values (thin lines) are quite jittery. I also marked a few rule changes that in my opinion could play a role in the changes in importance of various aspects of play.

Correlation through history
Correlation through history /

What we can see:

  • The biggest correlation is between age and SRS. More in this in a minute
  • Rebound percentages: Protecting the defensive glass has had positive correlation to SRS throughout the decades. There was a small correlation between offensive rebounding and SRS until the middle of the nineties, but this disappeared. I don’t want to speculate too much about why other than to observe that it happened.
  • Free throw rate: Defensive free throw rate has always been negatively correlated to SRS (which makes sense). There is a dip for offensive free throw rate in the early 2000’s. This is related to two things. First, teams that have more free throws tend to have more turnovers. And secondly and especially during the beginning of this millennium, teams that shot more three’s (like Nash’s Suns) tend to get to the line less frequently.
  • Turnovers: It is definitely good to avoid turnovers. The correlation between forcing turnovers and winning remains less clear. Perhaps for both offensive rebounding and forcing turnovers, a team needs to be able to do those things “efficiently”[3. Not giving up fast breaks off of offensive rebound attempts, or making poor gambles for steals and giving up open shots.] for them to be a net positive. That said, teams are always helped by taking care of the ball offensively.
  • Pace: For most years, pace did not correlate to playing well in any shape or form. But you can see the nice dip between 1985 and 2000, where slower teams tend to play better. In my opinion, this both started and ended with the old illegal defense rules, plus the changes to the amount of allowed hand checking and the five second rule. I wanted to insert a video of Charles Barkley or Mark Jackson sticking their butt into a defender for 10 seconds to get from the three point line to the zone but couldn’t find any.
  • The three pointer has become just as important over as defensive rebounding over the last few years.

Returning to the influence of age. The question with correlation is always a bit of a ‘chicken vs egg’ problem.[5. Note: Obviously the egg. Eggs exist much longer than chicken! – Ed: I have no idea what Hannes is talking about here! ] Team age is also related to roster continuity. If 80% of a teams’ players remain the same over four years (a la Spurs), they all age by four years together.

It stands to reason that teams tend remain around the same average age as long as they are not particularly good. Once a squad starts to become better, several things kick in. One is the desire to not fix something that isn’t broken – good teams retain their core, maintaining some continuity and thus grow old together. Another aspect is that better teams draft lower, making it less likely that to add productive young talent given the rapidly diminishing returns as you move down draft order. Finally, good teams, especially self-styled contenders often fill their last few sport with veterans, And so the aging process follows the winning process.

As NBA Empires Rise and Fall

The chart below, is similar to the first one, but with a line for team age added underneath. It’s might be slightly visually confusing at first, but I hope it is somewhat understandable.[6. I needed to rescale age to fit the same color scheme: Average team age is 26.9 so I subtracted that from team age and then multiplied by 2 to match scales. Ages are as of the end of the season in question.] As a rule of thumb: If a team age is dark red, it’s average age is around 30 and if it is dark blue it is around 24. I added some notable players and the years that they got to the team as well.  So the 2003 draft class of LeBron, Dwyane and Co. is marked as ’04:

NBA empires
NBA empires /

Notable is how much faster team strength changes than does team age age. Roster continuity stats aren’t included, as the charts would become visually unwieldy. That is a comparison for a future post perhaps.

But in many cases there is a clear pattern. For a bad team, a star rookie is drafted yet the team remains below average. Around year three or four, team strengths as measured by SRS increases drastically. Subsequently, as the team reaches and tries to maintain contender status average age starts to increase as well. At the end of the cycler, there is often have a sharp decrease of SRS while the average player age remains fairly old. Followed by the team bottoming out in the next two years with average age going back down,[7. Except for the 2002-2005 Knicks.] and the circle of life starts anew.

Unless you are the San Antonio Spurs, in which case you simply get new superstar or two every 7 years and continue as if no time had passed at all.